{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 29210,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.423485107839781e-05,
      "grad_norm": 0.0,
      "learning_rate": 2.280501710376283e-08,
      "loss": 16.9497,
      "step": 1
    },
    {
      "epoch": 6.846970215679562e-05,
      "grad_norm": 0.0,
      "learning_rate": 4.561003420752566e-08,
      "loss": 16.9908,
      "step": 2
    },
    {
      "epoch": 0.00010270455323519343,
      "grad_norm": 0.0,
      "learning_rate": 6.841505131128849e-08,
      "loss": 17.0152,
      "step": 3
    },
    {
      "epoch": 0.00013693940431359124,
      "grad_norm": 0.0,
      "learning_rate": 9.122006841505132e-08,
      "loss": 16.9132,
      "step": 4
    },
    {
      "epoch": 0.00017117425539198904,
      "grad_norm": 0.0,
      "learning_rate": 1.1402508551881415e-07,
      "loss": 16.9594,
      "step": 5
    },
    {
      "epoch": 0.00020540910647038685,
      "grad_norm": 0.0,
      "learning_rate": 1.3683010262257699e-07,
      "loss": 16.9724,
      "step": 6
    },
    {
      "epoch": 0.00023964395754878467,
      "grad_norm": 0.0,
      "learning_rate": 1.596351197263398e-07,
      "loss": 16.971,
      "step": 7
    },
    {
      "epoch": 0.0002738788086271825,
      "grad_norm": 0.0,
      "learning_rate": 1.8244013683010264e-07,
      "loss": 16.9601,
      "step": 8
    },
    {
      "epoch": 0.0003081136597055803,
      "grad_norm": 0.0,
      "learning_rate": 2.0524515393386548e-07,
      "loss": 16.9157,
      "step": 9
    },
    {
      "epoch": 0.00034234851078397807,
      "grad_norm": 0.0,
      "learning_rate": 2.280501710376283e-07,
      "loss": 16.9174,
      "step": 10
    },
    {
      "epoch": 0.0003765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2.5085518814139116e-07,
      "loss": 16.7159,
      "step": 11
    },
    {
      "epoch": 0.0004108182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2.7366020524515397e-07,
      "loss": 16.7359,
      "step": 12
    },
    {
      "epoch": 0.0004450530640191715,
      "grad_norm": 0.0,
      "learning_rate": 2.964652223489168e-07,
      "loss": 16.6819,
      "step": 13
    },
    {
      "epoch": 0.00047928791509756934,
      "grad_norm": 0.0,
      "learning_rate": 3.192702394526796e-07,
      "loss": 16.6507,
      "step": 14
    },
    {
      "epoch": 0.0005135227661759672,
      "grad_norm": 0.0,
      "learning_rate": 3.4207525655644247e-07,
      "loss": 16.1523,
      "step": 15
    },
    {
      "epoch": 0.000547757617254365,
      "grad_norm": 0.0,
      "learning_rate": 3.648802736602053e-07,
      "loss": 16.1374,
      "step": 16
    },
    {
      "epoch": 0.0005819924683327628,
      "grad_norm": 0.0,
      "learning_rate": 3.876852907639681e-07,
      "loss": 16.1157,
      "step": 17
    },
    {
      "epoch": 0.0006162273194111606,
      "grad_norm": 0.0,
      "learning_rate": 4.1049030786773096e-07,
      "loss": 16.0654,
      "step": 18
    },
    {
      "epoch": 0.0006504621704895583,
      "grad_norm": 0.0,
      "learning_rate": 4.3329532497149377e-07,
      "loss": 16.0585,
      "step": 19
    },
    {
      "epoch": 0.0006846970215679561,
      "grad_norm": 0.0,
      "learning_rate": 4.561003420752566e-07,
      "loss": 15.6925,
      "step": 20
    },
    {
      "epoch": 0.000718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 4.789053591790194e-07,
      "loss": 15.4189,
      "step": 21
    },
    {
      "epoch": 0.0007531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 5.017103762827823e-07,
      "loss": 15.357,
      "step": 22
    },
    {
      "epoch": 0.0007874015748031496,
      "grad_norm": 0.0,
      "learning_rate": 5.245153933865451e-07,
      "loss": 15.3389,
      "step": 23
    },
    {
      "epoch": 0.0008216364258815474,
      "grad_norm": 0.0,
      "learning_rate": 5.473204104903079e-07,
      "loss": 15.3316,
      "step": 24
    },
    {
      "epoch": 0.0008558712769599452,
      "grad_norm": 0.0,
      "learning_rate": 5.701254275940708e-07,
      "loss": 15.2372,
      "step": 25
    },
    {
      "epoch": 0.000890106128038343,
      "grad_norm": 0.0,
      "learning_rate": 5.929304446978336e-07,
      "loss": 15.1788,
      "step": 26
    },
    {
      "epoch": 0.0009243409791167409,
      "grad_norm": 0.0,
      "learning_rate": 6.157354618015964e-07,
      "loss": 14.9348,
      "step": 27
    },
    {
      "epoch": 0.0009585758301951387,
      "grad_norm": 0.0,
      "learning_rate": 6.385404789053592e-07,
      "loss": 14.6999,
      "step": 28
    },
    {
      "epoch": 0.0009928106812735365,
      "grad_norm": 0.0,
      "learning_rate": 6.61345496009122e-07,
      "loss": 14.4748,
      "step": 29
    },
    {
      "epoch": 0.0010270455323519343,
      "grad_norm": 0.0,
      "learning_rate": 6.841505131128849e-07,
      "loss": 14.3622,
      "step": 30
    },
    {
      "epoch": 0.0010612803834303321,
      "grad_norm": 0.0,
      "learning_rate": 7.069555302166476e-07,
      "loss": 14.2679,
      "step": 31
    },
    {
      "epoch": 0.00109551523450873,
      "grad_norm": 0.0,
      "learning_rate": 7.297605473204106e-07,
      "loss": 14.1828,
      "step": 32
    },
    {
      "epoch": 0.0011297500855871278,
      "grad_norm": 0.0,
      "learning_rate": 7.525655644241734e-07,
      "loss": 14.1624,
      "step": 33
    },
    {
      "epoch": 0.0011639849366655256,
      "grad_norm": 0.0,
      "learning_rate": 7.753705815279362e-07,
      "loss": 14.085,
      "step": 34
    },
    {
      "epoch": 0.0011982197877439234,
      "grad_norm": 0.0,
      "learning_rate": 7.98175598631699e-07,
      "loss": 14.0349,
      "step": 35
    },
    {
      "epoch": 0.0012324546388223212,
      "grad_norm": 0.0,
      "learning_rate": 8.209806157354619e-07,
      "loss": 14.044,
      "step": 36
    },
    {
      "epoch": 0.0012666894899007188,
      "grad_norm": 0.0,
      "learning_rate": 8.437856328392246e-07,
      "loss": 13.989,
      "step": 37
    },
    {
      "epoch": 0.0013009243409791166,
      "grad_norm": 0.0,
      "learning_rate": 8.665906499429875e-07,
      "loss": 13.8885,
      "step": 38
    },
    {
      "epoch": 0.0013351591920575145,
      "grad_norm": 0.0,
      "learning_rate": 8.893956670467504e-07,
      "loss": 13.8438,
      "step": 39
    },
    {
      "epoch": 0.0013693940431359123,
      "grad_norm": 0.0,
      "learning_rate": 9.122006841505132e-07,
      "loss": 13.7697,
      "step": 40
    },
    {
      "epoch": 0.00140362889421431,
      "grad_norm": 0.0,
      "learning_rate": 9.35005701254276e-07,
      "loss": 13.713,
      "step": 41
    },
    {
      "epoch": 0.001437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 9.578107183580388e-07,
      "loss": 13.7485,
      "step": 42
    },
    {
      "epoch": 0.0014720985963711057,
      "grad_norm": 0.0,
      "learning_rate": 9.806157354618017e-07,
      "loss": 13.6831,
      "step": 43
    },
    {
      "epoch": 0.0015063334474495036,
      "grad_norm": 0.0,
      "learning_rate": 1.0034207525655646e-06,
      "loss": 13.6553,
      "step": 44
    },
    {
      "epoch": 0.0015405682985279014,
      "grad_norm": 0.0,
      "learning_rate": 1.0262257696693273e-06,
      "loss": 13.5975,
      "step": 45
    },
    {
      "epoch": 0.0015748031496062992,
      "grad_norm": 0.0,
      "learning_rate": 1.0490307867730903e-06,
      "loss": 13.6359,
      "step": 46
    },
    {
      "epoch": 0.001609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 1.071835803876853e-06,
      "loss": 13.5956,
      "step": 47
    },
    {
      "epoch": 0.0016432728517630948,
      "grad_norm": 0.0,
      "learning_rate": 1.0946408209806159e-06,
      "loss": 13.5445,
      "step": 48
    },
    {
      "epoch": 0.0016775077028414926,
      "grad_norm": 0.0,
      "learning_rate": 1.1174458380843786e-06,
      "loss": 13.5401,
      "step": 49
    },
    {
      "epoch": 0.0017117425539198905,
      "grad_norm": 0.0,
      "learning_rate": 1.1402508551881415e-06,
      "loss": 13.4559,
      "step": 50
    },
    {
      "epoch": 0.0017459774049982883,
      "grad_norm": 0.0,
      "learning_rate": 1.1630558722919042e-06,
      "loss": 13.3973,
      "step": 51
    },
    {
      "epoch": 0.001780212256076686,
      "grad_norm": 0.0,
      "learning_rate": 1.1858608893956671e-06,
      "loss": 13.4151,
      "step": 52
    },
    {
      "epoch": 0.001814447107155084,
      "grad_norm": 0.0,
      "learning_rate": 1.2086659064994299e-06,
      "loss": 13.3778,
      "step": 53
    },
    {
      "epoch": 0.0018486819582334817,
      "grad_norm": 0.0,
      "learning_rate": 1.2314709236031928e-06,
      "loss": 13.3314,
      "step": 54
    },
    {
      "epoch": 0.0018829168093118796,
      "grad_norm": 0.0,
      "learning_rate": 1.2542759407069557e-06,
      "loss": 13.2611,
      "step": 55
    },
    {
      "epoch": 0.0019171516603902774,
      "grad_norm": 0.0,
      "learning_rate": 1.2770809578107184e-06,
      "loss": 13.2678,
      "step": 56
    },
    {
      "epoch": 0.0019513865114686752,
      "grad_norm": 0.0,
      "learning_rate": 1.2998859749144813e-06,
      "loss": 13.2371,
      "step": 57
    },
    {
      "epoch": 0.001985621362547073,
      "grad_norm": 0.0,
      "learning_rate": 1.322690992018244e-06,
      "loss": 13.1974,
      "step": 58
    },
    {
      "epoch": 0.0020198562136254706,
      "grad_norm": 0.0,
      "learning_rate": 1.3454960091220072e-06,
      "loss": 13.2028,
      "step": 59
    },
    {
      "epoch": 0.0020540910647038686,
      "grad_norm": 0.0,
      "learning_rate": 1.3683010262257699e-06,
      "loss": 13.2047,
      "step": 60
    },
    {
      "epoch": 0.0020883259157822662,
      "grad_norm": 0.0,
      "learning_rate": 1.3911060433295326e-06,
      "loss": 13.1855,
      "step": 61
    },
    {
      "epoch": 0.0021225607668606643,
      "grad_norm": 0.0,
      "learning_rate": 1.4139110604332953e-06,
      "loss": 13.1385,
      "step": 62
    },
    {
      "epoch": 0.002156795617939062,
      "grad_norm": 0.0,
      "learning_rate": 1.4367160775370584e-06,
      "loss": 13.1529,
      "step": 63
    },
    {
      "epoch": 0.00219103046901746,
      "grad_norm": 0.0,
      "learning_rate": 1.4595210946408211e-06,
      "loss": 13.1052,
      "step": 64
    },
    {
      "epoch": 0.0022252653200958575,
      "grad_norm": 0.0,
      "learning_rate": 1.4823261117445838e-06,
      "loss": 13.0344,
      "step": 65
    },
    {
      "epoch": 0.0022595001711742555,
      "grad_norm": 0.0,
      "learning_rate": 1.5051311288483467e-06,
      "loss": 13.079,
      "step": 66
    },
    {
      "epoch": 0.002293735022252653,
      "grad_norm": 0.0,
      "learning_rate": 1.5279361459521097e-06,
      "loss": 13.0506,
      "step": 67
    },
    {
      "epoch": 0.002327969873331051,
      "grad_norm": 0.0,
      "learning_rate": 1.5507411630558724e-06,
      "loss": 13.0298,
      "step": 68
    },
    {
      "epoch": 0.002362204724409449,
      "grad_norm": 0.0,
      "learning_rate": 1.5735461801596353e-06,
      "loss": 13.0197,
      "step": 69
    },
    {
      "epoch": 0.002396439575487847,
      "grad_norm": 0.0,
      "learning_rate": 1.596351197263398e-06,
      "loss": 13.0331,
      "step": 70
    },
    {
      "epoch": 0.0024306744265662444,
      "grad_norm": 0.0,
      "learning_rate": 1.6191562143671611e-06,
      "loss": 12.9746,
      "step": 71
    },
    {
      "epoch": 0.0024649092776446425,
      "grad_norm": 0.0,
      "learning_rate": 1.6419612314709238e-06,
      "loss": 12.9919,
      "step": 72
    },
    {
      "epoch": 0.00249914412872304,
      "grad_norm": 0.0,
      "learning_rate": 1.6647662485746865e-06,
      "loss": 12.9697,
      "step": 73
    },
    {
      "epoch": 0.0025333789798014377,
      "grad_norm": 0.0,
      "learning_rate": 1.6875712656784492e-06,
      "loss": 12.9525,
      "step": 74
    },
    {
      "epoch": 0.0025676138308798357,
      "grad_norm": 0.0,
      "learning_rate": 1.7103762827822124e-06,
      "loss": 12.9016,
      "step": 75
    },
    {
      "epoch": 0.0026018486819582333,
      "grad_norm": 0.0,
      "learning_rate": 1.733181299885975e-06,
      "loss": 12.866,
      "step": 76
    },
    {
      "epoch": 0.0026360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 1.7559863169897378e-06,
      "loss": 12.8592,
      "step": 77
    },
    {
      "epoch": 0.002670318384115029,
      "grad_norm": 0.0,
      "learning_rate": 1.7787913340935007e-06,
      "loss": 12.7891,
      "step": 78
    },
    {
      "epoch": 0.002704553235193427,
      "grad_norm": 0.0,
      "learning_rate": 1.8015963511972636e-06,
      "loss": 12.7798,
      "step": 79
    },
    {
      "epoch": 0.0027387880862718246,
      "grad_norm": 0.0,
      "learning_rate": 1.8244013683010263e-06,
      "loss": 12.7816,
      "step": 80
    },
    {
      "epoch": 0.0027730229373502226,
      "grad_norm": 0.0,
      "learning_rate": 1.8472063854047893e-06,
      "loss": 12.799,
      "step": 81
    },
    {
      "epoch": 0.00280725778842862,
      "grad_norm": 0.0,
      "learning_rate": 1.870011402508552e-06,
      "loss": 12.7246,
      "step": 82
    },
    {
      "epoch": 0.0028414926395070182,
      "grad_norm": 0.0,
      "learning_rate": 1.8928164196123149e-06,
      "loss": 12.7012,
      "step": 83
    },
    {
      "epoch": 0.002875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 1.9156214367160776e-06,
      "loss": 12.6992,
      "step": 84
    },
    {
      "epoch": 0.002909962341663814,
      "grad_norm": 0.0,
      "learning_rate": 1.9384264538198407e-06,
      "loss": 12.7006,
      "step": 85
    },
    {
      "epoch": 0.0029441971927422115,
      "grad_norm": 0.0,
      "learning_rate": 1.9612314709236034e-06,
      "loss": 12.6914,
      "step": 86
    },
    {
      "epoch": 0.0029784320438206095,
      "grad_norm": 0.0,
      "learning_rate": 1.984036488027366e-06,
      "loss": 12.5901,
      "step": 87
    },
    {
      "epoch": 0.003012666894899007,
      "grad_norm": 0.0,
      "learning_rate": 2.0068415051311293e-06,
      "loss": 12.6078,
      "step": 88
    },
    {
      "epoch": 0.003046901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2.029646522234892e-06,
      "loss": 12.5567,
      "step": 89
    },
    {
      "epoch": 0.0030811365970558027,
      "grad_norm": 0.0,
      "learning_rate": 2.0524515393386547e-06,
      "loss": 12.5746,
      "step": 90
    },
    {
      "epoch": 0.0031153714481342008,
      "grad_norm": 0.0,
      "learning_rate": 2.075256556442418e-06,
      "loss": 12.5977,
      "step": 91
    },
    {
      "epoch": 0.0031496062992125984,
      "grad_norm": 0.0,
      "learning_rate": 2.0980615735461805e-06,
      "loss": 12.567,
      "step": 92
    },
    {
      "epoch": 0.0031838411502909964,
      "grad_norm": 0.0,
      "learning_rate": 2.1208665906499432e-06,
      "loss": 12.5045,
      "step": 93
    },
    {
      "epoch": 0.003218076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2.143671607753706e-06,
      "loss": 12.4686,
      "step": 94
    },
    {
      "epoch": 0.003252310852447792,
      "grad_norm": 0.0,
      "learning_rate": 2.166476624857469e-06,
      "loss": 12.5066,
      "step": 95
    },
    {
      "epoch": 0.0032865457035261897,
      "grad_norm": 0.0,
      "learning_rate": 2.1892816419612318e-06,
      "loss": 12.4745,
      "step": 96
    },
    {
      "epoch": 0.0033207805546045873,
      "grad_norm": 0.0,
      "learning_rate": 2.2120866590649945e-06,
      "loss": 12.3975,
      "step": 97
    },
    {
      "epoch": 0.0033550154056829853,
      "grad_norm": 0.0,
      "learning_rate": 2.234891676168757e-06,
      "loss": 12.4296,
      "step": 98
    },
    {
      "epoch": 0.003389250256761383,
      "grad_norm": 0.0,
      "learning_rate": 2.25769669327252e-06,
      "loss": 12.3443,
      "step": 99
    },
    {
      "epoch": 0.003423485107839781,
      "grad_norm": 0.0,
      "learning_rate": 2.280501710376283e-06,
      "loss": 12.3815,
      "step": 100
    },
    {
      "epoch": 0.0034577199589181785,
      "grad_norm": 0.0,
      "learning_rate": 2.3033067274800457e-06,
      "loss": 12.2967,
      "step": 101
    },
    {
      "epoch": 0.0034919548099965766,
      "grad_norm": 0.0,
      "learning_rate": 2.3261117445838084e-06,
      "loss": 12.3305,
      "step": 102
    },
    {
      "epoch": 0.003526189661074974,
      "grad_norm": 0.0,
      "learning_rate": 2.348916761687571e-06,
      "loss": 12.2826,
      "step": 103
    },
    {
      "epoch": 0.003560424512153372,
      "grad_norm": 0.0,
      "learning_rate": 2.3717217787913343e-06,
      "loss": 12.3055,
      "step": 104
    },
    {
      "epoch": 0.00359465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2.394526795895097e-06,
      "loss": 12.2532,
      "step": 105
    },
    {
      "epoch": 0.003628894214310168,
      "grad_norm": 0.0,
      "learning_rate": 2.4173318129988597e-06,
      "loss": 12.2598,
      "step": 106
    },
    {
      "epoch": 0.0036631290653885654,
      "grad_norm": 0.0,
      "learning_rate": 2.440136830102623e-06,
      "loss": 12.2127,
      "step": 107
    },
    {
      "epoch": 0.0036973639164669635,
      "grad_norm": 0.0,
      "learning_rate": 2.4629418472063855e-06,
      "loss": 12.1973,
      "step": 108
    },
    {
      "epoch": 0.003731598767545361,
      "grad_norm": 0.0,
      "learning_rate": 2.4857468643101482e-06,
      "loss": 12.1387,
      "step": 109
    },
    {
      "epoch": 0.003765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2.5085518814139114e-06,
      "loss": 12.1154,
      "step": 110
    },
    {
      "epoch": 0.0038000684697021567,
      "grad_norm": 0.0,
      "learning_rate": 2.531356898517674e-06,
      "loss": 12.1051,
      "step": 111
    },
    {
      "epoch": 0.0038343033207805547,
      "grad_norm": 0.0,
      "learning_rate": 2.554161915621437e-06,
      "loss": 12.0488,
      "step": 112
    },
    {
      "epoch": 0.0038685381718589523,
      "grad_norm": 0.0,
      "learning_rate": 2.5769669327251995e-06,
      "loss": 12.0142,
      "step": 113
    },
    {
      "epoch": 0.0039027730229373504,
      "grad_norm": 0.0,
      "learning_rate": 2.5997719498289626e-06,
      "loss": 11.9763,
      "step": 114
    },
    {
      "epoch": 0.003937007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2.6225769669327258e-06,
      "loss": 11.9788,
      "step": 115
    },
    {
      "epoch": 0.003971242725094146,
      "grad_norm": 0.0,
      "learning_rate": 2.645381984036488e-06,
      "loss": 11.9041,
      "step": 116
    },
    {
      "epoch": 0.004005477576172544,
      "grad_norm": 0.0,
      "learning_rate": 2.668187001140251e-06,
      "loss": 11.8885,
      "step": 117
    },
    {
      "epoch": 0.004039712427250941,
      "grad_norm": 0.0,
      "learning_rate": 2.6909920182440143e-06,
      "loss": 11.8413,
      "step": 118
    },
    {
      "epoch": 0.004073947278329339,
      "grad_norm": 0.0,
      "learning_rate": 2.7137970353477766e-06,
      "loss": 11.8186,
      "step": 119
    },
    {
      "epoch": 0.004108182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2.7366020524515397e-06,
      "loss": 11.8007,
      "step": 120
    },
    {
      "epoch": 0.004142416980486135,
      "grad_norm": 0.0,
      "learning_rate": 2.759407069555302e-06,
      "loss": 11.8273,
      "step": 121
    },
    {
      "epoch": 0.0041766518315645325,
      "grad_norm": 0.0,
      "learning_rate": 2.782212086659065e-06,
      "loss": 11.7073,
      "step": 122
    },
    {
      "epoch": 0.0042108866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2.8050171037628283e-06,
      "loss": 11.6982,
      "step": 123
    },
    {
      "epoch": 0.0042451215337213286,
      "grad_norm": 0.0,
      "learning_rate": 2.8278221208665906e-06,
      "loss": 11.6873,
      "step": 124
    },
    {
      "epoch": 0.004279356384799726,
      "grad_norm": 0.0,
      "learning_rate": 2.8506271379703537e-06,
      "loss": 11.6757,
      "step": 125
    },
    {
      "epoch": 0.004313591235878124,
      "grad_norm": 0.0,
      "learning_rate": 2.873432155074117e-06,
      "loss": 11.6086,
      "step": 126
    },
    {
      "epoch": 0.004347826086956522,
      "grad_norm": 0.0,
      "learning_rate": 2.896237172177879e-06,
      "loss": 11.5616,
      "step": 127
    },
    {
      "epoch": 0.00438206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2.9190421892816422e-06,
      "loss": 11.5754,
      "step": 128
    },
    {
      "epoch": 0.004416295789113317,
      "grad_norm": 0.0,
      "learning_rate": 2.941847206385405e-06,
      "loss": 11.5319,
      "step": 129
    },
    {
      "epoch": 0.004450530640191715,
      "grad_norm": 0.0,
      "learning_rate": 2.9646522234891676e-06,
      "loss": 11.4735,
      "step": 130
    },
    {
      "epoch": 0.004484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2.9874572405929308e-06,
      "loss": 11.4144,
      "step": 131
    },
    {
      "epoch": 0.004519000342348511,
      "grad_norm": 0.0,
      "learning_rate": 3.0102622576966935e-06,
      "loss": 11.4144,
      "step": 132
    },
    {
      "epoch": 0.004553235193426908,
      "grad_norm": 0.0,
      "learning_rate": 3.033067274800456e-06,
      "loss": 11.3439,
      "step": 133
    },
    {
      "epoch": 0.004587470044505306,
      "grad_norm": 0.0,
      "learning_rate": 3.0558722919042193e-06,
      "loss": 11.2345,
      "step": 134
    },
    {
      "epoch": 0.004621704895583704,
      "grad_norm": 0.0,
      "learning_rate": 3.078677309007982e-06,
      "loss": 11.2309,
      "step": 135
    },
    {
      "epoch": 0.004655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 3.1014823261117447e-06,
      "loss": 11.1568,
      "step": 136
    },
    {
      "epoch": 0.0046901745977404995,
      "grad_norm": 0.0,
      "learning_rate": 3.1242873432155074e-06,
      "loss": 11.1392,
      "step": 137
    },
    {
      "epoch": 0.004724409448818898,
      "grad_norm": 0.0,
      "learning_rate": 3.1470923603192706e-06,
      "loss": 11.1451,
      "step": 138
    },
    {
      "epoch": 0.004758644299897296,
      "grad_norm": 0.0,
      "learning_rate": 3.1698973774230333e-06,
      "loss": 11.0842,
      "step": 139
    },
    {
      "epoch": 0.004792879150975694,
      "grad_norm": 0.0,
      "learning_rate": 3.192702394526796e-06,
      "loss": 10.9917,
      "step": 140
    },
    {
      "epoch": 0.004827114002054091,
      "grad_norm": 0.0,
      "learning_rate": 3.215507411630559e-06,
      "loss": 10.9778,
      "step": 141
    },
    {
      "epoch": 0.004861348853132489,
      "grad_norm": 0.0,
      "learning_rate": 3.2383124287343223e-06,
      "loss": 10.9127,
      "step": 142
    },
    {
      "epoch": 0.004895583704210887,
      "grad_norm": 0.0,
      "learning_rate": 3.2611174458380845e-06,
      "loss": 10.8538,
      "step": 143
    },
    {
      "epoch": 0.004929818555289285,
      "grad_norm": 0.0,
      "learning_rate": 3.2839224629418477e-06,
      "loss": 10.7905,
      "step": 144
    },
    {
      "epoch": 0.004964053406367682,
      "grad_norm": 0.0,
      "learning_rate": 3.30672748004561e-06,
      "loss": 10.7589,
      "step": 145
    },
    {
      "epoch": 0.00499828825744608,
      "grad_norm": 0.0,
      "learning_rate": 3.329532497149373e-06,
      "loss": 10.7099,
      "step": 146
    },
    {
      "epoch": 0.005032523108524478,
      "grad_norm": 0.0,
      "learning_rate": 3.3523375142531362e-06,
      "loss": 10.6294,
      "step": 147
    },
    {
      "epoch": 0.005066757959602875,
      "grad_norm": 0.0,
      "learning_rate": 3.3751425313568985e-06,
      "loss": 10.5788,
      "step": 148
    },
    {
      "epoch": 0.005100992810681273,
      "grad_norm": 0.0,
      "learning_rate": 3.3979475484606616e-06,
      "loss": 10.5198,
      "step": 149
    },
    {
      "epoch": 0.005135227661759671,
      "grad_norm": 0.0,
      "learning_rate": 3.4207525655644248e-06,
      "loss": 10.4927,
      "step": 150
    },
    {
      "epoch": 0.005169462512838069,
      "grad_norm": 0.0,
      "learning_rate": 3.443557582668187e-06,
      "loss": 10.3712,
      "step": 151
    },
    {
      "epoch": 0.005203697363916467,
      "grad_norm": 0.0,
      "learning_rate": 3.46636259977195e-06,
      "loss": 10.3287,
      "step": 152
    },
    {
      "epoch": 0.005237932214994865,
      "grad_norm": 0.0,
      "learning_rate": 3.489167616875713e-06,
      "loss": 10.2728,
      "step": 153
    },
    {
      "epoch": 0.005272167066073263,
      "grad_norm": 0.0,
      "learning_rate": 3.5119726339794756e-06,
      "loss": 10.2248,
      "step": 154
    },
    {
      "epoch": 0.005306401917151661,
      "grad_norm": 0.0,
      "learning_rate": 3.5347776510832387e-06,
      "loss": 10.1053,
      "step": 155
    },
    {
      "epoch": 0.005340636768230058,
      "grad_norm": 0.0,
      "learning_rate": 3.5575826681870014e-06,
      "loss": 10.0951,
      "step": 156
    },
    {
      "epoch": 0.005374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 3.580387685290764e-06,
      "loss": 10.021,
      "step": 157
    },
    {
      "epoch": 0.005409106470386854,
      "grad_norm": 0.0,
      "learning_rate": 3.6031927023945273e-06,
      "loss": 9.9165,
      "step": 158
    },
    {
      "epoch": 0.005443341321465252,
      "grad_norm": 0.0,
      "learning_rate": 3.62599771949829e-06,
      "loss": 9.8768,
      "step": 159
    },
    {
      "epoch": 0.005477576172543649,
      "grad_norm": 0.0,
      "learning_rate": 3.6488027366020527e-06,
      "loss": 9.7713,
      "step": 160
    },
    {
      "epoch": 0.005511811023622047,
      "grad_norm": 0.0,
      "learning_rate": 3.6716077537058154e-06,
      "loss": 9.7465,
      "step": 161
    },
    {
      "epoch": 0.005546045874700445,
      "grad_norm": 0.0,
      "learning_rate": 3.6944127708095785e-06,
      "loss": 9.6846,
      "step": 162
    },
    {
      "epoch": 0.005580280725778843,
      "grad_norm": 0.0,
      "learning_rate": 3.7172177879133412e-06,
      "loss": 9.5294,
      "step": 163
    },
    {
      "epoch": 0.00561451557685724,
      "grad_norm": 0.0,
      "learning_rate": 3.740022805017104e-06,
      "loss": 9.537,
      "step": 164
    },
    {
      "epoch": 0.0056487504279356384,
      "grad_norm": 0.0,
      "learning_rate": 3.762827822120867e-06,
      "loss": 9.47,
      "step": 165
    },
    {
      "epoch": 0.0056829852790140365,
      "grad_norm": 0.0,
      "learning_rate": 3.7856328392246298e-06,
      "loss": 9.3593,
      "step": 166
    },
    {
      "epoch": 0.0057172201300924345,
      "grad_norm": 0.0,
      "learning_rate": 3.8084378563283925e-06,
      "loss": 9.2919,
      "step": 167
    },
    {
      "epoch": 0.005751454981170832,
      "grad_norm": 0.0,
      "learning_rate": 3.831242873432155e-06,
      "loss": 9.2205,
      "step": 168
    },
    {
      "epoch": 0.00578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 3.854047890535918e-06,
      "loss": 9.2157,
      "step": 169
    },
    {
      "epoch": 0.005819924683327628,
      "grad_norm": 0.0,
      "learning_rate": 3.8768529076396815e-06,
      "loss": 9.0748,
      "step": 170
    },
    {
      "epoch": 0.005854159534406025,
      "grad_norm": 0.0,
      "learning_rate": 3.899657924743444e-06,
      "loss": 9.0083,
      "step": 171
    },
    {
      "epoch": 0.005888394385484423,
      "grad_norm": 0.0,
      "learning_rate": 3.922462941847207e-06,
      "loss": 8.987,
      "step": 172
    },
    {
      "epoch": 0.005922629236562821,
      "grad_norm": 0.0,
      "learning_rate": 3.94526795895097e-06,
      "loss": 8.8983,
      "step": 173
    },
    {
      "epoch": 0.005956864087641219,
      "grad_norm": 0.0,
      "learning_rate": 3.968072976054732e-06,
      "loss": 8.8113,
      "step": 174
    },
    {
      "epoch": 0.005991098938719616,
      "grad_norm": 0.0,
      "learning_rate": 3.990877993158495e-06,
      "loss": 8.7419,
      "step": 175
    },
    {
      "epoch": 0.006025333789798014,
      "grad_norm": 0.0,
      "learning_rate": 4.0136830102622585e-06,
      "loss": 8.6601,
      "step": 176
    },
    {
      "epoch": 0.006059568640876412,
      "grad_norm": 0.0,
      "learning_rate": 4.036488027366021e-06,
      "loss": 8.5807,
      "step": 177
    },
    {
      "epoch": 0.00609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 4.059293044469784e-06,
      "loss": 8.5539,
      "step": 178
    },
    {
      "epoch": 0.0061280383430332075,
      "grad_norm": 0.0,
      "learning_rate": 4.082098061573547e-06,
      "loss": 8.438,
      "step": 179
    },
    {
      "epoch": 0.0061622731941116055,
      "grad_norm": 0.0,
      "learning_rate": 4.104903078677309e-06,
      "loss": 8.3772,
      "step": 180
    },
    {
      "epoch": 0.0061965080451900035,
      "grad_norm": 0.0,
      "learning_rate": 4.1277080957810725e-06,
      "loss": 8.275,
      "step": 181
    },
    {
      "epoch": 0.0062307428962684016,
      "grad_norm": 0.0,
      "learning_rate": 4.150513112884836e-06,
      "loss": 8.189,
      "step": 182
    },
    {
      "epoch": 0.006264977747346799,
      "grad_norm": 0.0,
      "learning_rate": 4.173318129988598e-06,
      "loss": 8.18,
      "step": 183
    },
    {
      "epoch": 0.006299212598425197,
      "grad_norm": 0.0,
      "learning_rate": 4.196123147092361e-06,
      "loss": 8.0723,
      "step": 184
    },
    {
      "epoch": 0.006333447449503595,
      "grad_norm": 0.0,
      "learning_rate": 4.218928164196123e-06,
      "loss": 8.0351,
      "step": 185
    },
    {
      "epoch": 0.006367682300581993,
      "grad_norm": 0.0,
      "learning_rate": 4.2417331812998865e-06,
      "loss": 7.9146,
      "step": 186
    },
    {
      "epoch": 0.00640191715166039,
      "grad_norm": 0.0,
      "learning_rate": 4.26453819840365e-06,
      "loss": 7.8746,
      "step": 187
    },
    {
      "epoch": 0.006436152002738788,
      "grad_norm": 0.0,
      "learning_rate": 4.287343215507412e-06,
      "loss": 7.7714,
      "step": 188
    },
    {
      "epoch": 0.006470386853817186,
      "grad_norm": 0.0,
      "learning_rate": 4.310148232611175e-06,
      "loss": 7.7248,
      "step": 189
    },
    {
      "epoch": 0.006504621704895584,
      "grad_norm": 0.0,
      "learning_rate": 4.332953249714938e-06,
      "loss": 7.6204,
      "step": 190
    },
    {
      "epoch": 0.006538856555973981,
      "grad_norm": 0.0,
      "learning_rate": 4.3557582668187004e-06,
      "loss": 7.5722,
      "step": 191
    },
    {
      "epoch": 0.006573091407052379,
      "grad_norm": 0.0,
      "learning_rate": 4.3785632839224636e-06,
      "loss": 7.4425,
      "step": 192
    },
    {
      "epoch": 0.006607326258130777,
      "grad_norm": 0.0,
      "learning_rate": 4.401368301026226e-06,
      "loss": 7.4246,
      "step": 193
    },
    {
      "epoch": 0.0066415611092091745,
      "grad_norm": 0.0,
      "learning_rate": 4.424173318129989e-06,
      "loss": 7.3293,
      "step": 194
    },
    {
      "epoch": 0.0066757959602875725,
      "grad_norm": 0.0,
      "learning_rate": 4.446978335233752e-06,
      "loss": 7.2893,
      "step": 195
    },
    {
      "epoch": 0.006710030811365971,
      "grad_norm": 0.0,
      "learning_rate": 4.469783352337514e-06,
      "loss": 7.1756,
      "step": 196
    },
    {
      "epoch": 0.006744265662444369,
      "grad_norm": 0.0,
      "learning_rate": 4.4925883694412775e-06,
      "loss": 7.0833,
      "step": 197
    },
    {
      "epoch": 0.006778500513522766,
      "grad_norm": 0.0,
      "learning_rate": 4.51539338654504e-06,
      "loss": 6.9904,
      "step": 198
    },
    {
      "epoch": 0.006812735364601164,
      "grad_norm": 0.0,
      "learning_rate": 4.538198403648803e-06,
      "loss": 6.9277,
      "step": 199
    },
    {
      "epoch": 0.006846970215679562,
      "grad_norm": 0.0,
      "learning_rate": 4.561003420752566e-06,
      "loss": 6.903,
      "step": 200
    },
    {
      "epoch": 0.00688120506675796,
      "grad_norm": 0.0,
      "learning_rate": 4.583808437856328e-06,
      "loss": 6.7679,
      "step": 201
    },
    {
      "epoch": 0.006915439917836357,
      "grad_norm": 0.0,
      "learning_rate": 4.6066134549600915e-06,
      "loss": 6.716,
      "step": 202
    },
    {
      "epoch": 0.006949674768914755,
      "grad_norm": 0.0,
      "learning_rate": 4.629418472063855e-06,
      "loss": 6.6204,
      "step": 203
    },
    {
      "epoch": 0.006983909619993153,
      "grad_norm": 0.0,
      "learning_rate": 4.652223489167617e-06,
      "loss": 6.5151,
      "step": 204
    },
    {
      "epoch": 0.007018144471071551,
      "grad_norm": 0.0,
      "learning_rate": 4.67502850627138e-06,
      "loss": 6.4573,
      "step": 205
    },
    {
      "epoch": 0.007052379322149948,
      "grad_norm": 0.0,
      "learning_rate": 4.697833523375142e-06,
      "loss": 6.4229,
      "step": 206
    },
    {
      "epoch": 0.007086614173228346,
      "grad_norm": 0.0,
      "learning_rate": 4.7206385404789054e-06,
      "loss": 6.3133,
      "step": 207
    },
    {
      "epoch": 0.007120849024306744,
      "grad_norm": 0.0,
      "learning_rate": 4.7434435575826686e-06,
      "loss": 6.2777,
      "step": 208
    },
    {
      "epoch": 0.007155083875385142,
      "grad_norm": 0.0,
      "learning_rate": 4.766248574686431e-06,
      "loss": 6.2085,
      "step": 209
    },
    {
      "epoch": 0.00718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 4.789053591790194e-06,
      "loss": 6.08,
      "step": 210
    },
    {
      "epoch": 0.007223553577541938,
      "grad_norm": 0.0,
      "learning_rate": 4.811858608893957e-06,
      "loss": 6.0608,
      "step": 211
    },
    {
      "epoch": 0.007257788428620336,
      "grad_norm": 0.0,
      "learning_rate": 4.834663625997719e-06,
      "loss": 5.9678,
      "step": 212
    },
    {
      "epoch": 0.007292023279698734,
      "grad_norm": 0.0,
      "learning_rate": 4.8574686431014825e-06,
      "loss": 5.9577,
      "step": 213
    },
    {
      "epoch": 0.007326258130777131,
      "grad_norm": 0.0,
      "learning_rate": 4.880273660205246e-06,
      "loss": 5.814,
      "step": 214
    },
    {
      "epoch": 0.007360492981855529,
      "grad_norm": 0.0,
      "learning_rate": 4.903078677309008e-06,
      "loss": 5.7939,
      "step": 215
    },
    {
      "epoch": 0.007394727832933927,
      "grad_norm": 0.0,
      "learning_rate": 4.925883694412771e-06,
      "loss": 5.7234,
      "step": 216
    },
    {
      "epoch": 0.007428962684012325,
      "grad_norm": 0.0,
      "learning_rate": 4.948688711516534e-06,
      "loss": 5.7115,
      "step": 217
    },
    {
      "epoch": 0.007463197535090722,
      "grad_norm": 0.0,
      "learning_rate": 4.9714937286202965e-06,
      "loss": 5.6385,
      "step": 218
    },
    {
      "epoch": 0.00749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 4.99429874572406e-06,
      "loss": 5.551,
      "step": 219
    },
    {
      "epoch": 0.007531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 5.017103762827823e-06,
      "loss": 5.4802,
      "step": 220
    },
    {
      "epoch": 0.007565902088325915,
      "grad_norm": 0.0,
      "learning_rate": 5.039908779931585e-06,
      "loss": 5.3928,
      "step": 221
    },
    {
      "epoch": 0.007600136939404313,
      "grad_norm": 0.0,
      "learning_rate": 5.062713797035348e-06,
      "loss": 5.4128,
      "step": 222
    },
    {
      "epoch": 0.0076343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 5.085518814139111e-06,
      "loss": 5.3814,
      "step": 223
    },
    {
      "epoch": 0.0076686066415611095,
      "grad_norm": 0.0,
      "learning_rate": 5.108323831242874e-06,
      "loss": 5.314,
      "step": 224
    },
    {
      "epoch": 0.007702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 5.131128848346637e-06,
      "loss": 5.2669,
      "step": 225
    },
    {
      "epoch": 0.007737076343717905,
      "grad_norm": 0.0,
      "learning_rate": 5.153933865450399e-06,
      "loss": 5.2501,
      "step": 226
    },
    {
      "epoch": 0.007771311194796303,
      "grad_norm": 0.0,
      "learning_rate": 5.176738882554162e-06,
      "loss": 5.2234,
      "step": 227
    },
    {
      "epoch": 0.007805546045874701,
      "grad_norm": 0.0,
      "learning_rate": 5.199543899657925e-06,
      "loss": 5.1776,
      "step": 228
    },
    {
      "epoch": 0.007839780896953098,
      "grad_norm": 0.0,
      "learning_rate": 5.222348916761688e-06,
      "loss": 5.1113,
      "step": 229
    },
    {
      "epoch": 0.007874015748031496,
      "grad_norm": 0.0,
      "learning_rate": 5.2451539338654515e-06,
      "loss": 5.1042,
      "step": 230
    },
    {
      "epoch": 0.007908250599109894,
      "grad_norm": 0.0,
      "learning_rate": 5.267958950969214e-06,
      "loss": 5.0469,
      "step": 231
    },
    {
      "epoch": 0.007942485450188292,
      "grad_norm": 0.0,
      "learning_rate": 5.290763968072976e-06,
      "loss": 5.0439,
      "step": 232
    },
    {
      "epoch": 0.00797672030126669,
      "grad_norm": 0.0,
      "learning_rate": 5.313568985176739e-06,
      "loss": 5.0323,
      "step": 233
    },
    {
      "epoch": 0.008010955152345088,
      "grad_norm": 0.0,
      "learning_rate": 5.336374002280502e-06,
      "loss": 4.9716,
      "step": 234
    },
    {
      "epoch": 0.008045190003423484,
      "grad_norm": 0.0,
      "learning_rate": 5.3591790193842655e-06,
      "loss": 4.9236,
      "step": 235
    },
    {
      "epoch": 0.008079424854501882,
      "grad_norm": 0.0,
      "learning_rate": 5.381984036488029e-06,
      "loss": 4.9915,
      "step": 236
    },
    {
      "epoch": 0.00811365970558028,
      "grad_norm": 0.0,
      "learning_rate": 5.40478905359179e-06,
      "loss": 5.0077,
      "step": 237
    },
    {
      "epoch": 0.008147894556658678,
      "grad_norm": 0.0,
      "learning_rate": 5.427594070695553e-06,
      "loss": 4.9315,
      "step": 238
    },
    {
      "epoch": 0.008182129407737077,
      "grad_norm": 0.0,
      "learning_rate": 5.450399087799316e-06,
      "loss": 4.8879,
      "step": 239
    },
    {
      "epoch": 0.008216364258815475,
      "grad_norm": 0.0,
      "learning_rate": 5.4732041049030794e-06,
      "loss": 4.848,
      "step": 240
    },
    {
      "epoch": 0.008250599109893873,
      "grad_norm": 0.0,
      "learning_rate": 5.496009122006843e-06,
      "loss": 4.8794,
      "step": 241
    },
    {
      "epoch": 0.00828483396097227,
      "grad_norm": 0.0,
      "learning_rate": 5.518814139110604e-06,
      "loss": 4.8529,
      "step": 242
    },
    {
      "epoch": 0.008319068812050667,
      "grad_norm": 0.0,
      "learning_rate": 5.541619156214367e-06,
      "loss": 4.8483,
      "step": 243
    },
    {
      "epoch": 0.008353303663129065,
      "grad_norm": 0.0,
      "learning_rate": 5.56442417331813e-06,
      "loss": 4.8921,
      "step": 244
    },
    {
      "epoch": 0.008387538514207463,
      "grad_norm": 0.0,
      "learning_rate": 5.587229190421893e-06,
      "loss": 4.8137,
      "step": 245
    },
    {
      "epoch": 0.008421773365285861,
      "grad_norm": 0.0,
      "learning_rate": 5.6100342075256565e-06,
      "loss": 4.8151,
      "step": 246
    },
    {
      "epoch": 0.008456008216364259,
      "grad_norm": 0.0,
      "learning_rate": 5.63283922462942e-06,
      "loss": 4.7709,
      "step": 247
    },
    {
      "epoch": 0.008490243067442657,
      "grad_norm": 0.0,
      "learning_rate": 5.655644241733181e-06,
      "loss": 4.7937,
      "step": 248
    },
    {
      "epoch": 0.008524477918521055,
      "grad_norm": 0.0,
      "learning_rate": 5.678449258836944e-06,
      "loss": 4.7492,
      "step": 249
    },
    {
      "epoch": 0.008558712769599451,
      "grad_norm": 0.0,
      "learning_rate": 5.701254275940707e-06,
      "loss": 4.7135,
      "step": 250
    },
    {
      "epoch": 0.00859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 5.7240592930444705e-06,
      "loss": 4.7496,
      "step": 251
    },
    {
      "epoch": 0.008627182471756248,
      "grad_norm": 0.0,
      "learning_rate": 5.746864310148234e-06,
      "loss": 4.6829,
      "step": 252
    },
    {
      "epoch": 0.008661417322834646,
      "grad_norm": 0.0,
      "learning_rate": 5.769669327251995e-06,
      "loss": 4.7136,
      "step": 253
    },
    {
      "epoch": 0.008695652173913044,
      "grad_norm": 0.0,
      "learning_rate": 5.792474344355758e-06,
      "loss": 4.6901,
      "step": 254
    },
    {
      "epoch": 0.008729887024991442,
      "grad_norm": 0.0,
      "learning_rate": 5.815279361459521e-06,
      "loss": 4.6799,
      "step": 255
    },
    {
      "epoch": 0.00876412187606984,
      "grad_norm": 0.0,
      "learning_rate": 5.8380843785632845e-06,
      "loss": 4.6995,
      "step": 256
    },
    {
      "epoch": 0.008798356727148238,
      "grad_norm": 0.0,
      "learning_rate": 5.860889395667048e-06,
      "loss": 4.6567,
      "step": 257
    },
    {
      "epoch": 0.008832591578226634,
      "grad_norm": 0.0,
      "learning_rate": 5.88369441277081e-06,
      "loss": 4.6226,
      "step": 258
    },
    {
      "epoch": 0.008866826429305032,
      "grad_norm": 0.0,
      "learning_rate": 5.906499429874573e-06,
      "loss": 4.6138,
      "step": 259
    },
    {
      "epoch": 0.00890106128038343,
      "grad_norm": 0.0,
      "learning_rate": 5.929304446978335e-06,
      "loss": 4.6404,
      "step": 260
    },
    {
      "epoch": 0.008935296131461828,
      "grad_norm": 0.0,
      "learning_rate": 5.9521094640820984e-06,
      "loss": 4.5885,
      "step": 261
    },
    {
      "epoch": 0.008969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 5.9749144811858616e-06,
      "loss": 4.6169,
      "step": 262
    },
    {
      "epoch": 0.009003765833618624,
      "grad_norm": 0.0,
      "learning_rate": 5.997719498289625e-06,
      "loss": 4.8227,
      "step": 263
    },
    {
      "epoch": 0.009038000684697022,
      "grad_norm": 0.0,
      "learning_rate": 6.020524515393387e-06,
      "loss": 4.8149,
      "step": 264
    },
    {
      "epoch": 0.00907223553577542,
      "grad_norm": 0.0,
      "learning_rate": 6.04332953249715e-06,
      "loss": 4.9912,
      "step": 265
    },
    {
      "epoch": 0.009106470386853817,
      "grad_norm": 0.0,
      "learning_rate": 6.066134549600912e-06,
      "loss": 4.9474,
      "step": 266
    },
    {
      "epoch": 0.009140705237932215,
      "grad_norm": 0.0,
      "learning_rate": 6.0889395667046755e-06,
      "loss": 4.8881,
      "step": 267
    },
    {
      "epoch": 0.009174940089010613,
      "grad_norm": 0.0,
      "learning_rate": 6.111744583808439e-06,
      "loss": 4.8854,
      "step": 268
    },
    {
      "epoch": 0.00920917494008901,
      "grad_norm": 0.0,
      "learning_rate": 6.134549600912201e-06,
      "loss": 4.8643,
      "step": 269
    },
    {
      "epoch": 0.009243409791167409,
      "grad_norm": 0.0,
      "learning_rate": 6.157354618015964e-06,
      "loss": 4.7798,
      "step": 270
    },
    {
      "epoch": 0.009277644642245807,
      "grad_norm": 0.0,
      "learning_rate": 6.180159635119727e-06,
      "loss": 4.8092,
      "step": 271
    },
    {
      "epoch": 0.009311879493324205,
      "grad_norm": 0.0,
      "learning_rate": 6.2029646522234895e-06,
      "loss": 4.7451,
      "step": 272
    },
    {
      "epoch": 0.009346114344402601,
      "grad_norm": 0.0,
      "learning_rate": 6.225769669327253e-06,
      "loss": 4.7062,
      "step": 273
    },
    {
      "epoch": 0.009380349195480999,
      "grad_norm": 0.0,
      "learning_rate": 6.248574686431015e-06,
      "loss": 4.7327,
      "step": 274
    },
    {
      "epoch": 0.009414584046559397,
      "grad_norm": 0.0,
      "learning_rate": 6.271379703534778e-06,
      "loss": 4.7108,
      "step": 275
    },
    {
      "epoch": 0.009448818897637795,
      "grad_norm": 0.0,
      "learning_rate": 6.294184720638541e-06,
      "loss": 4.6231,
      "step": 276
    },
    {
      "epoch": 0.009483053748716193,
      "grad_norm": 0.0,
      "learning_rate": 6.316989737742304e-06,
      "loss": 4.6664,
      "step": 277
    },
    {
      "epoch": 0.009517288599794591,
      "grad_norm": 0.0,
      "learning_rate": 6.3397947548460666e-06,
      "loss": 4.6205,
      "step": 278
    },
    {
      "epoch": 0.00955152345087299,
      "grad_norm": 0.0,
      "learning_rate": 6.36259977194983e-06,
      "loss": 4.6306,
      "step": 279
    },
    {
      "epoch": 0.009585758301951387,
      "grad_norm": 0.0,
      "learning_rate": 6.385404789053592e-06,
      "loss": 4.5969,
      "step": 280
    },
    {
      "epoch": 0.009619993153029784,
      "grad_norm": 0.0,
      "learning_rate": 6.408209806157355e-06,
      "loss": 4.5955,
      "step": 281
    },
    {
      "epoch": 0.009654228004108182,
      "grad_norm": 0.0,
      "learning_rate": 6.431014823261118e-06,
      "loss": 4.5701,
      "step": 282
    },
    {
      "epoch": 0.00968846285518658,
      "grad_norm": 0.0,
      "learning_rate": 6.453819840364881e-06,
      "loss": 4.5032,
      "step": 283
    },
    {
      "epoch": 0.009722697706264978,
      "grad_norm": 0.0,
      "learning_rate": 6.4766248574686445e-06,
      "loss": 4.57,
      "step": 284
    },
    {
      "epoch": 0.009756932557343376,
      "grad_norm": 0.0,
      "learning_rate": 6.499429874572406e-06,
      "loss": 4.5605,
      "step": 285
    },
    {
      "epoch": 0.009791167408421774,
      "grad_norm": 0.0,
      "learning_rate": 6.522234891676169e-06,
      "loss": 4.4918,
      "step": 286
    },
    {
      "epoch": 0.009825402259500172,
      "grad_norm": 0.0,
      "learning_rate": 6.545039908779932e-06,
      "loss": 4.5186,
      "step": 287
    },
    {
      "epoch": 0.00985963711057857,
      "grad_norm": 0.0,
      "learning_rate": 6.567844925883695e-06,
      "loss": 4.4784,
      "step": 288
    },
    {
      "epoch": 0.009893871961656966,
      "grad_norm": 0.0,
      "learning_rate": 6.5906499429874585e-06,
      "loss": 4.5121,
      "step": 289
    },
    {
      "epoch": 0.009928106812735364,
      "grad_norm": 0.0,
      "learning_rate": 6.61345496009122e-06,
      "loss": 4.4438,
      "step": 290
    },
    {
      "epoch": 0.009962341663813762,
      "grad_norm": 0.0,
      "learning_rate": 6.636259977194983e-06,
      "loss": 4.4407,
      "step": 291
    },
    {
      "epoch": 0.00999657651489216,
      "grad_norm": 0.0,
      "learning_rate": 6.659064994298746e-06,
      "loss": 4.4552,
      "step": 292
    },
    {
      "epoch": 0.010030811365970558,
      "grad_norm": 0.0,
      "learning_rate": 6.681870011402509e-06,
      "loss": 4.4234,
      "step": 293
    },
    {
      "epoch": 0.010065046217048956,
      "grad_norm": 0.0,
      "learning_rate": 6.7046750285062724e-06,
      "loss": 4.4445,
      "step": 294
    },
    {
      "epoch": 0.010099281068127354,
      "grad_norm": 0.0,
      "learning_rate": 6.727480045610034e-06,
      "loss": 4.4316,
      "step": 295
    },
    {
      "epoch": 0.01013351591920575,
      "grad_norm": 0.0,
      "learning_rate": 6.750285062713797e-06,
      "loss": 4.3862,
      "step": 296
    },
    {
      "epoch": 0.010167750770284149,
      "grad_norm": 0.0,
      "learning_rate": 6.77309007981756e-06,
      "loss": 4.3931,
      "step": 297
    },
    {
      "epoch": 0.010201985621362547,
      "grad_norm": 0.0,
      "learning_rate": 6.795895096921323e-06,
      "loss": 4.3684,
      "step": 298
    },
    {
      "epoch": 0.010236220472440945,
      "grad_norm": 0.0,
      "learning_rate": 6.818700114025086e-06,
      "loss": 4.357,
      "step": 299
    },
    {
      "epoch": 0.010270455323519343,
      "grad_norm": 0.0,
      "learning_rate": 6.8415051311288495e-06,
      "loss": 4.3549,
      "step": 300
    },
    {
      "epoch": 0.01030469017459774,
      "grad_norm": 0.0,
      "learning_rate": 6.864310148232611e-06,
      "loss": 4.3478,
      "step": 301
    },
    {
      "epoch": 0.010338925025676139,
      "grad_norm": 0.0,
      "learning_rate": 6.887115165336374e-06,
      "loss": 4.3135,
      "step": 302
    },
    {
      "epoch": 0.010373159876754537,
      "grad_norm": 0.0,
      "learning_rate": 6.909920182440137e-06,
      "loss": 4.3606,
      "step": 303
    },
    {
      "epoch": 0.010407394727832933,
      "grad_norm": 0.0,
      "learning_rate": 6.9327251995439e-06,
      "loss": 4.3499,
      "step": 304
    },
    {
      "epoch": 0.010441629578911331,
      "grad_norm": 0.0,
      "learning_rate": 6.9555302166476635e-06,
      "loss": 4.325,
      "step": 305
    },
    {
      "epoch": 0.01047586442998973,
      "grad_norm": 0.0,
      "learning_rate": 6.978335233751426e-06,
      "loss": 4.2953,
      "step": 306
    },
    {
      "epoch": 0.010510099281068127,
      "grad_norm": 0.0,
      "learning_rate": 7.001140250855188e-06,
      "loss": 4.2868,
      "step": 307
    },
    {
      "epoch": 0.010544334132146525,
      "grad_norm": 0.0,
      "learning_rate": 7.023945267958951e-06,
      "loss": 4.2655,
      "step": 308
    },
    {
      "epoch": 0.010578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 7.046750285062714e-06,
      "loss": 4.2651,
      "step": 309
    },
    {
      "epoch": 0.010612803834303321,
      "grad_norm": 0.0,
      "learning_rate": 7.0695553021664774e-06,
      "loss": 4.282,
      "step": 310
    },
    {
      "epoch": 0.01064703868538172,
      "grad_norm": 0.0,
      "learning_rate": 7.09236031927024e-06,
      "loss": 4.2398,
      "step": 311
    },
    {
      "epoch": 0.010681273536460116,
      "grad_norm": 0.0,
      "learning_rate": 7.115165336374003e-06,
      "loss": 4.2964,
      "step": 312
    },
    {
      "epoch": 0.010715508387538514,
      "grad_norm": 0.0,
      "learning_rate": 7.137970353477766e-06,
      "loss": 4.1958,
      "step": 313
    },
    {
      "epoch": 0.010749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 7.160775370581528e-06,
      "loss": 4.2772,
      "step": 314
    },
    {
      "epoch": 0.01078397808969531,
      "grad_norm": 0.0,
      "learning_rate": 7.183580387685291e-06,
      "loss": 4.2175,
      "step": 315
    },
    {
      "epoch": 0.010818212940773708,
      "grad_norm": 0.0,
      "learning_rate": 7.2063854047890545e-06,
      "loss": 4.2337,
      "step": 316
    },
    {
      "epoch": 0.010852447791852106,
      "grad_norm": 0.0,
      "learning_rate": 7.229190421892817e-06,
      "loss": 4.276,
      "step": 317
    },
    {
      "epoch": 0.010886682642930504,
      "grad_norm": 0.0,
      "learning_rate": 7.25199543899658e-06,
      "loss": 4.2443,
      "step": 318
    },
    {
      "epoch": 0.0109209174940089,
      "grad_norm": 0.0,
      "learning_rate": 7.274800456100343e-06,
      "loss": 4.2562,
      "step": 319
    },
    {
      "epoch": 0.010955152345087298,
      "grad_norm": 0.0,
      "learning_rate": 7.297605473204105e-06,
      "loss": 4.2347,
      "step": 320
    },
    {
      "epoch": 0.010989387196165696,
      "grad_norm": 0.0,
      "learning_rate": 7.3204104903078685e-06,
      "loss": 4.2429,
      "step": 321
    },
    {
      "epoch": 0.011023622047244094,
      "grad_norm": 0.0,
      "learning_rate": 7.343215507411631e-06,
      "loss": 4.2239,
      "step": 322
    },
    {
      "epoch": 0.011057856898322492,
      "grad_norm": 0.0,
      "learning_rate": 7.366020524515394e-06,
      "loss": 4.193,
      "step": 323
    },
    {
      "epoch": 0.01109209174940089,
      "grad_norm": 0.0,
      "learning_rate": 7.388825541619157e-06,
      "loss": 4.2047,
      "step": 324
    },
    {
      "epoch": 0.011126326600479288,
      "grad_norm": 0.0,
      "learning_rate": 7.41163055872292e-06,
      "loss": 4.186,
      "step": 325
    },
    {
      "epoch": 0.011160561451557686,
      "grad_norm": 0.0,
      "learning_rate": 7.4344355758266825e-06,
      "loss": 4.2014,
      "step": 326
    },
    {
      "epoch": 0.011194796302636083,
      "grad_norm": 0.0,
      "learning_rate": 7.457240592930445e-06,
      "loss": 4.203,
      "step": 327
    },
    {
      "epoch": 0.01122903115371448,
      "grad_norm": 0.0,
      "learning_rate": 7.480045610034208e-06,
      "loss": 4.1546,
      "step": 328
    },
    {
      "epoch": 0.011263266004792879,
      "grad_norm": 0.0,
      "learning_rate": 7.502850627137971e-06,
      "loss": 4.1367,
      "step": 329
    },
    {
      "epoch": 0.011297500855871277,
      "grad_norm": 0.0,
      "learning_rate": 7.525655644241734e-06,
      "loss": 4.1821,
      "step": 330
    },
    {
      "epoch": 0.011331735706949675,
      "grad_norm": 0.0,
      "learning_rate": 7.548460661345497e-06,
      "loss": 4.2008,
      "step": 331
    },
    {
      "epoch": 0.011365970558028073,
      "grad_norm": 0.0,
      "learning_rate": 7.5712656784492596e-06,
      "loss": 4.1535,
      "step": 332
    },
    {
      "epoch": 0.011400205409106471,
      "grad_norm": 0.0,
      "learning_rate": 7.594070695553022e-06,
      "loss": 4.1817,
      "step": 333
    },
    {
      "epoch": 0.011434440260184869,
      "grad_norm": 0.0,
      "learning_rate": 7.616875712656785e-06,
      "loss": 4.1776,
      "step": 334
    },
    {
      "epoch": 0.011468675111263265,
      "grad_norm": 0.0,
      "learning_rate": 7.639680729760547e-06,
      "loss": 4.1358,
      "step": 335
    },
    {
      "epoch": 0.011502909962341663,
      "grad_norm": 0.0,
      "learning_rate": 7.66248574686431e-06,
      "loss": 4.1792,
      "step": 336
    },
    {
      "epoch": 0.011537144813420061,
      "grad_norm": 0.0,
      "learning_rate": 7.685290763968074e-06,
      "loss": 4.1465,
      "step": 337
    },
    {
      "epoch": 0.01157137966449846,
      "grad_norm": 0.0,
      "learning_rate": 7.708095781071837e-06,
      "loss": 4.1515,
      "step": 338
    },
    {
      "epoch": 0.011605614515576857,
      "grad_norm": 0.0,
      "learning_rate": 7.7309007981756e-06,
      "loss": 4.1347,
      "step": 339
    },
    {
      "epoch": 0.011639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 7.753705815279363e-06,
      "loss": 4.1539,
      "step": 340
    },
    {
      "epoch": 0.011674084217733654,
      "grad_norm": 0.0,
      "learning_rate": 7.776510832383124e-06,
      "loss": 4.1132,
      "step": 341
    },
    {
      "epoch": 0.01170831906881205,
      "grad_norm": 0.0,
      "learning_rate": 7.799315849486887e-06,
      "loss": 4.1374,
      "step": 342
    },
    {
      "epoch": 0.011742553919890448,
      "grad_norm": 0.0,
      "learning_rate": 7.82212086659065e-06,
      "loss": 4.1123,
      "step": 343
    },
    {
      "epoch": 0.011776788770968846,
      "grad_norm": 0.0,
      "learning_rate": 7.844925883694414e-06,
      "loss": 4.095,
      "step": 344
    },
    {
      "epoch": 0.011811023622047244,
      "grad_norm": 0.0,
      "learning_rate": 7.867730900798177e-06,
      "loss": 4.0882,
      "step": 345
    },
    {
      "epoch": 0.011845258473125642,
      "grad_norm": 0.0,
      "learning_rate": 7.89053591790194e-06,
      "loss": 4.0568,
      "step": 346
    },
    {
      "epoch": 0.01187949332420404,
      "grad_norm": 0.0,
      "learning_rate": 7.913340935005701e-06,
      "loss": 4.0684,
      "step": 347
    },
    {
      "epoch": 0.011913728175282438,
      "grad_norm": 0.0,
      "learning_rate": 7.936145952109465e-06,
      "loss": 4.0956,
      "step": 348
    },
    {
      "epoch": 0.011947963026360836,
      "grad_norm": 0.0,
      "learning_rate": 7.958950969213228e-06,
      "loss": 4.1132,
      "step": 349
    },
    {
      "epoch": 0.011982197877439232,
      "grad_norm": 0.0,
      "learning_rate": 7.98175598631699e-06,
      "loss": 4.0904,
      "step": 350
    },
    {
      "epoch": 0.01201643272851763,
      "grad_norm": 0.0,
      "learning_rate": 8.004561003420754e-06,
      "loss": 4.0988,
      "step": 351
    },
    {
      "epoch": 0.012050667579596028,
      "grad_norm": 0.0,
      "learning_rate": 8.027366020524517e-06,
      "loss": 4.074,
      "step": 352
    },
    {
      "epoch": 0.012084902430674426,
      "grad_norm": 0.0,
      "learning_rate": 8.050171037628279e-06,
      "loss": 4.0516,
      "step": 353
    },
    {
      "epoch": 0.012119137281752825,
      "grad_norm": 0.0,
      "learning_rate": 8.072976054732042e-06,
      "loss": 4.0701,
      "step": 354
    },
    {
      "epoch": 0.012153372132831223,
      "grad_norm": 0.0,
      "learning_rate": 8.095781071835805e-06,
      "loss": 4.0702,
      "step": 355
    },
    {
      "epoch": 0.01218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 8.118586088939568e-06,
      "loss": 4.0486,
      "step": 356
    },
    {
      "epoch": 0.012221841834988019,
      "grad_norm": 0.0,
      "learning_rate": 8.141391106043331e-06,
      "loss": 4.0658,
      "step": 357
    },
    {
      "epoch": 0.012256076686066415,
      "grad_norm": 0.0,
      "learning_rate": 8.164196123147094e-06,
      "loss": 4.0745,
      "step": 358
    },
    {
      "epoch": 0.012290311537144813,
      "grad_norm": 0.0,
      "learning_rate": 8.187001140250856e-06,
      "loss": 4.0817,
      "step": 359
    },
    {
      "epoch": 0.012324546388223211,
      "grad_norm": 0.0,
      "learning_rate": 8.209806157354619e-06,
      "loss": 4.0866,
      "step": 360
    },
    {
      "epoch": 0.012358781239301609,
      "grad_norm": 0.0,
      "learning_rate": 8.232611174458382e-06,
      "loss": 4.0641,
      "step": 361
    },
    {
      "epoch": 0.012393016090380007,
      "grad_norm": 0.0,
      "learning_rate": 8.255416191562145e-06,
      "loss": 4.0154,
      "step": 362
    },
    {
      "epoch": 0.012427250941458405,
      "grad_norm": 0.0,
      "learning_rate": 8.278221208665908e-06,
      "loss": 4.0741,
      "step": 363
    },
    {
      "epoch": 0.012461485792536803,
      "grad_norm": 0.0,
      "learning_rate": 8.301026225769671e-06,
      "loss": 4.0312,
      "step": 364
    },
    {
      "epoch": 0.0124957206436152,
      "grad_norm": 0.0,
      "learning_rate": 8.323831242873433e-06,
      "loss": 4.063,
      "step": 365
    },
    {
      "epoch": 0.012529955494693597,
      "grad_norm": 0.0,
      "learning_rate": 8.346636259977196e-06,
      "loss": 4.048,
      "step": 366
    },
    {
      "epoch": 0.012564190345771995,
      "grad_norm": 0.0,
      "learning_rate": 8.369441277080959e-06,
      "loss": 4.0455,
      "step": 367
    },
    {
      "epoch": 0.012598425196850394,
      "grad_norm": 0.0,
      "learning_rate": 8.392246294184722e-06,
      "loss": 3.9963,
      "step": 368
    },
    {
      "epoch": 0.012632660047928792,
      "grad_norm": 0.0,
      "learning_rate": 8.415051311288485e-06,
      "loss": 4.0582,
      "step": 369
    },
    {
      "epoch": 0.01266689489900719,
      "grad_norm": 0.0,
      "learning_rate": 8.437856328392247e-06,
      "loss": 4.0301,
      "step": 370
    },
    {
      "epoch": 0.012701129750085588,
      "grad_norm": 0.0,
      "learning_rate": 8.46066134549601e-06,
      "loss": 4.0538,
      "step": 371
    },
    {
      "epoch": 0.012735364601163986,
      "grad_norm": 0.0,
      "learning_rate": 8.483466362599773e-06,
      "loss": 4.0417,
      "step": 372
    },
    {
      "epoch": 0.012769599452242382,
      "grad_norm": 0.0,
      "learning_rate": 8.506271379703536e-06,
      "loss": 4.0618,
      "step": 373
    },
    {
      "epoch": 0.01280383430332078,
      "grad_norm": 0.0,
      "learning_rate": 8.5290763968073e-06,
      "loss": 4.1071,
      "step": 374
    },
    {
      "epoch": 0.012838069154399178,
      "grad_norm": 0.0,
      "learning_rate": 8.55188141391106e-06,
      "loss": 4.1415,
      "step": 375
    },
    {
      "epoch": 0.012872304005477576,
      "grad_norm": 0.0,
      "learning_rate": 8.574686431014824e-06,
      "loss": 4.2668,
      "step": 376
    },
    {
      "epoch": 0.012906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 8.597491448118587e-06,
      "loss": 4.5912,
      "step": 377
    },
    {
      "epoch": 0.012940773707634372,
      "grad_norm": 0.0,
      "learning_rate": 8.62029646522235e-06,
      "loss": 4.6147,
      "step": 378
    },
    {
      "epoch": 0.01297500855871277,
      "grad_norm": 0.0,
      "learning_rate": 8.643101482326113e-06,
      "loss": 4.5765,
      "step": 379
    },
    {
      "epoch": 0.013009243409791168,
      "grad_norm": 0.0,
      "learning_rate": 8.665906499429876e-06,
      "loss": 4.5088,
      "step": 380
    },
    {
      "epoch": 0.013043478260869565,
      "grad_norm": 0.0,
      "learning_rate": 8.688711516533638e-06,
      "loss": 4.3695,
      "step": 381
    },
    {
      "epoch": 0.013077713111947963,
      "grad_norm": 0.0,
      "learning_rate": 8.711516533637401e-06,
      "loss": 4.4502,
      "step": 382
    },
    {
      "epoch": 0.01311194796302636,
      "grad_norm": 0.0,
      "learning_rate": 8.734321550741164e-06,
      "loss": 4.6694,
      "step": 383
    },
    {
      "epoch": 0.013146182814104759,
      "grad_norm": 0.0,
      "learning_rate": 8.757126567844927e-06,
      "loss": 4.6249,
      "step": 384
    },
    {
      "epoch": 0.013180417665183157,
      "grad_norm": 0.0,
      "learning_rate": 8.77993158494869e-06,
      "loss": 4.6671,
      "step": 385
    },
    {
      "epoch": 0.013214652516261555,
      "grad_norm": 0.0,
      "learning_rate": 8.802736602052452e-06,
      "loss": 4.7157,
      "step": 386
    },
    {
      "epoch": 0.013248887367339953,
      "grad_norm": 0.0,
      "learning_rate": 8.825541619156215e-06,
      "loss": 4.6432,
      "step": 387
    },
    {
      "epoch": 0.013283122218418349,
      "grad_norm": 0.0,
      "learning_rate": 8.848346636259978e-06,
      "loss": 4.6292,
      "step": 388
    },
    {
      "epoch": 0.013317357069496747,
      "grad_norm": 0.0,
      "learning_rate": 8.871151653363741e-06,
      "loss": 4.5846,
      "step": 389
    },
    {
      "epoch": 0.013351591920575145,
      "grad_norm": 0.0,
      "learning_rate": 8.893956670467504e-06,
      "loss": 4.5311,
      "step": 390
    },
    {
      "epoch": 0.013385826771653543,
      "grad_norm": 0.0,
      "learning_rate": 8.916761687571266e-06,
      "loss": 4.4575,
      "step": 391
    },
    {
      "epoch": 0.013420061622731941,
      "grad_norm": 0.0,
      "learning_rate": 8.939566704675029e-06,
      "loss": 4.4109,
      "step": 392
    },
    {
      "epoch": 0.01345429647381034,
      "grad_norm": 0.0,
      "learning_rate": 8.962371721778792e-06,
      "loss": 4.3093,
      "step": 393
    },
    {
      "epoch": 0.013488531324888737,
      "grad_norm": 0.0,
      "learning_rate": 8.985176738882555e-06,
      "loss": 4.2751,
      "step": 394
    },
    {
      "epoch": 0.013522766175967135,
      "grad_norm": 0.0,
      "learning_rate": 9.007981755986318e-06,
      "loss": 4.1934,
      "step": 395
    },
    {
      "epoch": 0.013557001027045532,
      "grad_norm": 0.0,
      "learning_rate": 9.03078677309008e-06,
      "loss": 4.1652,
      "step": 396
    },
    {
      "epoch": 0.01359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 9.053591790193843e-06,
      "loss": 4.1355,
      "step": 397
    },
    {
      "epoch": 0.013625470729202328,
      "grad_norm": 0.0,
      "learning_rate": 9.076396807297606e-06,
      "loss": 4.1258,
      "step": 398
    },
    {
      "epoch": 0.013659705580280726,
      "grad_norm": 0.0,
      "learning_rate": 9.099201824401369e-06,
      "loss": 4.1279,
      "step": 399
    },
    {
      "epoch": 0.013693940431359124,
      "grad_norm": 0.0,
      "learning_rate": 9.122006841505132e-06,
      "loss": 4.1242,
      "step": 400
    },
    {
      "epoch": 0.013728175282437522,
      "grad_norm": 0.0,
      "learning_rate": 9.144811858608895e-06,
      "loss": 4.0783,
      "step": 401
    },
    {
      "epoch": 0.01376241013351592,
      "grad_norm": 0.0,
      "learning_rate": 9.167616875712657e-06,
      "loss": 4.0407,
      "step": 402
    },
    {
      "epoch": 0.013796644984594318,
      "grad_norm": 0.0,
      "learning_rate": 9.19042189281642e-06,
      "loss": 4.0341,
      "step": 403
    },
    {
      "epoch": 0.013830879835672714,
      "grad_norm": 0.0,
      "learning_rate": 9.213226909920183e-06,
      "loss": 4.0559,
      "step": 404
    },
    {
      "epoch": 0.013865114686751112,
      "grad_norm": 0.0,
      "learning_rate": 9.236031927023946e-06,
      "loss": 4.0788,
      "step": 405
    },
    {
      "epoch": 0.01389934953782951,
      "grad_norm": 0.0,
      "learning_rate": 9.25883694412771e-06,
      "loss": 4.0364,
      "step": 406
    },
    {
      "epoch": 0.013933584388907908,
      "grad_norm": 0.0,
      "learning_rate": 9.28164196123147e-06,
      "loss": 4.0667,
      "step": 407
    },
    {
      "epoch": 0.013967819239986306,
      "grad_norm": 0.0,
      "learning_rate": 9.304446978335234e-06,
      "loss": 4.0569,
      "step": 408
    },
    {
      "epoch": 0.014002054091064704,
      "grad_norm": 0.0,
      "learning_rate": 9.327251995438997e-06,
      "loss": 4.0342,
      "step": 409
    },
    {
      "epoch": 0.014036288942143102,
      "grad_norm": 0.0,
      "learning_rate": 9.35005701254276e-06,
      "loss": 4.0535,
      "step": 410
    },
    {
      "epoch": 0.0140705237932215,
      "grad_norm": 0.0,
      "learning_rate": 9.372862029646523e-06,
      "loss": 4.0837,
      "step": 411
    },
    {
      "epoch": 0.014104758644299897,
      "grad_norm": 0.0,
      "learning_rate": 9.395667046750285e-06,
      "loss": 3.9875,
      "step": 412
    },
    {
      "epoch": 0.014138993495378295,
      "grad_norm": 0.0,
      "learning_rate": 9.418472063854048e-06,
      "loss": 3.9983,
      "step": 413
    },
    {
      "epoch": 0.014173228346456693,
      "grad_norm": 0.0,
      "learning_rate": 9.441277080957811e-06,
      "loss": 4.0127,
      "step": 414
    },
    {
      "epoch": 0.01420746319753509,
      "grad_norm": 0.0,
      "learning_rate": 9.464082098061574e-06,
      "loss": 3.9781,
      "step": 415
    },
    {
      "epoch": 0.014241698048613489,
      "grad_norm": 0.0,
      "learning_rate": 9.486887115165337e-06,
      "loss": 4.0032,
      "step": 416
    },
    {
      "epoch": 0.014275932899691887,
      "grad_norm": 0.0,
      "learning_rate": 9.5096921322691e-06,
      "loss": 3.9916,
      "step": 417
    },
    {
      "epoch": 0.014310167750770285,
      "grad_norm": 0.0,
      "learning_rate": 9.532497149372862e-06,
      "loss": 3.9815,
      "step": 418
    },
    {
      "epoch": 0.014344402601848681,
      "grad_norm": 0.0,
      "learning_rate": 9.555302166476625e-06,
      "loss": 4.0026,
      "step": 419
    },
    {
      "epoch": 0.01437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 9.578107183580388e-06,
      "loss": 4.0215,
      "step": 420
    },
    {
      "epoch": 0.014412872304005477,
      "grad_norm": 0.0,
      "learning_rate": 9.600912200684151e-06,
      "loss": 3.9758,
      "step": 421
    },
    {
      "epoch": 0.014447107155083875,
      "grad_norm": 0.0,
      "learning_rate": 9.623717217787914e-06,
      "loss": 3.9718,
      "step": 422
    },
    {
      "epoch": 0.014481342006162273,
      "grad_norm": 0.0,
      "learning_rate": 9.646522234891676e-06,
      "loss": 3.9921,
      "step": 423
    },
    {
      "epoch": 0.014515576857240671,
      "grad_norm": 0.0,
      "learning_rate": 9.669327251995439e-06,
      "loss": 3.9546,
      "step": 424
    },
    {
      "epoch": 0.01454981170831907,
      "grad_norm": 0.0,
      "learning_rate": 9.692132269099202e-06,
      "loss": 3.9439,
      "step": 425
    },
    {
      "epoch": 0.014584046559397467,
      "grad_norm": 0.0,
      "learning_rate": 9.714937286202965e-06,
      "loss": 3.9529,
      "step": 426
    },
    {
      "epoch": 0.014618281410475864,
      "grad_norm": 0.0,
      "learning_rate": 9.737742303306728e-06,
      "loss": 3.9832,
      "step": 427
    },
    {
      "epoch": 0.014652516261554262,
      "grad_norm": 0.0,
      "learning_rate": 9.760547320410491e-06,
      "loss": 3.978,
      "step": 428
    },
    {
      "epoch": 0.01468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 9.783352337514254e-06,
      "loss": 3.9346,
      "step": 429
    },
    {
      "epoch": 0.014720985963711058,
      "grad_norm": 0.0,
      "learning_rate": 9.806157354618016e-06,
      "loss": 3.9659,
      "step": 430
    },
    {
      "epoch": 0.014755220814789456,
      "grad_norm": 0.0,
      "learning_rate": 9.828962371721779e-06,
      "loss": 3.9543,
      "step": 431
    },
    {
      "epoch": 0.014789455665867854,
      "grad_norm": 0.0,
      "learning_rate": 9.851767388825542e-06,
      "loss": 3.9102,
      "step": 432
    },
    {
      "epoch": 0.014823690516946252,
      "grad_norm": 0.0,
      "learning_rate": 9.874572405929305e-06,
      "loss": 3.9335,
      "step": 433
    },
    {
      "epoch": 0.01485792536802465,
      "grad_norm": 0.0,
      "learning_rate": 9.897377423033068e-06,
      "loss": 3.9318,
      "step": 434
    },
    {
      "epoch": 0.014892160219103046,
      "grad_norm": 0.0,
      "learning_rate": 9.920182440136832e-06,
      "loss": 3.9815,
      "step": 435
    },
    {
      "epoch": 0.014926395070181444,
      "grad_norm": 0.0,
      "learning_rate": 9.942987457240593e-06,
      "loss": 3.9701,
      "step": 436
    },
    {
      "epoch": 0.014960629921259842,
      "grad_norm": 0.0,
      "learning_rate": 9.965792474344356e-06,
      "loss": 3.9181,
      "step": 437
    },
    {
      "epoch": 0.01499486477233824,
      "grad_norm": 0.0,
      "learning_rate": 9.98859749144812e-06,
      "loss": 3.9713,
      "step": 438
    },
    {
      "epoch": 0.015029099623416638,
      "grad_norm": 0.0,
      "learning_rate": 1.0011402508551882e-05,
      "loss": 3.9664,
      "step": 439
    },
    {
      "epoch": 0.015063334474495036,
      "grad_norm": 0.0,
      "learning_rate": 1.0034207525655646e-05,
      "loss": 3.923,
      "step": 440
    },
    {
      "epoch": 0.015097569325573434,
      "grad_norm": 0.0,
      "learning_rate": 1.0057012542759409e-05,
      "loss": 3.9602,
      "step": 441
    },
    {
      "epoch": 0.01513180417665183,
      "grad_norm": 0.0,
      "learning_rate": 1.007981755986317e-05,
      "loss": 3.977,
      "step": 442
    },
    {
      "epoch": 0.015166039027730229,
      "grad_norm": 0.0,
      "learning_rate": 1.0102622576966933e-05,
      "loss": 3.9907,
      "step": 443
    },
    {
      "epoch": 0.015200273878808627,
      "grad_norm": 0.0,
      "learning_rate": 1.0125427594070696e-05,
      "loss": 3.9691,
      "step": 444
    },
    {
      "epoch": 0.015234508729887025,
      "grad_norm": 0.0,
      "learning_rate": 1.014823261117446e-05,
      "loss": 3.9542,
      "step": 445
    },
    {
      "epoch": 0.015268743580965423,
      "grad_norm": 0.0,
      "learning_rate": 1.0171037628278223e-05,
      "loss": 3.9736,
      "step": 446
    },
    {
      "epoch": 0.015302978432043821,
      "grad_norm": 0.0,
      "learning_rate": 1.0193842645381984e-05,
      "loss": 3.9395,
      "step": 447
    },
    {
      "epoch": 0.015337213283122219,
      "grad_norm": 0.0,
      "learning_rate": 1.0216647662485747e-05,
      "loss": 3.964,
      "step": 448
    },
    {
      "epoch": 0.015371448134200617,
      "grad_norm": 0.0,
      "learning_rate": 1.023945267958951e-05,
      "loss": 3.9582,
      "step": 449
    },
    {
      "epoch": 0.015405682985279013,
      "grad_norm": 0.0,
      "learning_rate": 1.0262257696693273e-05,
      "loss": 3.9171,
      "step": 450
    },
    {
      "epoch": 0.015439917836357411,
      "grad_norm": 0.0,
      "learning_rate": 1.0285062713797037e-05,
      "loss": 3.9559,
      "step": 451
    },
    {
      "epoch": 0.01547415268743581,
      "grad_norm": 0.0,
      "learning_rate": 1.0307867730900798e-05,
      "loss": 3.8935,
      "step": 452
    },
    {
      "epoch": 0.015508387538514207,
      "grad_norm": 0.0,
      "learning_rate": 1.0330672748004563e-05,
      "loss": 3.9525,
      "step": 453
    },
    {
      "epoch": 0.015542622389592605,
      "grad_norm": 0.0,
      "learning_rate": 1.0353477765108324e-05,
      "loss": 3.935,
      "step": 454
    },
    {
      "epoch": 0.015576857240671003,
      "grad_norm": 0.0,
      "learning_rate": 1.0376282782212087e-05,
      "loss": 3.9479,
      "step": 455
    },
    {
      "epoch": 0.015611092091749402,
      "grad_norm": 0.0,
      "learning_rate": 1.039908779931585e-05,
      "loss": 3.9225,
      "step": 456
    },
    {
      "epoch": 0.0156453269428278,
      "grad_norm": 0.0,
      "learning_rate": 1.0421892816419614e-05,
      "loss": 3.9284,
      "step": 457
    },
    {
      "epoch": 0.015679561793906196,
      "grad_norm": 0.0,
      "learning_rate": 1.0444697833523377e-05,
      "loss": 3.9704,
      "step": 458
    },
    {
      "epoch": 0.015713796644984596,
      "grad_norm": 0.0,
      "learning_rate": 1.0467502850627138e-05,
      "loss": 3.9589,
      "step": 459
    },
    {
      "epoch": 0.015748031496062992,
      "grad_norm": 0.0,
      "learning_rate": 1.0490307867730903e-05,
      "loss": 3.9516,
      "step": 460
    },
    {
      "epoch": 0.015782266347141388,
      "grad_norm": 0.0,
      "learning_rate": 1.0513112884834664e-05,
      "loss": 3.9265,
      "step": 461
    },
    {
      "epoch": 0.015816501198219788,
      "grad_norm": 0.0,
      "learning_rate": 1.0535917901938428e-05,
      "loss": 3.9059,
      "step": 462
    },
    {
      "epoch": 0.015850736049298184,
      "grad_norm": 0.0,
      "learning_rate": 1.055872291904219e-05,
      "loss": 3.9009,
      "step": 463
    },
    {
      "epoch": 0.015884970900376584,
      "grad_norm": 0.0,
      "learning_rate": 1.0581527936145952e-05,
      "loss": 3.9157,
      "step": 464
    },
    {
      "epoch": 0.01591920575145498,
      "grad_norm": 0.0,
      "learning_rate": 1.0604332953249717e-05,
      "loss": 3.9323,
      "step": 465
    },
    {
      "epoch": 0.01595344060253338,
      "grad_norm": 0.0,
      "learning_rate": 1.0627137970353478e-05,
      "loss": 3.9114,
      "step": 466
    },
    {
      "epoch": 0.015987675453611776,
      "grad_norm": 0.0,
      "learning_rate": 1.0649942987457242e-05,
      "loss": 3.9219,
      "step": 467
    },
    {
      "epoch": 0.016021910304690176,
      "grad_norm": 0.0,
      "learning_rate": 1.0672748004561005e-05,
      "loss": 3.9244,
      "step": 468
    },
    {
      "epoch": 0.016056145155768572,
      "grad_norm": 0.0,
      "learning_rate": 1.0695553021664766e-05,
      "loss": 3.9148,
      "step": 469
    },
    {
      "epoch": 0.01609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 1.0718358038768531e-05,
      "loss": 3.9104,
      "step": 470
    },
    {
      "epoch": 0.01612461485792537,
      "grad_norm": 0.0,
      "learning_rate": 1.0741163055872292e-05,
      "loss": 3.9307,
      "step": 471
    },
    {
      "epoch": 0.016158849709003765,
      "grad_norm": 0.0,
      "learning_rate": 1.0763968072976057e-05,
      "loss": 3.8739,
      "step": 472
    },
    {
      "epoch": 0.016193084560082165,
      "grad_norm": 0.0,
      "learning_rate": 1.0786773090079819e-05,
      "loss": 3.8998,
      "step": 473
    },
    {
      "epoch": 0.01622731941116056,
      "grad_norm": 0.0,
      "learning_rate": 1.080957810718358e-05,
      "loss": 3.9097,
      "step": 474
    },
    {
      "epoch": 0.01626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 1.0832383124287345e-05,
      "loss": 3.9231,
      "step": 475
    },
    {
      "epoch": 0.016295789113317357,
      "grad_norm": 0.0,
      "learning_rate": 1.0855188141391106e-05,
      "loss": 3.8845,
      "step": 476
    },
    {
      "epoch": 0.016330023964395753,
      "grad_norm": 0.0,
      "learning_rate": 1.0877993158494871e-05,
      "loss": 3.8932,
      "step": 477
    },
    {
      "epoch": 0.016364258815474153,
      "grad_norm": 0.0,
      "learning_rate": 1.0900798175598633e-05,
      "loss": 3.9084,
      "step": 478
    },
    {
      "epoch": 0.01639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 1.0923603192702394e-05,
      "loss": 3.9055,
      "step": 479
    },
    {
      "epoch": 0.01643272851763095,
      "grad_norm": 0.0,
      "learning_rate": 1.0946408209806159e-05,
      "loss": 3.9019,
      "step": 480
    },
    {
      "epoch": 0.016466963368709345,
      "grad_norm": 0.0,
      "learning_rate": 1.096921322690992e-05,
      "loss": 3.8932,
      "step": 481
    },
    {
      "epoch": 0.016501198219787745,
      "grad_norm": 0.0,
      "learning_rate": 1.0992018244013685e-05,
      "loss": 3.9298,
      "step": 482
    },
    {
      "epoch": 0.01653543307086614,
      "grad_norm": 0.0,
      "learning_rate": 1.1014823261117447e-05,
      "loss": 3.8759,
      "step": 483
    },
    {
      "epoch": 0.01656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 1.1037628278221208e-05,
      "loss": 3.8566,
      "step": 484
    },
    {
      "epoch": 0.016603902773022938,
      "grad_norm": 0.0,
      "learning_rate": 1.1060433295324973e-05,
      "loss": 3.8621,
      "step": 485
    },
    {
      "epoch": 0.016638137624101334,
      "grad_norm": 0.0,
      "learning_rate": 1.1083238312428734e-05,
      "loss": 3.8474,
      "step": 486
    },
    {
      "epoch": 0.016672372475179734,
      "grad_norm": 0.0,
      "learning_rate": 1.1106043329532499e-05,
      "loss": 3.9009,
      "step": 487
    },
    {
      "epoch": 0.01670660732625813,
      "grad_norm": 0.0,
      "learning_rate": 1.112884834663626e-05,
      "loss": 3.858,
      "step": 488
    },
    {
      "epoch": 0.01674084217733653,
      "grad_norm": 0.0,
      "learning_rate": 1.1151653363740022e-05,
      "loss": 3.9006,
      "step": 489
    },
    {
      "epoch": 0.016775077028414926,
      "grad_norm": 0.0,
      "learning_rate": 1.1174458380843787e-05,
      "loss": 3.885,
      "step": 490
    },
    {
      "epoch": 0.016809311879493326,
      "grad_norm": 0.0,
      "learning_rate": 1.1197263397947548e-05,
      "loss": 3.8843,
      "step": 491
    },
    {
      "epoch": 0.016843546730571722,
      "grad_norm": 0.0,
      "learning_rate": 1.1220068415051313e-05,
      "loss": 3.8547,
      "step": 492
    },
    {
      "epoch": 0.01687778158165012,
      "grad_norm": 0.0,
      "learning_rate": 1.1242873432155075e-05,
      "loss": 3.8684,
      "step": 493
    },
    {
      "epoch": 0.016912016432728518,
      "grad_norm": 0.0,
      "learning_rate": 1.126567844925884e-05,
      "loss": 3.8156,
      "step": 494
    },
    {
      "epoch": 0.016946251283806914,
      "grad_norm": 0.0,
      "learning_rate": 1.12884834663626e-05,
      "loss": 3.8372,
      "step": 495
    },
    {
      "epoch": 0.016980486134885314,
      "grad_norm": 0.0,
      "learning_rate": 1.1311288483466362e-05,
      "loss": 3.842,
      "step": 496
    },
    {
      "epoch": 0.01701472098596371,
      "grad_norm": 0.0,
      "learning_rate": 1.1334093500570127e-05,
      "loss": 3.8343,
      "step": 497
    },
    {
      "epoch": 0.01704895583704211,
      "grad_norm": 0.0,
      "learning_rate": 1.1356898517673888e-05,
      "loss": 3.9196,
      "step": 498
    },
    {
      "epoch": 0.017083190688120507,
      "grad_norm": 0.0,
      "learning_rate": 1.1379703534777653e-05,
      "loss": 3.8452,
      "step": 499
    },
    {
      "epoch": 0.017117425539198903,
      "grad_norm": 0.0,
      "learning_rate": 1.1402508551881415e-05,
      "loss": 3.8947,
      "step": 500
    },
    {
      "epoch": 0.017151660390277303,
      "grad_norm": 0.0,
      "learning_rate": 1.1425313568985176e-05,
      "loss": 3.8937,
      "step": 501
    },
    {
      "epoch": 0.0171858952413557,
      "grad_norm": 0.0,
      "learning_rate": 1.1448118586088941e-05,
      "loss": 3.8576,
      "step": 502
    },
    {
      "epoch": 0.0172201300924341,
      "grad_norm": 0.0,
      "learning_rate": 1.1470923603192702e-05,
      "loss": 3.8794,
      "step": 503
    },
    {
      "epoch": 0.017254364943512495,
      "grad_norm": 0.0,
      "learning_rate": 1.1493728620296467e-05,
      "loss": 3.8263,
      "step": 504
    },
    {
      "epoch": 0.017288599794590895,
      "grad_norm": 0.0,
      "learning_rate": 1.1516533637400229e-05,
      "loss": 3.8758,
      "step": 505
    },
    {
      "epoch": 0.01732283464566929,
      "grad_norm": 0.0,
      "learning_rate": 1.153933865450399e-05,
      "loss": 3.8626,
      "step": 506
    },
    {
      "epoch": 0.01735706949674769,
      "grad_norm": 0.0,
      "learning_rate": 1.1562143671607755e-05,
      "loss": 3.8533,
      "step": 507
    },
    {
      "epoch": 0.017391304347826087,
      "grad_norm": 0.0,
      "learning_rate": 1.1584948688711516e-05,
      "loss": 3.8212,
      "step": 508
    },
    {
      "epoch": 0.017425539198904483,
      "grad_norm": 0.0,
      "learning_rate": 1.1607753705815281e-05,
      "loss": 3.8831,
      "step": 509
    },
    {
      "epoch": 0.017459774049982883,
      "grad_norm": 0.0,
      "learning_rate": 1.1630558722919043e-05,
      "loss": 3.8779,
      "step": 510
    },
    {
      "epoch": 0.01749400890106128,
      "grad_norm": 0.0,
      "learning_rate": 1.1653363740022806e-05,
      "loss": 3.8825,
      "step": 511
    },
    {
      "epoch": 0.01752824375213968,
      "grad_norm": 0.0,
      "learning_rate": 1.1676168757126569e-05,
      "loss": 3.8357,
      "step": 512
    },
    {
      "epoch": 0.017562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 1.169897377423033e-05,
      "loss": 3.8648,
      "step": 513
    },
    {
      "epoch": 0.017596713454296475,
      "grad_norm": 0.0,
      "learning_rate": 1.1721778791334095e-05,
      "loss": 3.8418,
      "step": 514
    },
    {
      "epoch": 0.01763094830537487,
      "grad_norm": 0.0,
      "learning_rate": 1.1744583808437857e-05,
      "loss": 3.8194,
      "step": 515
    },
    {
      "epoch": 0.017665183156453268,
      "grad_norm": 0.0,
      "learning_rate": 1.176738882554162e-05,
      "loss": 3.8263,
      "step": 516
    },
    {
      "epoch": 0.017699418007531668,
      "grad_norm": 0.0,
      "learning_rate": 1.1790193842645383e-05,
      "loss": 3.8852,
      "step": 517
    },
    {
      "epoch": 0.017733652858610064,
      "grad_norm": 0.0,
      "learning_rate": 1.1812998859749146e-05,
      "loss": 3.8478,
      "step": 518
    },
    {
      "epoch": 0.017767887709688464,
      "grad_norm": 0.0,
      "learning_rate": 1.1835803876852909e-05,
      "loss": 3.8341,
      "step": 519
    },
    {
      "epoch": 0.01780212256076686,
      "grad_norm": 0.0,
      "learning_rate": 1.185860889395667e-05,
      "loss": 3.8438,
      "step": 520
    },
    {
      "epoch": 0.01783635741184526,
      "grad_norm": 0.0,
      "learning_rate": 1.1881413911060434e-05,
      "loss": 3.847,
      "step": 521
    },
    {
      "epoch": 0.017870592262923656,
      "grad_norm": 0.0,
      "learning_rate": 1.1904218928164197e-05,
      "loss": 3.8477,
      "step": 522
    },
    {
      "epoch": 0.017904827114002052,
      "grad_norm": 0.0,
      "learning_rate": 1.192702394526796e-05,
      "loss": 3.838,
      "step": 523
    },
    {
      "epoch": 0.017939061965080452,
      "grad_norm": 0.0,
      "learning_rate": 1.1949828962371723e-05,
      "loss": 3.8266,
      "step": 524
    },
    {
      "epoch": 0.01797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 1.1972633979475485e-05,
      "loss": 3.8296,
      "step": 525
    },
    {
      "epoch": 0.01800753166723725,
      "grad_norm": 0.0,
      "learning_rate": 1.199543899657925e-05,
      "loss": 3.8192,
      "step": 526
    },
    {
      "epoch": 0.018041766518315645,
      "grad_norm": 0.0,
      "learning_rate": 1.201824401368301e-05,
      "loss": 3.8329,
      "step": 527
    },
    {
      "epoch": 0.018076001369394044,
      "grad_norm": 0.0,
      "learning_rate": 1.2041049030786774e-05,
      "loss": 3.831,
      "step": 528
    },
    {
      "epoch": 0.01811023622047244,
      "grad_norm": 0.0,
      "learning_rate": 1.2063854047890537e-05,
      "loss": 3.8188,
      "step": 529
    },
    {
      "epoch": 0.01814447107155084,
      "grad_norm": 0.0,
      "learning_rate": 1.20866590649943e-05,
      "loss": 3.7851,
      "step": 530
    },
    {
      "epoch": 0.018178705922629237,
      "grad_norm": 0.0,
      "learning_rate": 1.2109464082098063e-05,
      "loss": 3.8194,
      "step": 531
    },
    {
      "epoch": 0.018212940773707633,
      "grad_norm": 0.0,
      "learning_rate": 1.2132269099201825e-05,
      "loss": 3.8018,
      "step": 532
    },
    {
      "epoch": 0.018247175624786033,
      "grad_norm": 0.0,
      "learning_rate": 1.2155074116305588e-05,
      "loss": 3.8342,
      "step": 533
    },
    {
      "epoch": 0.01828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 1.2177879133409351e-05,
      "loss": 3.8173,
      "step": 534
    },
    {
      "epoch": 0.01831564532694283,
      "grad_norm": 0.0,
      "learning_rate": 1.2200684150513114e-05,
      "loss": 3.7828,
      "step": 535
    },
    {
      "epoch": 0.018349880178021225,
      "grad_norm": 0.0,
      "learning_rate": 1.2223489167616877e-05,
      "loss": 3.8038,
      "step": 536
    },
    {
      "epoch": 0.018384115029099625,
      "grad_norm": 0.0,
      "learning_rate": 1.2246294184720639e-05,
      "loss": 3.827,
      "step": 537
    },
    {
      "epoch": 0.01841834988017802,
      "grad_norm": 0.0,
      "learning_rate": 1.2269099201824402e-05,
      "loss": 3.8343,
      "step": 538
    },
    {
      "epoch": 0.018452584731256418,
      "grad_norm": 0.0,
      "learning_rate": 1.2291904218928165e-05,
      "loss": 3.8246,
      "step": 539
    },
    {
      "epoch": 0.018486819582334817,
      "grad_norm": 0.0,
      "learning_rate": 1.2314709236031928e-05,
      "loss": 3.8333,
      "step": 540
    },
    {
      "epoch": 0.018521054433413214,
      "grad_norm": 0.0,
      "learning_rate": 1.2337514253135691e-05,
      "loss": 3.8201,
      "step": 541
    },
    {
      "epoch": 0.018555289284491613,
      "grad_norm": 0.0,
      "learning_rate": 1.2360319270239454e-05,
      "loss": 3.807,
      "step": 542
    },
    {
      "epoch": 0.01858952413557001,
      "grad_norm": 0.0,
      "learning_rate": 1.2383124287343216e-05,
      "loss": 3.7997,
      "step": 543
    },
    {
      "epoch": 0.01862375898664841,
      "grad_norm": 0.0,
      "learning_rate": 1.2405929304446979e-05,
      "loss": 3.8073,
      "step": 544
    },
    {
      "epoch": 0.018657993837726806,
      "grad_norm": 0.0,
      "learning_rate": 1.2428734321550742e-05,
      "loss": 3.7992,
      "step": 545
    },
    {
      "epoch": 0.018692228688805202,
      "grad_norm": 0.0,
      "learning_rate": 1.2451539338654505e-05,
      "loss": 3.825,
      "step": 546
    },
    {
      "epoch": 0.018726463539883602,
      "grad_norm": 0.0,
      "learning_rate": 1.2474344355758268e-05,
      "loss": 3.7262,
      "step": 547
    },
    {
      "epoch": 0.018760698390961998,
      "grad_norm": 0.0,
      "learning_rate": 1.249714937286203e-05,
      "loss": 3.7951,
      "step": 548
    },
    {
      "epoch": 0.018794933242040398,
      "grad_norm": 0.0,
      "learning_rate": 1.2519954389965795e-05,
      "loss": 3.7694,
      "step": 549
    },
    {
      "epoch": 0.018829168093118794,
      "grad_norm": 0.0,
      "learning_rate": 1.2542759407069556e-05,
      "loss": 3.7568,
      "step": 550
    },
    {
      "epoch": 0.018863402944197194,
      "grad_norm": 0.0,
      "learning_rate": 1.256556442417332e-05,
      "loss": 3.8233,
      "step": 551
    },
    {
      "epoch": 0.01889763779527559,
      "grad_norm": 0.0,
      "learning_rate": 1.2588369441277082e-05,
      "loss": 3.8096,
      "step": 552
    },
    {
      "epoch": 0.01893187264635399,
      "grad_norm": 0.0,
      "learning_rate": 1.2611174458380844e-05,
      "loss": 3.787,
      "step": 553
    },
    {
      "epoch": 0.018966107497432386,
      "grad_norm": 0.0,
      "learning_rate": 1.2633979475484609e-05,
      "loss": 3.7962,
      "step": 554
    },
    {
      "epoch": 0.019000342348510783,
      "grad_norm": 0.0,
      "learning_rate": 1.265678449258837e-05,
      "loss": 3.8031,
      "step": 555
    },
    {
      "epoch": 0.019034577199589182,
      "grad_norm": 0.0,
      "learning_rate": 1.2679589509692133e-05,
      "loss": 3.7593,
      "step": 556
    },
    {
      "epoch": 0.01906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 1.2702394526795896e-05,
      "loss": 3.7473,
      "step": 557
    },
    {
      "epoch": 0.01910304690174598,
      "grad_norm": 0.0,
      "learning_rate": 1.272519954389966e-05,
      "loss": 3.7848,
      "step": 558
    },
    {
      "epoch": 0.019137281752824375,
      "grad_norm": 0.0,
      "learning_rate": 1.2748004561003423e-05,
      "loss": 3.7883,
      "step": 559
    },
    {
      "epoch": 0.019171516603902775,
      "grad_norm": 0.0,
      "learning_rate": 1.2770809578107184e-05,
      "loss": 3.7693,
      "step": 560
    },
    {
      "epoch": 0.01920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 1.2793614595210949e-05,
      "loss": 3.7661,
      "step": 561
    },
    {
      "epoch": 0.019239986306059567,
      "grad_norm": 0.0,
      "learning_rate": 1.281641961231471e-05,
      "loss": 3.7751,
      "step": 562
    },
    {
      "epoch": 0.019274221157137967,
      "grad_norm": 0.0,
      "learning_rate": 1.2839224629418473e-05,
      "loss": 3.7452,
      "step": 563
    },
    {
      "epoch": 0.019308456008216363,
      "grad_norm": 0.0,
      "learning_rate": 1.2862029646522236e-05,
      "loss": 3.7427,
      "step": 564
    },
    {
      "epoch": 0.019342690859294763,
      "grad_norm": 0.0,
      "learning_rate": 1.2884834663625998e-05,
      "loss": 3.7595,
      "step": 565
    },
    {
      "epoch": 0.01937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 1.2907639680729763e-05,
      "loss": 3.7283,
      "step": 566
    },
    {
      "epoch": 0.01941116056145156,
      "grad_norm": 0.0,
      "learning_rate": 1.2930444697833524e-05,
      "loss": 3.7334,
      "step": 567
    },
    {
      "epoch": 0.019445395412529955,
      "grad_norm": 0.0,
      "learning_rate": 1.2953249714937289e-05,
      "loss": 3.7685,
      "step": 568
    },
    {
      "epoch": 0.01947963026360835,
      "grad_norm": 0.0,
      "learning_rate": 1.297605473204105e-05,
      "loss": 3.753,
      "step": 569
    },
    {
      "epoch": 0.01951386511468675,
      "grad_norm": 0.0,
      "learning_rate": 1.2998859749144812e-05,
      "loss": 3.7491,
      "step": 570
    },
    {
      "epoch": 0.019548099965765148,
      "grad_norm": 0.0,
      "learning_rate": 1.3021664766248577e-05,
      "loss": 3.7849,
      "step": 571
    },
    {
      "epoch": 0.019582334816843548,
      "grad_norm": 0.0,
      "learning_rate": 1.3044469783352338e-05,
      "loss": 3.723,
      "step": 572
    },
    {
      "epoch": 0.019616569667921944,
      "grad_norm": 0.0,
      "learning_rate": 1.3067274800456103e-05,
      "loss": 3.7504,
      "step": 573
    },
    {
      "epoch": 0.019650804519000344,
      "grad_norm": 0.0,
      "learning_rate": 1.3090079817559864e-05,
      "loss": 3.7763,
      "step": 574
    },
    {
      "epoch": 0.01968503937007874,
      "grad_norm": 0.0,
      "learning_rate": 1.3112884834663626e-05,
      "loss": 3.7422,
      "step": 575
    },
    {
      "epoch": 0.01971927422115714,
      "grad_norm": 0.0,
      "learning_rate": 1.313568985176739e-05,
      "loss": 3.7159,
      "step": 576
    },
    {
      "epoch": 0.019753509072235536,
      "grad_norm": 0.0,
      "learning_rate": 1.3158494868871152e-05,
      "loss": 3.7395,
      "step": 577
    },
    {
      "epoch": 0.019787743923313932,
      "grad_norm": 0.0,
      "learning_rate": 1.3181299885974917e-05,
      "loss": 3.72,
      "step": 578
    },
    {
      "epoch": 0.019821978774392332,
      "grad_norm": 0.0,
      "learning_rate": 1.3204104903078678e-05,
      "loss": 3.7295,
      "step": 579
    },
    {
      "epoch": 0.01985621362547073,
      "grad_norm": 0.0,
      "learning_rate": 1.322690992018244e-05,
      "loss": 3.712,
      "step": 580
    },
    {
      "epoch": 0.019890448476549128,
      "grad_norm": 0.0,
      "learning_rate": 1.3249714937286205e-05,
      "loss": 3.742,
      "step": 581
    },
    {
      "epoch": 0.019924683327627524,
      "grad_norm": 0.0,
      "learning_rate": 1.3272519954389966e-05,
      "loss": 3.6996,
      "step": 582
    },
    {
      "epoch": 0.019958918178705924,
      "grad_norm": 0.0,
      "learning_rate": 1.3295324971493731e-05,
      "loss": 3.7189,
      "step": 583
    },
    {
      "epoch": 0.01999315302978432,
      "grad_norm": 0.0,
      "learning_rate": 1.3318129988597492e-05,
      "loss": 3.7426,
      "step": 584
    },
    {
      "epoch": 0.020027387880862717,
      "grad_norm": 0.0,
      "learning_rate": 1.3340935005701254e-05,
      "loss": 3.7309,
      "step": 585
    },
    {
      "epoch": 0.020061622731941117,
      "grad_norm": 0.0,
      "learning_rate": 1.3363740022805019e-05,
      "loss": 3.6876,
      "step": 586
    },
    {
      "epoch": 0.020095857583019513,
      "grad_norm": 0.0,
      "learning_rate": 1.338654503990878e-05,
      "loss": 3.747,
      "step": 587
    },
    {
      "epoch": 0.020130092434097913,
      "grad_norm": 0.0,
      "learning_rate": 1.3409350057012545e-05,
      "loss": 3.7203,
      "step": 588
    },
    {
      "epoch": 0.02016432728517631,
      "grad_norm": 0.0,
      "learning_rate": 1.3432155074116306e-05,
      "loss": 3.7087,
      "step": 589
    },
    {
      "epoch": 0.02019856213625471,
      "grad_norm": 0.0,
      "learning_rate": 1.3454960091220068e-05,
      "loss": 3.707,
      "step": 590
    },
    {
      "epoch": 0.020232796987333105,
      "grad_norm": 0.0,
      "learning_rate": 1.3477765108323833e-05,
      "loss": 3.7484,
      "step": 591
    },
    {
      "epoch": 0.0202670318384115,
      "grad_norm": 0.0,
      "learning_rate": 1.3500570125427594e-05,
      "loss": 3.7168,
      "step": 592
    },
    {
      "epoch": 0.0203012666894899,
      "grad_norm": 0.0,
      "learning_rate": 1.3523375142531359e-05,
      "loss": 3.7064,
      "step": 593
    },
    {
      "epoch": 0.020335501540568297,
      "grad_norm": 0.0,
      "learning_rate": 1.354618015963512e-05,
      "loss": 3.7317,
      "step": 594
    },
    {
      "epoch": 0.020369736391646697,
      "grad_norm": 0.0,
      "learning_rate": 1.3568985176738885e-05,
      "loss": 3.7544,
      "step": 595
    },
    {
      "epoch": 0.020403971242725093,
      "grad_norm": 0.0,
      "learning_rate": 1.3591790193842647e-05,
      "loss": 3.7294,
      "step": 596
    },
    {
      "epoch": 0.020438206093803493,
      "grad_norm": 0.0,
      "learning_rate": 1.3614595210946408e-05,
      "loss": 3.7472,
      "step": 597
    },
    {
      "epoch": 0.02047244094488189,
      "grad_norm": 0.0,
      "learning_rate": 1.3637400228050173e-05,
      "loss": 3.7144,
      "step": 598
    },
    {
      "epoch": 0.02050667579596029,
      "grad_norm": 0.0,
      "learning_rate": 1.3660205245153934e-05,
      "loss": 3.752,
      "step": 599
    },
    {
      "epoch": 0.020540910647038686,
      "grad_norm": 0.0,
      "learning_rate": 1.3683010262257699e-05,
      "loss": 3.729,
      "step": 600
    },
    {
      "epoch": 0.020575145498117082,
      "grad_norm": 0.0,
      "learning_rate": 1.370581527936146e-05,
      "loss": 3.6919,
      "step": 601
    },
    {
      "epoch": 0.02060938034919548,
      "grad_norm": 0.0,
      "learning_rate": 1.3728620296465222e-05,
      "loss": 3.692,
      "step": 602
    },
    {
      "epoch": 0.020643615200273878,
      "grad_norm": 0.0,
      "learning_rate": 1.3751425313568987e-05,
      "loss": 3.7041,
      "step": 603
    },
    {
      "epoch": 0.020677850051352278,
      "grad_norm": 0.0,
      "learning_rate": 1.3774230330672748e-05,
      "loss": 3.6746,
      "step": 604
    },
    {
      "epoch": 0.020712084902430674,
      "grad_norm": 0.0,
      "learning_rate": 1.3797035347776513e-05,
      "loss": 3.6816,
      "step": 605
    },
    {
      "epoch": 0.020746319753509074,
      "grad_norm": 0.0,
      "learning_rate": 1.3819840364880274e-05,
      "loss": 3.6949,
      "step": 606
    },
    {
      "epoch": 0.02078055460458747,
      "grad_norm": 0.0,
      "learning_rate": 1.3842645381984038e-05,
      "loss": 3.6702,
      "step": 607
    },
    {
      "epoch": 0.020814789455665866,
      "grad_norm": 0.0,
      "learning_rate": 1.38654503990878e-05,
      "loss": 3.6759,
      "step": 608
    },
    {
      "epoch": 0.020849024306744266,
      "grad_norm": 0.0,
      "learning_rate": 1.3888255416191562e-05,
      "loss": 3.6639,
      "step": 609
    },
    {
      "epoch": 0.020883259157822662,
      "grad_norm": 0.0,
      "learning_rate": 1.3911060433295327e-05,
      "loss": 3.7,
      "step": 610
    },
    {
      "epoch": 0.020917494008901062,
      "grad_norm": 0.0,
      "learning_rate": 1.3933865450399088e-05,
      "loss": 3.699,
      "step": 611
    },
    {
      "epoch": 0.02095172885997946,
      "grad_norm": 0.0,
      "learning_rate": 1.3956670467502852e-05,
      "loss": 3.6897,
      "step": 612
    },
    {
      "epoch": 0.02098596371105786,
      "grad_norm": 0.0,
      "learning_rate": 1.3979475484606615e-05,
      "loss": 3.64,
      "step": 613
    },
    {
      "epoch": 0.021020198562136255,
      "grad_norm": 0.0,
      "learning_rate": 1.4002280501710376e-05,
      "loss": 3.6747,
      "step": 614
    },
    {
      "epoch": 0.02105443341321465,
      "grad_norm": 0.0,
      "learning_rate": 1.4025085518814141e-05,
      "loss": 3.6879,
      "step": 615
    },
    {
      "epoch": 0.02108866826429305,
      "grad_norm": 0.0,
      "learning_rate": 1.4047890535917902e-05,
      "loss": 3.6797,
      "step": 616
    },
    {
      "epoch": 0.021122903115371447,
      "grad_norm": 0.0,
      "learning_rate": 1.4070695553021665e-05,
      "loss": 3.6744,
      "step": 617
    },
    {
      "epoch": 0.021157137966449847,
      "grad_norm": 0.0,
      "learning_rate": 1.4093500570125429e-05,
      "loss": 3.6909,
      "step": 618
    },
    {
      "epoch": 0.021191372817528243,
      "grad_norm": 0.0,
      "learning_rate": 1.4116305587229192e-05,
      "loss": 3.6714,
      "step": 619
    },
    {
      "epoch": 0.021225607668606643,
      "grad_norm": 0.0,
      "learning_rate": 1.4139110604332955e-05,
      "loss": 3.6744,
      "step": 620
    },
    {
      "epoch": 0.02125984251968504,
      "grad_norm": 0.0,
      "learning_rate": 1.4161915621436716e-05,
      "loss": 3.6636,
      "step": 621
    },
    {
      "epoch": 0.02129407737076344,
      "grad_norm": 0.0,
      "learning_rate": 1.418472063854048e-05,
      "loss": 3.7041,
      "step": 622
    },
    {
      "epoch": 0.021328312221841835,
      "grad_norm": 0.0,
      "learning_rate": 1.4207525655644243e-05,
      "loss": 3.6944,
      "step": 623
    },
    {
      "epoch": 0.02136254707292023,
      "grad_norm": 0.0,
      "learning_rate": 1.4230330672748006e-05,
      "loss": 3.6541,
      "step": 624
    },
    {
      "epoch": 0.02139678192399863,
      "grad_norm": 0.0,
      "learning_rate": 1.4253135689851769e-05,
      "loss": 3.6657,
      "step": 625
    },
    {
      "epoch": 0.021431016775077028,
      "grad_norm": 0.0,
      "learning_rate": 1.4275940706955532e-05,
      "loss": 3.6517,
      "step": 626
    },
    {
      "epoch": 0.021465251626155427,
      "grad_norm": 0.0,
      "learning_rate": 1.4298745724059295e-05,
      "loss": 3.6188,
      "step": 627
    },
    {
      "epoch": 0.021499486477233824,
      "grad_norm": 0.0,
      "learning_rate": 1.4321550741163057e-05,
      "loss": 3.754,
      "step": 628
    },
    {
      "epoch": 0.021533721328312223,
      "grad_norm": 0.0,
      "learning_rate": 1.434435575826682e-05,
      "loss": 3.6528,
      "step": 629
    },
    {
      "epoch": 0.02156795617939062,
      "grad_norm": 0.0,
      "learning_rate": 1.4367160775370583e-05,
      "loss": 3.6875,
      "step": 630
    },
    {
      "epoch": 0.021602191030469016,
      "grad_norm": 0.0,
      "learning_rate": 1.4389965792474346e-05,
      "loss": 3.6855,
      "step": 631
    },
    {
      "epoch": 0.021636425881547416,
      "grad_norm": 0.0,
      "learning_rate": 1.4412770809578109e-05,
      "loss": 3.6278,
      "step": 632
    },
    {
      "epoch": 0.021670660732625812,
      "grad_norm": 0.0,
      "learning_rate": 1.443557582668187e-05,
      "loss": 3.6501,
      "step": 633
    },
    {
      "epoch": 0.021704895583704212,
      "grad_norm": 0.0,
      "learning_rate": 1.4458380843785634e-05,
      "loss": 3.6278,
      "step": 634
    },
    {
      "epoch": 0.021739130434782608,
      "grad_norm": 0.0,
      "learning_rate": 1.4481185860889397e-05,
      "loss": 3.6729,
      "step": 635
    },
    {
      "epoch": 0.021773365285861008,
      "grad_norm": 0.0,
      "learning_rate": 1.450399087799316e-05,
      "loss": 3.6527,
      "step": 636
    },
    {
      "epoch": 0.021807600136939404,
      "grad_norm": 0.0,
      "learning_rate": 1.4526795895096923e-05,
      "loss": 3.6598,
      "step": 637
    },
    {
      "epoch": 0.0218418349880178,
      "grad_norm": 0.0,
      "learning_rate": 1.4549600912200686e-05,
      "loss": 3.6598,
      "step": 638
    },
    {
      "epoch": 0.0218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 1.4572405929304448e-05,
      "loss": 3.651,
      "step": 639
    },
    {
      "epoch": 0.021910304690174597,
      "grad_norm": 0.0,
      "learning_rate": 1.459521094640821e-05,
      "loss": 3.6474,
      "step": 640
    },
    {
      "epoch": 0.021944539541252996,
      "grad_norm": 0.0,
      "learning_rate": 1.4618015963511974e-05,
      "loss": 3.6556,
      "step": 641
    },
    {
      "epoch": 0.021978774392331393,
      "grad_norm": 0.0,
      "learning_rate": 1.4640820980615737e-05,
      "loss": 3.6585,
      "step": 642
    },
    {
      "epoch": 0.022013009243409792,
      "grad_norm": 0.0,
      "learning_rate": 1.46636259977195e-05,
      "loss": 3.6275,
      "step": 643
    },
    {
      "epoch": 0.02204724409448819,
      "grad_norm": 0.0,
      "learning_rate": 1.4686431014823262e-05,
      "loss": 3.6338,
      "step": 644
    },
    {
      "epoch": 0.02208147894556659,
      "grad_norm": 0.0,
      "learning_rate": 1.4709236031927025e-05,
      "loss": 3.6101,
      "step": 645
    },
    {
      "epoch": 0.022115713796644985,
      "grad_norm": 0.0,
      "learning_rate": 1.4732041049030788e-05,
      "loss": 3.6107,
      "step": 646
    },
    {
      "epoch": 0.02214994864772338,
      "grad_norm": 0.0,
      "learning_rate": 1.4754846066134551e-05,
      "loss": 3.6422,
      "step": 647
    },
    {
      "epoch": 0.02218418349880178,
      "grad_norm": 0.0,
      "learning_rate": 1.4777651083238314e-05,
      "loss": 3.6247,
      "step": 648
    },
    {
      "epoch": 0.022218418349880177,
      "grad_norm": 0.0,
      "learning_rate": 1.4800456100342076e-05,
      "loss": 3.6007,
      "step": 649
    },
    {
      "epoch": 0.022252653200958577,
      "grad_norm": 0.0,
      "learning_rate": 1.482326111744584e-05,
      "loss": 3.5964,
      "step": 650
    },
    {
      "epoch": 0.022286888052036973,
      "grad_norm": 0.0,
      "learning_rate": 1.4846066134549602e-05,
      "loss": 3.5806,
      "step": 651
    },
    {
      "epoch": 0.022321122903115373,
      "grad_norm": 0.0,
      "learning_rate": 1.4868871151653365e-05,
      "loss": 3.6132,
      "step": 652
    },
    {
      "epoch": 0.02235535775419377,
      "grad_norm": 0.0,
      "learning_rate": 1.4891676168757128e-05,
      "loss": 3.5962,
      "step": 653
    },
    {
      "epoch": 0.022389592605272166,
      "grad_norm": 0.0,
      "learning_rate": 1.491448118586089e-05,
      "loss": 3.5864,
      "step": 654
    },
    {
      "epoch": 0.022423827456350565,
      "grad_norm": 0.0,
      "learning_rate": 1.4937286202964654e-05,
      "loss": 3.6361,
      "step": 655
    },
    {
      "epoch": 0.02245806230742896,
      "grad_norm": 0.0,
      "learning_rate": 1.4960091220068416e-05,
      "loss": 3.6024,
      "step": 656
    },
    {
      "epoch": 0.02249229715850736,
      "grad_norm": 0.0,
      "learning_rate": 1.498289623717218e-05,
      "loss": 3.6025,
      "step": 657
    },
    {
      "epoch": 0.022526532009585758,
      "grad_norm": 0.0,
      "learning_rate": 1.5005701254275942e-05,
      "loss": 3.6016,
      "step": 658
    },
    {
      "epoch": 0.022560766860664157,
      "grad_norm": 0.0,
      "learning_rate": 1.5028506271379705e-05,
      "loss": 3.604,
      "step": 659
    },
    {
      "epoch": 0.022595001711742554,
      "grad_norm": 0.0,
      "learning_rate": 1.5051311288483468e-05,
      "loss": 3.5995,
      "step": 660
    },
    {
      "epoch": 0.02262923656282095,
      "grad_norm": 0.0,
      "learning_rate": 1.507411630558723e-05,
      "loss": 3.6311,
      "step": 661
    },
    {
      "epoch": 0.02266347141389935,
      "grad_norm": 0.0,
      "learning_rate": 1.5096921322690995e-05,
      "loss": 3.6943,
      "step": 662
    },
    {
      "epoch": 0.022697706264977746,
      "grad_norm": 0.0,
      "learning_rate": 1.5119726339794756e-05,
      "loss": 3.6374,
      "step": 663
    },
    {
      "epoch": 0.022731941116056146,
      "grad_norm": 0.0,
      "learning_rate": 1.5142531356898519e-05,
      "loss": 3.6351,
      "step": 664
    },
    {
      "epoch": 0.022766175967134542,
      "grad_norm": 0.0,
      "learning_rate": 1.5165336374002282e-05,
      "loss": 3.6602,
      "step": 665
    },
    {
      "epoch": 0.022800410818212942,
      "grad_norm": 0.0,
      "learning_rate": 1.5188141391106044e-05,
      "loss": 3.5879,
      "step": 666
    },
    {
      "epoch": 0.02283464566929134,
      "grad_norm": 0.0,
      "learning_rate": 1.5210946408209808e-05,
      "loss": 3.6284,
      "step": 667
    },
    {
      "epoch": 0.022868880520369738,
      "grad_norm": 0.0,
      "learning_rate": 1.523375142531357e-05,
      "loss": 3.6261,
      "step": 668
    },
    {
      "epoch": 0.022903115371448134,
      "grad_norm": 0.0,
      "learning_rate": 1.5256556442417335e-05,
      "loss": 3.604,
      "step": 669
    },
    {
      "epoch": 0.02293735022252653,
      "grad_norm": 0.0,
      "learning_rate": 1.5279361459521094e-05,
      "loss": 3.6024,
      "step": 670
    },
    {
      "epoch": 0.02297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 1.5302166476624858e-05,
      "loss": 3.6155,
      "step": 671
    },
    {
      "epoch": 0.023005819924683327,
      "grad_norm": 0.0,
      "learning_rate": 1.532497149372862e-05,
      "loss": 3.6053,
      "step": 672
    },
    {
      "epoch": 0.023040054775761726,
      "grad_norm": 0.0,
      "learning_rate": 1.5347776510832384e-05,
      "loss": 3.5898,
      "step": 673
    },
    {
      "epoch": 0.023074289626840123,
      "grad_norm": 0.0,
      "learning_rate": 1.5370581527936147e-05,
      "loss": 3.5842,
      "step": 674
    },
    {
      "epoch": 0.023108524477918523,
      "grad_norm": 0.0,
      "learning_rate": 1.539338654503991e-05,
      "loss": 3.5987,
      "step": 675
    },
    {
      "epoch": 0.02314275932899692,
      "grad_norm": 0.0,
      "learning_rate": 1.5416191562143673e-05,
      "loss": 3.6016,
      "step": 676
    },
    {
      "epoch": 0.023176994180075315,
      "grad_norm": 0.0,
      "learning_rate": 1.5438996579247436e-05,
      "loss": 3.637,
      "step": 677
    },
    {
      "epoch": 0.023211229031153715,
      "grad_norm": 0.0,
      "learning_rate": 1.54618015963512e-05,
      "loss": 3.6105,
      "step": 678
    },
    {
      "epoch": 0.02324546388223211,
      "grad_norm": 0.0,
      "learning_rate": 1.5484606613454963e-05,
      "loss": 3.5789,
      "step": 679
    },
    {
      "epoch": 0.02327969873331051,
      "grad_norm": 0.0,
      "learning_rate": 1.5507411630558726e-05,
      "loss": 3.6315,
      "step": 680
    },
    {
      "epoch": 0.023313933584388907,
      "grad_norm": 0.0,
      "learning_rate": 1.5530216647662486e-05,
      "loss": 3.5982,
      "step": 681
    },
    {
      "epoch": 0.023348168435467307,
      "grad_norm": 0.0,
      "learning_rate": 1.555302166476625e-05,
      "loss": 3.588,
      "step": 682
    },
    {
      "epoch": 0.023382403286545703,
      "grad_norm": 0.0,
      "learning_rate": 1.5575826681870012e-05,
      "loss": 3.5843,
      "step": 683
    },
    {
      "epoch": 0.0234166381376241,
      "grad_norm": 0.0,
      "learning_rate": 1.5598631698973775e-05,
      "loss": 3.5772,
      "step": 684
    },
    {
      "epoch": 0.0234508729887025,
      "grad_norm": 0.0,
      "learning_rate": 1.5621436716077538e-05,
      "loss": 3.5751,
      "step": 685
    },
    {
      "epoch": 0.023485107839780896,
      "grad_norm": 0.0,
      "learning_rate": 1.56442417331813e-05,
      "loss": 3.5467,
      "step": 686
    },
    {
      "epoch": 0.023519342690859295,
      "grad_norm": 0.0,
      "learning_rate": 1.5667046750285064e-05,
      "loss": 3.5615,
      "step": 687
    },
    {
      "epoch": 0.023553577541937692,
      "grad_norm": 0.0,
      "learning_rate": 1.5689851767388827e-05,
      "loss": 3.5687,
      "step": 688
    },
    {
      "epoch": 0.02358781239301609,
      "grad_norm": 0.0,
      "learning_rate": 1.571265678449259e-05,
      "loss": 3.5848,
      "step": 689
    },
    {
      "epoch": 0.023622047244094488,
      "grad_norm": 0.0,
      "learning_rate": 1.5735461801596354e-05,
      "loss": 3.5655,
      "step": 690
    },
    {
      "epoch": 0.023656282095172888,
      "grad_norm": 0.0,
      "learning_rate": 1.5758266818700113e-05,
      "loss": 3.5677,
      "step": 691
    },
    {
      "epoch": 0.023690516946251284,
      "grad_norm": 0.0,
      "learning_rate": 1.578107183580388e-05,
      "loss": 3.5524,
      "step": 692
    },
    {
      "epoch": 0.02372475179732968,
      "grad_norm": 0.0,
      "learning_rate": 1.580387685290764e-05,
      "loss": 3.5579,
      "step": 693
    },
    {
      "epoch": 0.02375898664840808,
      "grad_norm": 0.0,
      "learning_rate": 1.5826681870011403e-05,
      "loss": 3.5488,
      "step": 694
    },
    {
      "epoch": 0.023793221499486476,
      "grad_norm": 0.0,
      "learning_rate": 1.5849486887115166e-05,
      "loss": 3.5685,
      "step": 695
    },
    {
      "epoch": 0.023827456350564876,
      "grad_norm": 0.0,
      "learning_rate": 1.587229190421893e-05,
      "loss": 3.5618,
      "step": 696
    },
    {
      "epoch": 0.023861691201643272,
      "grad_norm": 0.0,
      "learning_rate": 1.5895096921322692e-05,
      "loss": 3.5364,
      "step": 697
    },
    {
      "epoch": 0.023895926052721672,
      "grad_norm": 0.0,
      "learning_rate": 1.5917901938426455e-05,
      "loss": 3.551,
      "step": 698
    },
    {
      "epoch": 0.02393016090380007,
      "grad_norm": 0.0,
      "learning_rate": 1.594070695553022e-05,
      "loss": 3.5399,
      "step": 699
    },
    {
      "epoch": 0.023964395754878465,
      "grad_norm": 0.0,
      "learning_rate": 1.596351197263398e-05,
      "loss": 3.547,
      "step": 700
    },
    {
      "epoch": 0.023998630605956865,
      "grad_norm": 0.0,
      "learning_rate": 1.5986316989737745e-05,
      "loss": 3.5744,
      "step": 701
    },
    {
      "epoch": 0.02403286545703526,
      "grad_norm": 0.0,
      "learning_rate": 1.6009122006841508e-05,
      "loss": 3.5085,
      "step": 702
    },
    {
      "epoch": 0.02406710030811366,
      "grad_norm": 0.0,
      "learning_rate": 1.6031927023945268e-05,
      "loss": 3.5579,
      "step": 703
    },
    {
      "epoch": 0.024101335159192057,
      "grad_norm": 0.0,
      "learning_rate": 1.6054732041049034e-05,
      "loss": 3.5479,
      "step": 704
    },
    {
      "epoch": 0.024135570010270457,
      "grad_norm": 0.0,
      "learning_rate": 1.6077537058152794e-05,
      "loss": 3.5934,
      "step": 705
    },
    {
      "epoch": 0.024169804861348853,
      "grad_norm": 0.0,
      "learning_rate": 1.6100342075256557e-05,
      "loss": 3.5585,
      "step": 706
    },
    {
      "epoch": 0.02420403971242725,
      "grad_norm": 0.0,
      "learning_rate": 1.612314709236032e-05,
      "loss": 3.5702,
      "step": 707
    },
    {
      "epoch": 0.02423827456350565,
      "grad_norm": 0.0,
      "learning_rate": 1.6145952109464083e-05,
      "loss": 3.5891,
      "step": 708
    },
    {
      "epoch": 0.024272509414584045,
      "grad_norm": 0.0,
      "learning_rate": 1.6168757126567846e-05,
      "loss": 3.5252,
      "step": 709
    },
    {
      "epoch": 0.024306744265662445,
      "grad_norm": 0.0,
      "learning_rate": 1.619156214367161e-05,
      "loss": 3.5307,
      "step": 710
    },
    {
      "epoch": 0.02434097911674084,
      "grad_norm": 0.0,
      "learning_rate": 1.6214367160775373e-05,
      "loss": 3.5338,
      "step": 711
    },
    {
      "epoch": 0.02437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 1.6237172177879136e-05,
      "loss": 3.5236,
      "step": 712
    },
    {
      "epoch": 0.024409448818897637,
      "grad_norm": 0.0,
      "learning_rate": 1.6259977194982896e-05,
      "loss": 3.5048,
      "step": 713
    },
    {
      "epoch": 0.024443683669976037,
      "grad_norm": 0.0,
      "learning_rate": 1.6282782212086662e-05,
      "loss": 3.516,
      "step": 714
    },
    {
      "epoch": 0.024477918521054434,
      "grad_norm": 0.0,
      "learning_rate": 1.6305587229190422e-05,
      "loss": 3.5144,
      "step": 715
    },
    {
      "epoch": 0.02451215337213283,
      "grad_norm": 0.0,
      "learning_rate": 1.632839224629419e-05,
      "loss": 3.5386,
      "step": 716
    },
    {
      "epoch": 0.02454638822321123,
      "grad_norm": 0.0,
      "learning_rate": 1.6351197263397948e-05,
      "loss": 3.5166,
      "step": 717
    },
    {
      "epoch": 0.024580623074289626,
      "grad_norm": 0.0,
      "learning_rate": 1.637400228050171e-05,
      "loss": 3.5232,
      "step": 718
    },
    {
      "epoch": 0.024614857925368026,
      "grad_norm": 0.0,
      "learning_rate": 1.6396807297605474e-05,
      "loss": 3.52,
      "step": 719
    },
    {
      "epoch": 0.024649092776446422,
      "grad_norm": 0.0,
      "learning_rate": 1.6419612314709237e-05,
      "loss": 3.4741,
      "step": 720
    },
    {
      "epoch": 0.02468332762752482,
      "grad_norm": 0.0,
      "learning_rate": 1.6442417331813e-05,
      "loss": 3.5132,
      "step": 721
    },
    {
      "epoch": 0.024717562478603218,
      "grad_norm": 0.0,
      "learning_rate": 1.6465222348916764e-05,
      "loss": 3.5169,
      "step": 722
    },
    {
      "epoch": 0.024751797329681614,
      "grad_norm": 0.0,
      "learning_rate": 1.6488027366020524e-05,
      "loss": 3.4896,
      "step": 723
    },
    {
      "epoch": 0.024786032180760014,
      "grad_norm": 0.0,
      "learning_rate": 1.651083238312429e-05,
      "loss": 3.4981,
      "step": 724
    },
    {
      "epoch": 0.02482026703183841,
      "grad_norm": 0.0,
      "learning_rate": 1.653363740022805e-05,
      "loss": 3.4727,
      "step": 725
    },
    {
      "epoch": 0.02485450188291681,
      "grad_norm": 0.0,
      "learning_rate": 1.6556442417331816e-05,
      "loss": 3.452,
      "step": 726
    },
    {
      "epoch": 0.024888736733995206,
      "grad_norm": 0.0,
      "learning_rate": 1.6579247434435576e-05,
      "loss": 3.5303,
      "step": 727
    },
    {
      "epoch": 0.024922971585073606,
      "grad_norm": 0.0,
      "learning_rate": 1.6602052451539343e-05,
      "loss": 3.5084,
      "step": 728
    },
    {
      "epoch": 0.024957206436152003,
      "grad_norm": 0.0,
      "learning_rate": 1.6624857468643102e-05,
      "loss": 3.5041,
      "step": 729
    },
    {
      "epoch": 0.0249914412872304,
      "grad_norm": 0.0,
      "learning_rate": 1.6647662485746865e-05,
      "loss": 3.5341,
      "step": 730
    },
    {
      "epoch": 0.0250256761383088,
      "grad_norm": 0.0,
      "learning_rate": 1.667046750285063e-05,
      "loss": 3.4937,
      "step": 731
    },
    {
      "epoch": 0.025059910989387195,
      "grad_norm": 0.0,
      "learning_rate": 1.669327251995439e-05,
      "loss": 3.4667,
      "step": 732
    },
    {
      "epoch": 0.025094145840465595,
      "grad_norm": 0.0,
      "learning_rate": 1.6716077537058155e-05,
      "loss": 3.4762,
      "step": 733
    },
    {
      "epoch": 0.02512838069154399,
      "grad_norm": 0.0,
      "learning_rate": 1.6738882554161918e-05,
      "loss": 3.5008,
      "step": 734
    },
    {
      "epoch": 0.02516261554262239,
      "grad_norm": 0.0,
      "learning_rate": 1.6761687571265678e-05,
      "loss": 3.5153,
      "step": 735
    },
    {
      "epoch": 0.025196850393700787,
      "grad_norm": 0.0,
      "learning_rate": 1.6784492588369444e-05,
      "loss": 3.484,
      "step": 736
    },
    {
      "epoch": 0.025231085244779187,
      "grad_norm": 0.0,
      "learning_rate": 1.6807297605473204e-05,
      "loss": 3.5005,
      "step": 737
    },
    {
      "epoch": 0.025265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 1.683010262257697e-05,
      "loss": 3.447,
      "step": 738
    },
    {
      "epoch": 0.02529955494693598,
      "grad_norm": 0.0,
      "learning_rate": 1.685290763968073e-05,
      "loss": 3.4784,
      "step": 739
    },
    {
      "epoch": 0.02533378979801438,
      "grad_norm": 0.0,
      "learning_rate": 1.6875712656784493e-05,
      "loss": 3.4455,
      "step": 740
    },
    {
      "epoch": 0.025368024649092775,
      "grad_norm": 0.0,
      "learning_rate": 1.6898517673888256e-05,
      "loss": 3.507,
      "step": 741
    },
    {
      "epoch": 0.025402259500171175,
      "grad_norm": 0.0,
      "learning_rate": 1.692132269099202e-05,
      "loss": 3.4927,
      "step": 742
    },
    {
      "epoch": 0.02543649435124957,
      "grad_norm": 0.0,
      "learning_rate": 1.6944127708095783e-05,
      "loss": 3.4729,
      "step": 743
    },
    {
      "epoch": 0.02547072920232797,
      "grad_norm": 0.0,
      "learning_rate": 1.6966932725199546e-05,
      "loss": 3.4639,
      "step": 744
    },
    {
      "epoch": 0.025504964053406368,
      "grad_norm": 0.0,
      "learning_rate": 1.6989737742303306e-05,
      "loss": 3.4763,
      "step": 745
    },
    {
      "epoch": 0.025539198904484764,
      "grad_norm": 0.0,
      "learning_rate": 1.7012542759407072e-05,
      "loss": 3.481,
      "step": 746
    },
    {
      "epoch": 0.025573433755563164,
      "grad_norm": 0.0,
      "learning_rate": 1.7035347776510832e-05,
      "loss": 3.4681,
      "step": 747
    },
    {
      "epoch": 0.02560766860664156,
      "grad_norm": 0.0,
      "learning_rate": 1.70581527936146e-05,
      "loss": 3.452,
      "step": 748
    },
    {
      "epoch": 0.02564190345771996,
      "grad_norm": 0.0,
      "learning_rate": 1.7080957810718358e-05,
      "loss": 3.4654,
      "step": 749
    },
    {
      "epoch": 0.025676138308798356,
      "grad_norm": 0.0,
      "learning_rate": 1.710376282782212e-05,
      "loss": 3.47,
      "step": 750
    },
    {
      "epoch": 0.025710373159876756,
      "grad_norm": 0.0,
      "learning_rate": 1.7126567844925884e-05,
      "loss": 3.4596,
      "step": 751
    },
    {
      "epoch": 0.025744608010955152,
      "grad_norm": 0.0,
      "learning_rate": 1.7149372862029648e-05,
      "loss": 3.4585,
      "step": 752
    },
    {
      "epoch": 0.02577884286203355,
      "grad_norm": 0.0,
      "learning_rate": 1.717217787913341e-05,
      "loss": 3.4784,
      "step": 753
    },
    {
      "epoch": 0.025813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 1.7194982896237174e-05,
      "loss": 3.4497,
      "step": 754
    },
    {
      "epoch": 0.025847312564190345,
      "grad_norm": 0.0,
      "learning_rate": 1.7217787913340937e-05,
      "loss": 3.4606,
      "step": 755
    },
    {
      "epoch": 0.025881547415268744,
      "grad_norm": 0.0,
      "learning_rate": 1.72405929304447e-05,
      "loss": 3.4789,
      "step": 756
    },
    {
      "epoch": 0.02591578226634714,
      "grad_norm": 0.0,
      "learning_rate": 1.726339794754846e-05,
      "loss": 3.4955,
      "step": 757
    },
    {
      "epoch": 0.02595001711742554,
      "grad_norm": 0.0,
      "learning_rate": 1.7286202964652226e-05,
      "loss": 3.4565,
      "step": 758
    },
    {
      "epoch": 0.025984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 1.7309007981755986e-05,
      "loss": 3.5113,
      "step": 759
    },
    {
      "epoch": 0.026018486819582336,
      "grad_norm": 0.0,
      "learning_rate": 1.7331812998859753e-05,
      "loss": 3.4545,
      "step": 760
    },
    {
      "epoch": 0.026052721670660733,
      "grad_norm": 0.0,
      "learning_rate": 1.7354618015963512e-05,
      "loss": 3.4356,
      "step": 761
    },
    {
      "epoch": 0.02608695652173913,
      "grad_norm": 0.0,
      "learning_rate": 1.7377423033067275e-05,
      "loss": 3.4411,
      "step": 762
    },
    {
      "epoch": 0.02612119137281753,
      "grad_norm": 0.0,
      "learning_rate": 1.740022805017104e-05,
      "loss": 3.4374,
      "step": 763
    },
    {
      "epoch": 0.026155426223895925,
      "grad_norm": 0.0,
      "learning_rate": 1.7423033067274802e-05,
      "loss": 3.4496,
      "step": 764
    },
    {
      "epoch": 0.026189661074974325,
      "grad_norm": 0.0,
      "learning_rate": 1.7445838084378565e-05,
      "loss": 3.4419,
      "step": 765
    },
    {
      "epoch": 0.02622389592605272,
      "grad_norm": 0.0,
      "learning_rate": 1.7468643101482328e-05,
      "loss": 3.453,
      "step": 766
    },
    {
      "epoch": 0.02625813077713112,
      "grad_norm": 0.0,
      "learning_rate": 1.749144811858609e-05,
      "loss": 3.4572,
      "step": 767
    },
    {
      "epoch": 0.026292365628209517,
      "grad_norm": 0.0,
      "learning_rate": 1.7514253135689854e-05,
      "loss": 3.4478,
      "step": 768
    },
    {
      "epoch": 0.026326600479287914,
      "grad_norm": 0.0,
      "learning_rate": 1.7537058152793617e-05,
      "loss": 3.4614,
      "step": 769
    },
    {
      "epoch": 0.026360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 1.755986316989738e-05,
      "loss": 3.4075,
      "step": 770
    },
    {
      "epoch": 0.02639507018144471,
      "grad_norm": 0.0,
      "learning_rate": 1.758266818700114e-05,
      "loss": 3.433,
      "step": 771
    },
    {
      "epoch": 0.02642930503252311,
      "grad_norm": 0.0,
      "learning_rate": 1.7605473204104903e-05,
      "loss": 3.3804,
      "step": 772
    },
    {
      "epoch": 0.026463539883601506,
      "grad_norm": 0.0,
      "learning_rate": 1.7628278221208667e-05,
      "loss": 3.4522,
      "step": 773
    },
    {
      "epoch": 0.026497774734679905,
      "grad_norm": 0.0,
      "learning_rate": 1.765108323831243e-05,
      "loss": 3.4299,
      "step": 774
    },
    {
      "epoch": 0.0265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 1.7673888255416193e-05,
      "loss": 3.4526,
      "step": 775
    },
    {
      "epoch": 0.026566244436836698,
      "grad_norm": 0.0,
      "learning_rate": 1.7696693272519956e-05,
      "loss": 3.4185,
      "step": 776
    },
    {
      "epoch": 0.026600479287915098,
      "grad_norm": 0.0,
      "learning_rate": 1.771949828962372e-05,
      "loss": 3.4439,
      "step": 777
    },
    {
      "epoch": 0.026634714138993494,
      "grad_norm": 0.0,
      "learning_rate": 1.7742303306727482e-05,
      "loss": 3.434,
      "step": 778
    },
    {
      "epoch": 0.026668948990071894,
      "grad_norm": 0.0,
      "learning_rate": 1.7765108323831245e-05,
      "loss": 3.3884,
      "step": 779
    },
    {
      "epoch": 0.02670318384115029,
      "grad_norm": 0.0,
      "learning_rate": 1.778791334093501e-05,
      "loss": 3.4368,
      "step": 780
    },
    {
      "epoch": 0.02673741869222869,
      "grad_norm": 0.0,
      "learning_rate": 1.781071835803877e-05,
      "loss": 3.3913,
      "step": 781
    },
    {
      "epoch": 0.026771653543307086,
      "grad_norm": 0.0,
      "learning_rate": 1.783352337514253e-05,
      "loss": 3.411,
      "step": 782
    },
    {
      "epoch": 0.026805888394385486,
      "grad_norm": 0.0,
      "learning_rate": 1.7856328392246294e-05,
      "loss": 3.4476,
      "step": 783
    },
    {
      "epoch": 0.026840123245463882,
      "grad_norm": 0.0,
      "learning_rate": 1.7879133409350058e-05,
      "loss": 3.4574,
      "step": 784
    },
    {
      "epoch": 0.02687435809654228,
      "grad_norm": 0.0,
      "learning_rate": 1.790193842645382e-05,
      "loss": 3.4173,
      "step": 785
    },
    {
      "epoch": 0.02690859294762068,
      "grad_norm": 0.0,
      "learning_rate": 1.7924743443557584e-05,
      "loss": 3.4175,
      "step": 786
    },
    {
      "epoch": 0.026942827798699075,
      "grad_norm": 0.0,
      "learning_rate": 1.7947548460661347e-05,
      "loss": 3.4455,
      "step": 787
    },
    {
      "epoch": 0.026977062649777474,
      "grad_norm": 0.0,
      "learning_rate": 1.797035347776511e-05,
      "loss": 3.4331,
      "step": 788
    },
    {
      "epoch": 0.02701129750085587,
      "grad_norm": 0.0,
      "learning_rate": 1.7993158494868873e-05,
      "loss": 3.399,
      "step": 789
    },
    {
      "epoch": 0.02704553235193427,
      "grad_norm": 0.0,
      "learning_rate": 1.8015963511972636e-05,
      "loss": 3.4155,
      "step": 790
    },
    {
      "epoch": 0.027079767203012667,
      "grad_norm": 0.0,
      "learning_rate": 1.80387685290764e-05,
      "loss": 3.4004,
      "step": 791
    },
    {
      "epoch": 0.027114002054091063,
      "grad_norm": 0.0,
      "learning_rate": 1.806157354618016e-05,
      "loss": 3.4279,
      "step": 792
    },
    {
      "epoch": 0.027148236905169463,
      "grad_norm": 0.0,
      "learning_rate": 1.8084378563283926e-05,
      "loss": 3.3933,
      "step": 793
    },
    {
      "epoch": 0.02718247175624786,
      "grad_norm": 0.0,
      "learning_rate": 1.8107183580387685e-05,
      "loss": 3.3947,
      "step": 794
    },
    {
      "epoch": 0.02721670660732626,
      "grad_norm": 0.0,
      "learning_rate": 1.812998859749145e-05,
      "loss": 3.4034,
      "step": 795
    },
    {
      "epoch": 0.027250941458404655,
      "grad_norm": 0.0,
      "learning_rate": 1.8152793614595212e-05,
      "loss": 3.3863,
      "step": 796
    },
    {
      "epoch": 0.027285176309483055,
      "grad_norm": 0.0,
      "learning_rate": 1.8175598631698975e-05,
      "loss": 3.3729,
      "step": 797
    },
    {
      "epoch": 0.02731941116056145,
      "grad_norm": 0.0,
      "learning_rate": 1.8198403648802738e-05,
      "loss": 3.3837,
      "step": 798
    },
    {
      "epoch": 0.02735364601163985,
      "grad_norm": 0.0,
      "learning_rate": 1.82212086659065e-05,
      "loss": 3.383,
      "step": 799
    },
    {
      "epoch": 0.027387880862718247,
      "grad_norm": 0.0,
      "learning_rate": 1.8244013683010264e-05,
      "loss": 3.3816,
      "step": 800
    },
    {
      "epoch": 0.027422115713796644,
      "grad_norm": 0.0,
      "learning_rate": 1.8266818700114027e-05,
      "loss": 3.3623,
      "step": 801
    },
    {
      "epoch": 0.027456350564875043,
      "grad_norm": 0.0,
      "learning_rate": 1.828962371721779e-05,
      "loss": 3.4184,
      "step": 802
    },
    {
      "epoch": 0.02749058541595344,
      "grad_norm": 0.0,
      "learning_rate": 1.8312428734321554e-05,
      "loss": 3.3784,
      "step": 803
    },
    {
      "epoch": 0.02752482026703184,
      "grad_norm": 0.0,
      "learning_rate": 1.8335233751425313e-05,
      "loss": 3.401,
      "step": 804
    },
    {
      "epoch": 0.027559055118110236,
      "grad_norm": 0.0,
      "learning_rate": 1.835803876852908e-05,
      "loss": 3.3283,
      "step": 805
    },
    {
      "epoch": 0.027593289969188636,
      "grad_norm": 0.0,
      "learning_rate": 1.838084378563284e-05,
      "loss": 3.3548,
      "step": 806
    },
    {
      "epoch": 0.027627524820267032,
      "grad_norm": 0.0,
      "learning_rate": 1.8403648802736603e-05,
      "loss": 3.3714,
      "step": 807
    },
    {
      "epoch": 0.027661759671345428,
      "grad_norm": 0.0,
      "learning_rate": 1.8426453819840366e-05,
      "loss": 3.3503,
      "step": 808
    },
    {
      "epoch": 0.027695994522423828,
      "grad_norm": 0.0,
      "learning_rate": 1.844925883694413e-05,
      "loss": 3.3564,
      "step": 809
    },
    {
      "epoch": 0.027730229373502224,
      "grad_norm": 0.0,
      "learning_rate": 1.8472063854047892e-05,
      "loss": 3.3754,
      "step": 810
    },
    {
      "epoch": 0.027764464224580624,
      "grad_norm": 0.0,
      "learning_rate": 1.8494868871151655e-05,
      "loss": 3.3589,
      "step": 811
    },
    {
      "epoch": 0.02779869907565902,
      "grad_norm": 0.0,
      "learning_rate": 1.851767388825542e-05,
      "loss": 3.321,
      "step": 812
    },
    {
      "epoch": 0.02783293392673742,
      "grad_norm": 0.0,
      "learning_rate": 1.854047890535918e-05,
      "loss": 3.3618,
      "step": 813
    },
    {
      "epoch": 0.027867168777815816,
      "grad_norm": 0.0,
      "learning_rate": 1.856328392246294e-05,
      "loss": 3.3916,
      "step": 814
    },
    {
      "epoch": 0.027901403628894213,
      "grad_norm": 0.0,
      "learning_rate": 1.8586088939566708e-05,
      "loss": 3.3798,
      "step": 815
    },
    {
      "epoch": 0.027935638479972612,
      "grad_norm": 0.0,
      "learning_rate": 1.8608893956670468e-05,
      "loss": 3.3679,
      "step": 816
    },
    {
      "epoch": 0.02796987333105101,
      "grad_norm": 0.0,
      "learning_rate": 1.8631698973774234e-05,
      "loss": 3.3836,
      "step": 817
    },
    {
      "epoch": 0.02800410818212941,
      "grad_norm": 0.0,
      "learning_rate": 1.8654503990877994e-05,
      "loss": 3.3359,
      "step": 818
    },
    {
      "epoch": 0.028038343033207805,
      "grad_norm": 0.0,
      "learning_rate": 1.8677309007981757e-05,
      "loss": 3.3286,
      "step": 819
    },
    {
      "epoch": 0.028072577884286205,
      "grad_norm": 0.0,
      "learning_rate": 1.870011402508552e-05,
      "loss": 3.3416,
      "step": 820
    },
    {
      "epoch": 0.0281068127353646,
      "grad_norm": 0.0,
      "learning_rate": 1.8722919042189283e-05,
      "loss": 3.3662,
      "step": 821
    },
    {
      "epoch": 0.028141047586443,
      "grad_norm": 0.0,
      "learning_rate": 1.8745724059293046e-05,
      "loss": 3.3511,
      "step": 822
    },
    {
      "epoch": 0.028175282437521397,
      "grad_norm": 0.0,
      "learning_rate": 1.876852907639681e-05,
      "loss": 3.3499,
      "step": 823
    },
    {
      "epoch": 0.028209517288599793,
      "grad_norm": 0.0,
      "learning_rate": 1.879133409350057e-05,
      "loss": 3.3257,
      "step": 824
    },
    {
      "epoch": 0.028243752139678193,
      "grad_norm": 0.0,
      "learning_rate": 1.8814139110604336e-05,
      "loss": 3.3431,
      "step": 825
    },
    {
      "epoch": 0.02827798699075659,
      "grad_norm": 0.0,
      "learning_rate": 1.8836944127708096e-05,
      "loss": 3.344,
      "step": 826
    },
    {
      "epoch": 0.02831222184183499,
      "grad_norm": 0.0,
      "learning_rate": 1.8859749144811862e-05,
      "loss": 3.3382,
      "step": 827
    },
    {
      "epoch": 0.028346456692913385,
      "grad_norm": 0.0,
      "learning_rate": 1.8882554161915622e-05,
      "loss": 3.3278,
      "step": 828
    },
    {
      "epoch": 0.028380691543991785,
      "grad_norm": 0.0,
      "learning_rate": 1.8905359179019388e-05,
      "loss": 3.3334,
      "step": 829
    },
    {
      "epoch": 0.02841492639507018,
      "grad_norm": 0.0,
      "learning_rate": 1.8928164196123148e-05,
      "loss": 3.348,
      "step": 830
    },
    {
      "epoch": 0.028449161246148578,
      "grad_norm": 0.0,
      "learning_rate": 1.895096921322691e-05,
      "loss": 3.3276,
      "step": 831
    },
    {
      "epoch": 0.028483396097226978,
      "grad_norm": 0.0,
      "learning_rate": 1.8973774230330674e-05,
      "loss": 3.321,
      "step": 832
    },
    {
      "epoch": 0.028517630948305374,
      "grad_norm": 0.0,
      "learning_rate": 1.8996579247434437e-05,
      "loss": 3.3558,
      "step": 833
    },
    {
      "epoch": 0.028551865799383774,
      "grad_norm": 0.0,
      "learning_rate": 1.90193842645382e-05,
      "loss": 3.3546,
      "step": 834
    },
    {
      "epoch": 0.02858610065046217,
      "grad_norm": 0.0,
      "learning_rate": 1.9042189281641964e-05,
      "loss": 3.3385,
      "step": 835
    },
    {
      "epoch": 0.02862033550154057,
      "grad_norm": 0.0,
      "learning_rate": 1.9064994298745723e-05,
      "loss": 3.3447,
      "step": 836
    },
    {
      "epoch": 0.028654570352618966,
      "grad_norm": 0.0,
      "learning_rate": 1.908779931584949e-05,
      "loss": 3.3192,
      "step": 837
    },
    {
      "epoch": 0.028688805203697362,
      "grad_norm": 0.0,
      "learning_rate": 1.911060433295325e-05,
      "loss": 3.3357,
      "step": 838
    },
    {
      "epoch": 0.028723040054775762,
      "grad_norm": 0.0,
      "learning_rate": 1.9133409350057016e-05,
      "loss": 3.3251,
      "step": 839
    },
    {
      "epoch": 0.02875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 1.9156214367160776e-05,
      "loss": 3.327,
      "step": 840
    },
    {
      "epoch": 0.028791509756932558,
      "grad_norm": 0.0,
      "learning_rate": 1.917901938426454e-05,
      "loss": 3.3303,
      "step": 841
    },
    {
      "epoch": 0.028825744608010954,
      "grad_norm": 0.0,
      "learning_rate": 1.9201824401368302e-05,
      "loss": 3.3326,
      "step": 842
    },
    {
      "epoch": 0.028859979459089354,
      "grad_norm": 0.0,
      "learning_rate": 1.9224629418472065e-05,
      "loss": 3.3316,
      "step": 843
    },
    {
      "epoch": 0.02889421431016775,
      "grad_norm": 0.0,
      "learning_rate": 1.924743443557583e-05,
      "loss": 3.2824,
      "step": 844
    },
    {
      "epoch": 0.02892844916124615,
      "grad_norm": 0.0,
      "learning_rate": 1.927023945267959e-05,
      "loss": 3.3,
      "step": 845
    },
    {
      "epoch": 0.028962684012324547,
      "grad_norm": 0.0,
      "learning_rate": 1.929304446978335e-05,
      "loss": 3.3386,
      "step": 846
    },
    {
      "epoch": 0.028996918863402943,
      "grad_norm": 0.0,
      "learning_rate": 1.9315849486887118e-05,
      "loss": 3.2911,
      "step": 847
    },
    {
      "epoch": 0.029031153714481343,
      "grad_norm": 0.0,
      "learning_rate": 1.9338654503990878e-05,
      "loss": 3.3171,
      "step": 848
    },
    {
      "epoch": 0.02906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 1.9361459521094644e-05,
      "loss": 3.2823,
      "step": 849
    },
    {
      "epoch": 0.02909962341663814,
      "grad_norm": 0.0,
      "learning_rate": 1.9384264538198404e-05,
      "loss": 3.3092,
      "step": 850
    },
    {
      "epoch": 0.029133858267716535,
      "grad_norm": 0.0,
      "learning_rate": 1.9407069555302167e-05,
      "loss": 3.2831,
      "step": 851
    },
    {
      "epoch": 0.029168093118794935,
      "grad_norm": 0.0,
      "learning_rate": 1.942987457240593e-05,
      "loss": 3.2955,
      "step": 852
    },
    {
      "epoch": 0.02920232796987333,
      "grad_norm": 0.0,
      "learning_rate": 1.9452679589509693e-05,
      "loss": 3.3157,
      "step": 853
    },
    {
      "epoch": 0.029236562820951727,
      "grad_norm": 0.0,
      "learning_rate": 1.9475484606613456e-05,
      "loss": 3.2975,
      "step": 854
    },
    {
      "epoch": 0.029270797672030127,
      "grad_norm": 0.0,
      "learning_rate": 1.949828962371722e-05,
      "loss": 3.2821,
      "step": 855
    },
    {
      "epoch": 0.029305032523108523,
      "grad_norm": 0.0,
      "learning_rate": 1.9521094640820983e-05,
      "loss": 3.2889,
      "step": 856
    },
    {
      "epoch": 0.029339267374186923,
      "grad_norm": 0.0,
      "learning_rate": 1.9543899657924746e-05,
      "loss": 3.3337,
      "step": 857
    },
    {
      "epoch": 0.02937350222526532,
      "grad_norm": 0.0,
      "learning_rate": 1.956670467502851e-05,
      "loss": 3.2779,
      "step": 858
    },
    {
      "epoch": 0.02940773707634372,
      "grad_norm": 0.0,
      "learning_rate": 1.9589509692132272e-05,
      "loss": 3.2684,
      "step": 859
    },
    {
      "epoch": 0.029441971927422116,
      "grad_norm": 0.0,
      "learning_rate": 1.9612314709236032e-05,
      "loss": 3.2872,
      "step": 860
    },
    {
      "epoch": 0.029476206778500512,
      "grad_norm": 0.0,
      "learning_rate": 1.96351197263398e-05,
      "loss": 3.2874,
      "step": 861
    },
    {
      "epoch": 0.02951044162957891,
      "grad_norm": 0.0,
      "learning_rate": 1.9657924743443558e-05,
      "loss": 3.2669,
      "step": 862
    },
    {
      "epoch": 0.029544676480657308,
      "grad_norm": 0.0,
      "learning_rate": 1.968072976054732e-05,
      "loss": 3.2623,
      "step": 863
    },
    {
      "epoch": 0.029578911331735708,
      "grad_norm": 0.0,
      "learning_rate": 1.9703534777651084e-05,
      "loss": 3.2496,
      "step": 864
    },
    {
      "epoch": 0.029613146182814104,
      "grad_norm": 0.0,
      "learning_rate": 1.9726339794754847e-05,
      "loss": 3.2683,
      "step": 865
    },
    {
      "epoch": 0.029647381033892504,
      "grad_norm": 0.0,
      "learning_rate": 1.974914481185861e-05,
      "loss": 3.2889,
      "step": 866
    },
    {
      "epoch": 0.0296816158849709,
      "grad_norm": 0.0,
      "learning_rate": 1.9771949828962374e-05,
      "loss": 3.2819,
      "step": 867
    },
    {
      "epoch": 0.0297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 1.9794754846066137e-05,
      "loss": 3.3253,
      "step": 868
    },
    {
      "epoch": 0.029750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 1.98175598631699e-05,
      "loss": 3.2906,
      "step": 869
    },
    {
      "epoch": 0.029784320438206092,
      "grad_norm": 0.0,
      "learning_rate": 1.9840364880273663e-05,
      "loss": 3.3162,
      "step": 870
    },
    {
      "epoch": 0.029818555289284492,
      "grad_norm": 0.0,
      "learning_rate": 1.9863169897377426e-05,
      "loss": 3.2515,
      "step": 871
    },
    {
      "epoch": 0.02985279014036289,
      "grad_norm": 0.0,
      "learning_rate": 1.9885974914481186e-05,
      "loss": 3.3025,
      "step": 872
    },
    {
      "epoch": 0.02988702499144129,
      "grad_norm": 0.0,
      "learning_rate": 1.990877993158495e-05,
      "loss": 3.3039,
      "step": 873
    },
    {
      "epoch": 0.029921259842519685,
      "grad_norm": 0.0,
      "learning_rate": 1.9931584948688712e-05,
      "loss": 3.2831,
      "step": 874
    },
    {
      "epoch": 0.029955494693598084,
      "grad_norm": 0.0,
      "learning_rate": 1.9954389965792475e-05,
      "loss": 3.2711,
      "step": 875
    },
    {
      "epoch": 0.02998972954467648,
      "grad_norm": 0.0,
      "learning_rate": 1.997719498289624e-05,
      "loss": 3.2648,
      "step": 876
    },
    {
      "epoch": 0.030023964395754877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.301,
      "step": 877
    },
    {
      "epoch": 0.030058199246833277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2466,
      "step": 878
    },
    {
      "epoch": 0.030092434097911673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2757,
      "step": 879
    },
    {
      "epoch": 0.030126668948990073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2583,
      "step": 880
    },
    {
      "epoch": 0.03016090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.276,
      "step": 881
    },
    {
      "epoch": 0.03019513865114687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2569,
      "step": 882
    },
    {
      "epoch": 0.030229373502225265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2451,
      "step": 883
    },
    {
      "epoch": 0.03026360835330366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2309,
      "step": 884
    },
    {
      "epoch": 0.03029784320438206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2459,
      "step": 885
    },
    {
      "epoch": 0.030332078055460458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2565,
      "step": 886
    },
    {
      "epoch": 0.030366312906538857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2249,
      "step": 887
    },
    {
      "epoch": 0.030400547757617254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.229,
      "step": 888
    },
    {
      "epoch": 0.030434782608695653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2366,
      "step": 889
    },
    {
      "epoch": 0.03046901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2316,
      "step": 890
    },
    {
      "epoch": 0.03050325231085245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2334,
      "step": 891
    },
    {
      "epoch": 0.030537487161930846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.217,
      "step": 892
    },
    {
      "epoch": 0.030571722013009242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2449,
      "step": 893
    },
    {
      "epoch": 0.030605956864087642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.298,
      "step": 894
    },
    {
      "epoch": 0.030640191715166038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2796,
      "step": 895
    },
    {
      "epoch": 0.030674426566244438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2053,
      "step": 896
    },
    {
      "epoch": 0.030708661417322834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2463,
      "step": 897
    },
    {
      "epoch": 0.030742896268401234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2472,
      "step": 898
    },
    {
      "epoch": 0.03077713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2152,
      "step": 899
    },
    {
      "epoch": 0.030811365970558027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2608,
      "step": 900
    },
    {
      "epoch": 0.030845600821636426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2568,
      "step": 901
    },
    {
      "epoch": 0.030879835672714823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2298,
      "step": 902
    },
    {
      "epoch": 0.030914070523793222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2295,
      "step": 903
    },
    {
      "epoch": 0.03094830537487162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2399,
      "step": 904
    },
    {
      "epoch": 0.03098254022595002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2347,
      "step": 905
    },
    {
      "epoch": 0.031016775077028415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2152,
      "step": 906
    },
    {
      "epoch": 0.03105100992810681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2156,
      "step": 907
    },
    {
      "epoch": 0.03108524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2188,
      "step": 908
    },
    {
      "epoch": 0.031119479630263607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.207,
      "step": 909
    },
    {
      "epoch": 0.031153714481342007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2421,
      "step": 910
    },
    {
      "epoch": 0.031187949332420403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2108,
      "step": 911
    },
    {
      "epoch": 0.031222184183498803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2143,
      "step": 912
    },
    {
      "epoch": 0.0312564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.209,
      "step": 913
    },
    {
      "epoch": 0.0312906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2391,
      "step": 914
    },
    {
      "epoch": 0.031324888736733995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1794,
      "step": 915
    },
    {
      "epoch": 0.03135912358781239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1943,
      "step": 916
    },
    {
      "epoch": 0.03139335843889079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2167,
      "step": 917
    },
    {
      "epoch": 0.03142759328996919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1985,
      "step": 918
    },
    {
      "epoch": 0.03146182814104759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1961,
      "step": 919
    },
    {
      "epoch": 0.031496062992125984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2021,
      "step": 920
    },
    {
      "epoch": 0.03153029784320438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2261,
      "step": 921
    },
    {
      "epoch": 0.031564532694282776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1729,
      "step": 922
    },
    {
      "epoch": 0.03159876754536118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1607,
      "step": 923
    },
    {
      "epoch": 0.031633002396439576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2042,
      "step": 924
    },
    {
      "epoch": 0.03166723724751797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1828,
      "step": 925
    },
    {
      "epoch": 0.03170147209859637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.157,
      "step": 926
    },
    {
      "epoch": 0.03173570694967477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1607,
      "step": 927
    },
    {
      "epoch": 0.03176994180075317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1918,
      "step": 928
    },
    {
      "epoch": 0.031804176651831564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1849,
      "step": 929
    },
    {
      "epoch": 0.03183841150290996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1655,
      "step": 930
    },
    {
      "epoch": 0.03187264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1568,
      "step": 931
    },
    {
      "epoch": 0.03190688120506676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2095,
      "step": 932
    },
    {
      "epoch": 0.03194111605614516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2077,
      "step": 933
    },
    {
      "epoch": 0.03197535090722355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1779,
      "step": 934
    },
    {
      "epoch": 0.03200958575830195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.17,
      "step": 935
    },
    {
      "epoch": 0.03204382060938035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1654,
      "step": 936
    },
    {
      "epoch": 0.03207805546045875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.148,
      "step": 937
    },
    {
      "epoch": 0.032112290311537145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1508,
      "step": 938
    },
    {
      "epoch": 0.03214652516261554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1376,
      "step": 939
    },
    {
      "epoch": 0.03218076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1608,
      "step": 940
    },
    {
      "epoch": 0.03221499486477234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1381,
      "step": 941
    },
    {
      "epoch": 0.03224922971585074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1181,
      "step": 942
    },
    {
      "epoch": 0.03228346456692913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1387,
      "step": 943
    },
    {
      "epoch": 0.03231769941800753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1483,
      "step": 944
    },
    {
      "epoch": 0.032351934269085926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1717,
      "step": 945
    },
    {
      "epoch": 0.03238616912016433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1546,
      "step": 946
    },
    {
      "epoch": 0.032420403971242726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1852,
      "step": 947
    },
    {
      "epoch": 0.03245463882232112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1679,
      "step": 948
    },
    {
      "epoch": 0.03248887367339952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1965,
      "step": 949
    },
    {
      "epoch": 0.03252310852447792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1505,
      "step": 950
    },
    {
      "epoch": 0.03255734337555632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1759,
      "step": 951
    },
    {
      "epoch": 0.032591578226634714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1692,
      "step": 952
    },
    {
      "epoch": 0.03262581307771311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1465,
      "step": 953
    },
    {
      "epoch": 0.03266004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1297,
      "step": 954
    },
    {
      "epoch": 0.03269428277986991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1224,
      "step": 955
    },
    {
      "epoch": 0.032728517630948306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.176,
      "step": 956
    },
    {
      "epoch": 0.0327627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1175,
      "step": 957
    },
    {
      "epoch": 0.0327969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1554,
      "step": 958
    },
    {
      "epoch": 0.0328312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1385,
      "step": 959
    },
    {
      "epoch": 0.0328654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1495,
      "step": 960
    },
    {
      "epoch": 0.032899691886340295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.15,
      "step": 961
    },
    {
      "epoch": 0.03293392673741869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1369,
      "step": 962
    },
    {
      "epoch": 0.03296816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1347,
      "step": 963
    },
    {
      "epoch": 0.03300239643957549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1547,
      "step": 964
    },
    {
      "epoch": 0.03303663129065389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1253,
      "step": 965
    },
    {
      "epoch": 0.03307086614173228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1464,
      "step": 966
    },
    {
      "epoch": 0.03310510099281068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1473,
      "step": 967
    },
    {
      "epoch": 0.03313933584388908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1431,
      "step": 968
    },
    {
      "epoch": 0.03317357069496748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1647,
      "step": 969
    },
    {
      "epoch": 0.033207805546045875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1064,
      "step": 970
    },
    {
      "epoch": 0.03324204039712427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1342,
      "step": 971
    },
    {
      "epoch": 0.03327627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1013,
      "step": 972
    },
    {
      "epoch": 0.03331051009928107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1249,
      "step": 973
    },
    {
      "epoch": 0.03334474495035947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0985,
      "step": 974
    },
    {
      "epoch": 0.033378979801437864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1051,
      "step": 975
    },
    {
      "epoch": 0.03341321465251626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1233,
      "step": 976
    },
    {
      "epoch": 0.033447449503594656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.127,
      "step": 977
    },
    {
      "epoch": 0.03348168435467306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1194,
      "step": 978
    },
    {
      "epoch": 0.033515919205751456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.13,
      "step": 979
    },
    {
      "epoch": 0.03355015405682985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0862,
      "step": 980
    },
    {
      "epoch": 0.03358438890790825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1234,
      "step": 981
    },
    {
      "epoch": 0.03361862375898665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1361,
      "step": 982
    },
    {
      "epoch": 0.03365285861006505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.083,
      "step": 983
    },
    {
      "epoch": 0.033687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1552,
      "step": 984
    },
    {
      "epoch": 0.03372132831222184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1211,
      "step": 985
    },
    {
      "epoch": 0.03375556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1006,
      "step": 986
    },
    {
      "epoch": 0.03378979801437864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1024,
      "step": 987
    },
    {
      "epoch": 0.033824032865457036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1405,
      "step": 988
    },
    {
      "epoch": 0.03385826771653543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0894,
      "step": 989
    },
    {
      "epoch": 0.03389250256761383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0881,
      "step": 990
    },
    {
      "epoch": 0.03392673741869223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1109,
      "step": 991
    },
    {
      "epoch": 0.03396097226977063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0889,
      "step": 992
    },
    {
      "epoch": 0.033995207120849025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1098,
      "step": 993
    },
    {
      "epoch": 0.03402944197192742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1243,
      "step": 994
    },
    {
      "epoch": 0.03406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0993,
      "step": 995
    },
    {
      "epoch": 0.03409791167408422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.09,
      "step": 996
    },
    {
      "epoch": 0.03413214652516262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1064,
      "step": 997
    },
    {
      "epoch": 0.03416638137624101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0846,
      "step": 998
    },
    {
      "epoch": 0.03420061622731941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0658,
      "step": 999
    },
    {
      "epoch": 0.034234851078397806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0855,
      "step": 1000
    },
    {
      "epoch": 0.03426908592947621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1132,
      "step": 1001
    },
    {
      "epoch": 0.034303320780554605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0856,
      "step": 1002
    },
    {
      "epoch": 0.034337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0811,
      "step": 1003
    },
    {
      "epoch": 0.0343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0917,
      "step": 1004
    },
    {
      "epoch": 0.0344060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0888,
      "step": 1005
    },
    {
      "epoch": 0.0344402601848682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0919,
      "step": 1006
    },
    {
      "epoch": 0.034474495035946594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0913,
      "step": 1007
    },
    {
      "epoch": 0.03450872988702499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0742,
      "step": 1008
    },
    {
      "epoch": 0.034542964738103386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0674,
      "step": 1009
    },
    {
      "epoch": 0.03457719958918179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.06,
      "step": 1010
    },
    {
      "epoch": 0.034611434440260186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0685,
      "step": 1011
    },
    {
      "epoch": 0.03464566929133858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0549,
      "step": 1012
    },
    {
      "epoch": 0.03467990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0549,
      "step": 1013
    },
    {
      "epoch": 0.03471413899349538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0737,
      "step": 1014
    },
    {
      "epoch": 0.03474837384457378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0651,
      "step": 1015
    },
    {
      "epoch": 0.034782608695652174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0552,
      "step": 1016
    },
    {
      "epoch": 0.03481684354673057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0548,
      "step": 1017
    },
    {
      "epoch": 0.03485107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0998,
      "step": 1018
    },
    {
      "epoch": 0.03488531324888737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0501,
      "step": 1019
    },
    {
      "epoch": 0.034919548099965766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0692,
      "step": 1020
    },
    {
      "epoch": 0.03495378295104416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1087,
      "step": 1021
    },
    {
      "epoch": 0.03498801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.055,
      "step": 1022
    },
    {
      "epoch": 0.035022252653200955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0384,
      "step": 1023
    },
    {
      "epoch": 0.03505648750427936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0379,
      "step": 1024
    },
    {
      "epoch": 0.035090722355357755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0382,
      "step": 1025
    },
    {
      "epoch": 0.03512495720643615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.027,
      "step": 1026
    },
    {
      "epoch": 0.03515919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0263,
      "step": 1027
    },
    {
      "epoch": 0.03519342690859295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0754,
      "step": 1028
    },
    {
      "epoch": 0.03522766175967135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0447,
      "step": 1029
    },
    {
      "epoch": 0.03526189661074974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0239,
      "step": 1030
    },
    {
      "epoch": 0.03529613146182814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0297,
      "step": 1031
    },
    {
      "epoch": 0.035330366312906536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0358,
      "step": 1032
    },
    {
      "epoch": 0.03536460116398494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0099,
      "step": 1033
    },
    {
      "epoch": 0.035398836015063335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0472,
      "step": 1034
    },
    {
      "epoch": 0.03543307086614173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0415,
      "step": 1035
    },
    {
      "epoch": 0.03546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.044,
      "step": 1036
    },
    {
      "epoch": 0.03550154056829853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0336,
      "step": 1037
    },
    {
      "epoch": 0.03553577541937693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0411,
      "step": 1038
    },
    {
      "epoch": 0.035570010270455324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0128,
      "step": 1039
    },
    {
      "epoch": 0.03560424512153372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0413,
      "step": 1040
    },
    {
      "epoch": 0.03563847997261212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0427,
      "step": 1041
    },
    {
      "epoch": 0.03567271482369052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.049,
      "step": 1042
    },
    {
      "epoch": 0.035706949674768916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0278,
      "step": 1043
    },
    {
      "epoch": 0.03574118452584731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0132,
      "step": 1044
    },
    {
      "epoch": 0.03577541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.032,
      "step": 1045
    },
    {
      "epoch": 0.035809654228004105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0403,
      "step": 1046
    },
    {
      "epoch": 0.03584388907908251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0392,
      "step": 1047
    },
    {
      "epoch": 0.035878123930160905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0152,
      "step": 1048
    },
    {
      "epoch": 0.0359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0425,
      "step": 1049
    },
    {
      "epoch": 0.0359465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.032,
      "step": 1050
    },
    {
      "epoch": 0.0359808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0348,
      "step": 1051
    },
    {
      "epoch": 0.0360150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9793,
      "step": 1052
    },
    {
      "epoch": 0.03604929818555289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0016,
      "step": 1053
    },
    {
      "epoch": 0.03608353303663129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9715,
      "step": 1054
    },
    {
      "epoch": 0.036117767887709686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9978,
      "step": 1055
    },
    {
      "epoch": 0.03615200273878809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0017,
      "step": 1056
    },
    {
      "epoch": 0.036186237589866485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0222,
      "step": 1057
    },
    {
      "epoch": 0.03622047244094488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9989,
      "step": 1058
    },
    {
      "epoch": 0.03625470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0487,
      "step": 1059
    },
    {
      "epoch": 0.03628894214310168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0268,
      "step": 1060
    },
    {
      "epoch": 0.03632317699418008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0104,
      "step": 1061
    },
    {
      "epoch": 0.036357411845258474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0699,
      "step": 1062
    },
    {
      "epoch": 0.03639164669633687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0007,
      "step": 1063
    },
    {
      "epoch": 0.036425881547415266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0198,
      "step": 1064
    },
    {
      "epoch": 0.03646011639849367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0046,
      "step": 1065
    },
    {
      "epoch": 0.036494351249572066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0113,
      "step": 1066
    },
    {
      "epoch": 0.03652858610065046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0051,
      "step": 1067
    },
    {
      "epoch": 0.03656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0148,
      "step": 1068
    },
    {
      "epoch": 0.036597055802807255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0477,
      "step": 1069
    },
    {
      "epoch": 0.03663129065388566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9782,
      "step": 1070
    },
    {
      "epoch": 0.036665525504964054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9823,
      "step": 1071
    },
    {
      "epoch": 0.03669976035604245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9798,
      "step": 1072
    },
    {
      "epoch": 0.03673399520712085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9958,
      "step": 1073
    },
    {
      "epoch": 0.03676823005819925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0154,
      "step": 1074
    },
    {
      "epoch": 0.036802464909277646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9906,
      "step": 1075
    },
    {
      "epoch": 0.03683669976035604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9843,
      "step": 1076
    },
    {
      "epoch": 0.03687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0285,
      "step": 1077
    },
    {
      "epoch": 0.036905169462512835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9707,
      "step": 1078
    },
    {
      "epoch": 0.03693940431359124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9661,
      "step": 1079
    },
    {
      "epoch": 0.036973639164669635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0077,
      "step": 1080
    },
    {
      "epoch": 0.03700787401574803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.98,
      "step": 1081
    },
    {
      "epoch": 0.03704210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9684,
      "step": 1082
    },
    {
      "epoch": 0.03707634371790483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9539,
      "step": 1083
    },
    {
      "epoch": 0.03711057856898323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9609,
      "step": 1084
    },
    {
      "epoch": 0.03714481342006162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9913,
      "step": 1085
    },
    {
      "epoch": 0.03717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0206,
      "step": 1086
    },
    {
      "epoch": 0.037213283122218416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9957,
      "step": 1087
    },
    {
      "epoch": 0.03724751797329682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9888,
      "step": 1088
    },
    {
      "epoch": 0.037281752824375215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9912,
      "step": 1089
    },
    {
      "epoch": 0.03731598767545361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9549,
      "step": 1090
    },
    {
      "epoch": 0.03735022252653201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9813,
      "step": 1091
    },
    {
      "epoch": 0.037384457377610404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0016,
      "step": 1092
    },
    {
      "epoch": 0.03741869222868881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9758,
      "step": 1093
    },
    {
      "epoch": 0.037452927079767204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.989,
      "step": 1094
    },
    {
      "epoch": 0.0374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9876,
      "step": 1095
    },
    {
      "epoch": 0.037521396781923996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9724,
      "step": 1096
    },
    {
      "epoch": 0.0375556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9841,
      "step": 1097
    },
    {
      "epoch": 0.037589866484080796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9614,
      "step": 1098
    },
    {
      "epoch": 0.03762410133515919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9343,
      "step": 1099
    },
    {
      "epoch": 0.03765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.993,
      "step": 1100
    },
    {
      "epoch": 0.037692571037315985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9602,
      "step": 1101
    },
    {
      "epoch": 0.03772680588839439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9738,
      "step": 1102
    },
    {
      "epoch": 0.037761040739472784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9348,
      "step": 1103
    },
    {
      "epoch": 0.03779527559055118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9473,
      "step": 1104
    },
    {
      "epoch": 0.03782951044162958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9609,
      "step": 1105
    },
    {
      "epoch": 0.03786374529270798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9968,
      "step": 1106
    },
    {
      "epoch": 0.037897980143786376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9886,
      "step": 1107
    },
    {
      "epoch": 0.03793221499486477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.94,
      "step": 1108
    },
    {
      "epoch": 0.03796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.966,
      "step": 1109
    },
    {
      "epoch": 0.038000684697021565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9821,
      "step": 1110
    },
    {
      "epoch": 0.03803491954809997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9522,
      "step": 1111
    },
    {
      "epoch": 0.038069154399178365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9378,
      "step": 1112
    },
    {
      "epoch": 0.03810338925025676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9476,
      "step": 1113
    },
    {
      "epoch": 0.03813762410133516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9395,
      "step": 1114
    },
    {
      "epoch": 0.038171858952413554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9446,
      "step": 1115
    },
    {
      "epoch": 0.03820609380349196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9333,
      "step": 1116
    },
    {
      "epoch": 0.03824032865457035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.954,
      "step": 1117
    },
    {
      "epoch": 0.03827456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.92,
      "step": 1118
    },
    {
      "epoch": 0.038308798356727146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.93,
      "step": 1119
    },
    {
      "epoch": 0.03834303320780555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.968,
      "step": 1120
    },
    {
      "epoch": 0.038377268058883945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9638,
      "step": 1121
    },
    {
      "epoch": 0.03841150290996234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.949,
      "step": 1122
    },
    {
      "epoch": 0.03844573776104074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9529,
      "step": 1123
    },
    {
      "epoch": 0.038479972612119134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9454,
      "step": 1124
    },
    {
      "epoch": 0.03851420746319754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.958,
      "step": 1125
    },
    {
      "epoch": 0.038548442314275934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9326,
      "step": 1126
    },
    {
      "epoch": 0.03858267716535433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9325,
      "step": 1127
    },
    {
      "epoch": 0.038616912016432726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9585,
      "step": 1128
    },
    {
      "epoch": 0.03865114686751113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.932,
      "step": 1129
    },
    {
      "epoch": 0.038685381718589526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9359,
      "step": 1130
    },
    {
      "epoch": 0.03871961656966792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8981,
      "step": 1131
    },
    {
      "epoch": 0.03875385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9472,
      "step": 1132
    },
    {
      "epoch": 0.038788086271824715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9552,
      "step": 1133
    },
    {
      "epoch": 0.03882232112290312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9181,
      "step": 1134
    },
    {
      "epoch": 0.038856555973981514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9294,
      "step": 1135
    },
    {
      "epoch": 0.03889079082505991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.91,
      "step": 1136
    },
    {
      "epoch": 0.03892502567613831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.916,
      "step": 1137
    },
    {
      "epoch": 0.0389592605272167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9187,
      "step": 1138
    },
    {
      "epoch": 0.03899349537829511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9367,
      "step": 1139
    },
    {
      "epoch": 0.0390277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9023,
      "step": 1140
    },
    {
      "epoch": 0.0390619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9427,
      "step": 1141
    },
    {
      "epoch": 0.039096199931530295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9148,
      "step": 1142
    },
    {
      "epoch": 0.0391304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9219,
      "step": 1143
    },
    {
      "epoch": 0.039164669633687095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9338,
      "step": 1144
    },
    {
      "epoch": 0.03919890448476549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9198,
      "step": 1145
    },
    {
      "epoch": 0.03923313933584389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9244,
      "step": 1146
    },
    {
      "epoch": 0.039267374186922284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9402,
      "step": 1147
    },
    {
      "epoch": 0.03930160903800069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9268,
      "step": 1148
    },
    {
      "epoch": 0.03933584388907908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9111,
      "step": 1149
    },
    {
      "epoch": 0.03937007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.92,
      "step": 1150
    },
    {
      "epoch": 0.039404313591235876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9034,
      "step": 1151
    },
    {
      "epoch": 0.03943854844231428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9045,
      "step": 1152
    },
    {
      "epoch": 0.039472783293392676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8956,
      "step": 1153
    },
    {
      "epoch": 0.03950701814447107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.92,
      "step": 1154
    },
    {
      "epoch": 0.03954125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8892,
      "step": 1155
    },
    {
      "epoch": 0.039575487846627865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9146,
      "step": 1156
    },
    {
      "epoch": 0.03960972269770627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9111,
      "step": 1157
    },
    {
      "epoch": 0.039643957548784664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8897,
      "step": 1158
    },
    {
      "epoch": 0.03967819239986306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9118,
      "step": 1159
    },
    {
      "epoch": 0.03971242725094146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8842,
      "step": 1160
    },
    {
      "epoch": 0.03974666210201985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9103,
      "step": 1161
    },
    {
      "epoch": 0.039780896953098256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8945,
      "step": 1162
    },
    {
      "epoch": 0.03981513180417665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8896,
      "step": 1163
    },
    {
      "epoch": 0.03984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8958,
      "step": 1164
    },
    {
      "epoch": 0.039883601506333445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9068,
      "step": 1165
    },
    {
      "epoch": 0.03991783635741185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9053,
      "step": 1166
    },
    {
      "epoch": 0.039952071208490245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8676,
      "step": 1167
    },
    {
      "epoch": 0.03998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8674,
      "step": 1168
    },
    {
      "epoch": 0.04002054091064704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9126,
      "step": 1169
    },
    {
      "epoch": 0.040054775761725434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9128,
      "step": 1170
    },
    {
      "epoch": 0.04008901061280384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8743,
      "step": 1171
    },
    {
      "epoch": 0.04012324546388223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9098,
      "step": 1172
    },
    {
      "epoch": 0.04015748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9056,
      "step": 1173
    },
    {
      "epoch": 0.040191715166039026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8954,
      "step": 1174
    },
    {
      "epoch": 0.04022595001711743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9297,
      "step": 1175
    },
    {
      "epoch": 0.040260184868195825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8736,
      "step": 1176
    },
    {
      "epoch": 0.04029441971927422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9145,
      "step": 1177
    },
    {
      "epoch": 0.04032865457035262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8923,
      "step": 1178
    },
    {
      "epoch": 0.040362889421431014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8861,
      "step": 1179
    },
    {
      "epoch": 0.04039712427250942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8968,
      "step": 1180
    },
    {
      "epoch": 0.040431359123587814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8627,
      "step": 1181
    },
    {
      "epoch": 0.04046559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8787,
      "step": 1182
    },
    {
      "epoch": 0.040499828825744606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9154,
      "step": 1183
    },
    {
      "epoch": 0.040534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8966,
      "step": 1184
    },
    {
      "epoch": 0.040568298527901406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9173,
      "step": 1185
    },
    {
      "epoch": 0.0406025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.871,
      "step": 1186
    },
    {
      "epoch": 0.0406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8931,
      "step": 1187
    },
    {
      "epoch": 0.040671003081136595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8842,
      "step": 1188
    },
    {
      "epoch": 0.040705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8638,
      "step": 1189
    },
    {
      "epoch": 0.040739472783293394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8516,
      "step": 1190
    },
    {
      "epoch": 0.04077370763437179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8583,
      "step": 1191
    },
    {
      "epoch": 0.04080794248545019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8842,
      "step": 1192
    },
    {
      "epoch": 0.04084217733652858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8966,
      "step": 1193
    },
    {
      "epoch": 0.040876412187606986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8727,
      "step": 1194
    },
    {
      "epoch": 0.04091064703868538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8518,
      "step": 1195
    },
    {
      "epoch": 0.04094488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8601,
      "step": 1196
    },
    {
      "epoch": 0.040979116740842175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8697,
      "step": 1197
    },
    {
      "epoch": 0.04101335159192058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8764,
      "step": 1198
    },
    {
      "epoch": 0.041047586442998975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8608,
      "step": 1199
    },
    {
      "epoch": 0.04108182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8657,
      "step": 1200
    },
    {
      "epoch": 0.04111605614515577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8641,
      "step": 1201
    },
    {
      "epoch": 0.041150290996234164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.88,
      "step": 1202
    },
    {
      "epoch": 0.04118452584731257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8883,
      "step": 1203
    },
    {
      "epoch": 0.04121876069839096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8437,
      "step": 1204
    },
    {
      "epoch": 0.04125299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8428,
      "step": 1205
    },
    {
      "epoch": 0.041287230400547756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8513,
      "step": 1206
    },
    {
      "epoch": 0.04132146525162615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8736,
      "step": 1207
    },
    {
      "epoch": 0.041355700102704555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8695,
      "step": 1208
    },
    {
      "epoch": 0.04138993495378295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8464,
      "step": 1209
    },
    {
      "epoch": 0.04142416980486135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8775,
      "step": 1210
    },
    {
      "epoch": 0.041458404655939744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8719,
      "step": 1211
    },
    {
      "epoch": 0.04149263950701815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8737,
      "step": 1212
    },
    {
      "epoch": 0.041526874358096544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8565,
      "step": 1213
    },
    {
      "epoch": 0.04156110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8708,
      "step": 1214
    },
    {
      "epoch": 0.041595344060253336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8404,
      "step": 1215
    },
    {
      "epoch": 0.04162957891133173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8662,
      "step": 1216
    },
    {
      "epoch": 0.041663813762410136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8335,
      "step": 1217
    },
    {
      "epoch": 0.04169804861348853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8222,
      "step": 1218
    },
    {
      "epoch": 0.04173228346456693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8371,
      "step": 1219
    },
    {
      "epoch": 0.041766518315645325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8511,
      "step": 1220
    },
    {
      "epoch": 0.04180075316672373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8558,
      "step": 1221
    },
    {
      "epoch": 0.041834988017802124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8391,
      "step": 1222
    },
    {
      "epoch": 0.04186922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8779,
      "step": 1223
    },
    {
      "epoch": 0.04190345771995892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.831,
      "step": 1224
    },
    {
      "epoch": 0.04193769257103731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8354,
      "step": 1225
    },
    {
      "epoch": 0.04197192742211572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8495,
      "step": 1226
    },
    {
      "epoch": 0.04200616227319411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8365,
      "step": 1227
    },
    {
      "epoch": 0.04204039712427251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8284,
      "step": 1228
    },
    {
      "epoch": 0.042074631975350905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8327,
      "step": 1229
    },
    {
      "epoch": 0.0421088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8526,
      "step": 1230
    },
    {
      "epoch": 0.042143101677507705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7981,
      "step": 1231
    },
    {
      "epoch": 0.0421773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8127,
      "step": 1232
    },
    {
      "epoch": 0.0422115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8935,
      "step": 1233
    },
    {
      "epoch": 0.042245806230742894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8124,
      "step": 1234
    },
    {
      "epoch": 0.0422800410818213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8548,
      "step": 1235
    },
    {
      "epoch": 0.04231427593289969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8108,
      "step": 1236
    },
    {
      "epoch": 0.04234851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8163,
      "step": 1237
    },
    {
      "epoch": 0.042382745635056486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8479,
      "step": 1238
    },
    {
      "epoch": 0.04241698048613488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.841,
      "step": 1239
    },
    {
      "epoch": 0.042451215337213286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8633,
      "step": 1240
    },
    {
      "epoch": 0.04248545018829168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8774,
      "step": 1241
    },
    {
      "epoch": 0.04251968503937008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8181,
      "step": 1242
    },
    {
      "epoch": 0.042553919890448474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8537,
      "step": 1243
    },
    {
      "epoch": 0.04258815474152688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8164,
      "step": 1244
    },
    {
      "epoch": 0.042622389592605274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8544,
      "step": 1245
    },
    {
      "epoch": 0.04265662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8428,
      "step": 1246
    },
    {
      "epoch": 0.04269085929476207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8057,
      "step": 1247
    },
    {
      "epoch": 0.04272509414584046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8376,
      "step": 1248
    },
    {
      "epoch": 0.042759328996918866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8134,
      "step": 1249
    },
    {
      "epoch": 0.04279356384799726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.862,
      "step": 1250
    },
    {
      "epoch": 0.04282779869907566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8271,
      "step": 1251
    },
    {
      "epoch": 0.042862033550154055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8082,
      "step": 1252
    },
    {
      "epoch": 0.04289626840123245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.804,
      "step": 1253
    },
    {
      "epoch": 0.042930503252310855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.809,
      "step": 1254
    },
    {
      "epoch": 0.04296473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8103,
      "step": 1255
    },
    {
      "epoch": 0.04299897295446765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8193,
      "step": 1256
    },
    {
      "epoch": 0.04303320780554604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8121,
      "step": 1257
    },
    {
      "epoch": 0.04306744265662445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8157,
      "step": 1258
    },
    {
      "epoch": 0.04310167750770284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8085,
      "step": 1259
    },
    {
      "epoch": 0.04313591235878124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8279,
      "step": 1260
    },
    {
      "epoch": 0.043170147209859636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8243,
      "step": 1261
    },
    {
      "epoch": 0.04320438206093803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.827,
      "step": 1262
    },
    {
      "epoch": 0.043238616912016435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8126,
      "step": 1263
    },
    {
      "epoch": 0.04327285176309483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8318,
      "step": 1264
    },
    {
      "epoch": 0.04330708661417323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8352,
      "step": 1265
    },
    {
      "epoch": 0.043341321465251624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.832,
      "step": 1266
    },
    {
      "epoch": 0.04337555631633003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8124,
      "step": 1267
    },
    {
      "epoch": 0.043409791167408424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8403,
      "step": 1268
    },
    {
      "epoch": 0.04344402601848682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.798,
      "step": 1269
    },
    {
      "epoch": 0.043478260869565216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7905,
      "step": 1270
    },
    {
      "epoch": 0.04351249572064361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8262,
      "step": 1271
    },
    {
      "epoch": 0.043546730571722016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8025,
      "step": 1272
    },
    {
      "epoch": 0.04358096542280041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8071,
      "step": 1273
    },
    {
      "epoch": 0.04361520027387881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8271,
      "step": 1274
    },
    {
      "epoch": 0.043649435124957205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8171,
      "step": 1275
    },
    {
      "epoch": 0.0436836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8131,
      "step": 1276
    },
    {
      "epoch": 0.043717904827114004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.815,
      "step": 1277
    },
    {
      "epoch": 0.0437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8297,
      "step": 1278
    },
    {
      "epoch": 0.0437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7693,
      "step": 1279
    },
    {
      "epoch": 0.04382060938034919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7796,
      "step": 1280
    },
    {
      "epoch": 0.043854844231427596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.802,
      "step": 1281
    },
    {
      "epoch": 0.04388907908250599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.789,
      "step": 1282
    },
    {
      "epoch": 0.04392331393358439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7765,
      "step": 1283
    },
    {
      "epoch": 0.043957548784662785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7835,
      "step": 1284
    },
    {
      "epoch": 0.04399178363574118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8277,
      "step": 1285
    },
    {
      "epoch": 0.044026018486819585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.787,
      "step": 1286
    },
    {
      "epoch": 0.04406025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7896,
      "step": 1287
    },
    {
      "epoch": 0.04409448818897638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7739,
      "step": 1288
    },
    {
      "epoch": 0.044128723040054774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8009,
      "step": 1289
    },
    {
      "epoch": 0.04416295789113318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7778,
      "step": 1290
    },
    {
      "epoch": 0.04419719274221157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7811,
      "step": 1291
    },
    {
      "epoch": 0.04423142759328997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7955,
      "step": 1292
    },
    {
      "epoch": 0.044265662444368366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7944,
      "step": 1293
    },
    {
      "epoch": 0.04429989729544676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7967,
      "step": 1294
    },
    {
      "epoch": 0.044334132146525165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8139,
      "step": 1295
    },
    {
      "epoch": 0.04436836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7756,
      "step": 1296
    },
    {
      "epoch": 0.04440260184868196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7697,
      "step": 1297
    },
    {
      "epoch": 0.044436836699760354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.776,
      "step": 1298
    },
    {
      "epoch": 0.04447107155083875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7723,
      "step": 1299
    },
    {
      "epoch": 0.044505306401917154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.824,
      "step": 1300
    },
    {
      "epoch": 0.04453954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8007,
      "step": 1301
    },
    {
      "epoch": 0.044573776104073946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7999,
      "step": 1302
    },
    {
      "epoch": 0.04460801095515234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7927,
      "step": 1303
    },
    {
      "epoch": 0.044642245806230746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7809,
      "step": 1304
    },
    {
      "epoch": 0.04467648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7778,
      "step": 1305
    },
    {
      "epoch": 0.04471071550838754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7518,
      "step": 1306
    },
    {
      "epoch": 0.044744950359465935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7775,
      "step": 1307
    },
    {
      "epoch": 0.04477918521054433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7668,
      "step": 1308
    },
    {
      "epoch": 0.044813420061622734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7372,
      "step": 1309
    },
    {
      "epoch": 0.04484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7746,
      "step": 1310
    },
    {
      "epoch": 0.04488188976377953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7784,
      "step": 1311
    },
    {
      "epoch": 0.04491612461485792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7689,
      "step": 1312
    },
    {
      "epoch": 0.044950359465936326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7583,
      "step": 1313
    },
    {
      "epoch": 0.04498459431701472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7587,
      "step": 1314
    },
    {
      "epoch": 0.04501882916809312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7731,
      "step": 1315
    },
    {
      "epoch": 0.045053064019171515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7634,
      "step": 1316
    },
    {
      "epoch": 0.04508729887024991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7819,
      "step": 1317
    },
    {
      "epoch": 0.045121533721328315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7784,
      "step": 1318
    },
    {
      "epoch": 0.04515576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7674,
      "step": 1319
    },
    {
      "epoch": 0.04519000342348511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7865,
      "step": 1320
    },
    {
      "epoch": 0.045224238274563504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7968,
      "step": 1321
    },
    {
      "epoch": 0.0452584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7862,
      "step": 1322
    },
    {
      "epoch": 0.0452927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7713,
      "step": 1323
    },
    {
      "epoch": 0.0453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.764,
      "step": 1324
    },
    {
      "epoch": 0.045361177678877096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7878,
      "step": 1325
    },
    {
      "epoch": 0.04539541252995549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7758,
      "step": 1326
    },
    {
      "epoch": 0.045429647381033895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7327,
      "step": 1327
    },
    {
      "epoch": 0.04546388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7515,
      "step": 1328
    },
    {
      "epoch": 0.04549811708319069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7627,
      "step": 1329
    },
    {
      "epoch": 0.045532351934269084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7549,
      "step": 1330
    },
    {
      "epoch": 0.04556658678534748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7647,
      "step": 1331
    },
    {
      "epoch": 0.045600821636425884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7886,
      "step": 1332
    },
    {
      "epoch": 0.04563505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.784,
      "step": 1333
    },
    {
      "epoch": 0.04566929133858268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7613,
      "step": 1334
    },
    {
      "epoch": 0.04570352618966107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7434,
      "step": 1335
    },
    {
      "epoch": 0.045737761040739476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7467,
      "step": 1336
    },
    {
      "epoch": 0.04577199589181787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7281,
      "step": 1337
    },
    {
      "epoch": 0.04580623074289627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.735,
      "step": 1338
    },
    {
      "epoch": 0.045840465593974665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7515,
      "step": 1339
    },
    {
      "epoch": 0.04587470044505306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7494,
      "step": 1340
    },
    {
      "epoch": 0.045908935296131465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7652,
      "step": 1341
    },
    {
      "epoch": 0.04594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7387,
      "step": 1342
    },
    {
      "epoch": 0.04597740499828826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7477,
      "step": 1343
    },
    {
      "epoch": 0.04601163984936665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7507,
      "step": 1344
    },
    {
      "epoch": 0.04604587470044505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7483,
      "step": 1345
    },
    {
      "epoch": 0.04608010955152345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7693,
      "step": 1346
    },
    {
      "epoch": 0.04611434440260185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7822,
      "step": 1347
    },
    {
      "epoch": 0.046148579253680246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7986,
      "step": 1348
    },
    {
      "epoch": 0.04618281410475864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7435,
      "step": 1349
    },
    {
      "epoch": 0.046217048955837045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7411,
      "step": 1350
    },
    {
      "epoch": 0.04625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7344,
      "step": 1351
    },
    {
      "epoch": 0.04628551865799384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7358,
      "step": 1352
    },
    {
      "epoch": 0.046319753509072234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7417,
      "step": 1353
    },
    {
      "epoch": 0.04635398836015063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7494,
      "step": 1354
    },
    {
      "epoch": 0.046388223211229034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7379,
      "step": 1355
    },
    {
      "epoch": 0.04642245806230743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7613,
      "step": 1356
    },
    {
      "epoch": 0.046456692913385826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6948,
      "step": 1357
    },
    {
      "epoch": 0.04649092776446422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7349,
      "step": 1358
    },
    {
      "epoch": 0.046525162615542626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7428,
      "step": 1359
    },
    {
      "epoch": 0.04655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7113,
      "step": 1360
    },
    {
      "epoch": 0.04659363231769942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7222,
      "step": 1361
    },
    {
      "epoch": 0.046627867168777815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.749,
      "step": 1362
    },
    {
      "epoch": 0.04666210201985621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7501,
      "step": 1363
    },
    {
      "epoch": 0.046696336870934614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7434,
      "step": 1364
    },
    {
      "epoch": 0.04673057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7342,
      "step": 1365
    },
    {
      "epoch": 0.04676480657309141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7314,
      "step": 1366
    },
    {
      "epoch": 0.0467990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.707,
      "step": 1367
    },
    {
      "epoch": 0.0468332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7563,
      "step": 1368
    },
    {
      "epoch": 0.0468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7167,
      "step": 1369
    },
    {
      "epoch": 0.046901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7489,
      "step": 1370
    },
    {
      "epoch": 0.046935980828483395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7311,
      "step": 1371
    },
    {
      "epoch": 0.04697021567956179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7478,
      "step": 1372
    },
    {
      "epoch": 0.047004450530640195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7185,
      "step": 1373
    },
    {
      "epoch": 0.04703868538171859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7282,
      "step": 1374
    },
    {
      "epoch": 0.04707292023279699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7301,
      "step": 1375
    },
    {
      "epoch": 0.047107155083875384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7347,
      "step": 1376
    },
    {
      "epoch": 0.04714138993495378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7285,
      "step": 1377
    },
    {
      "epoch": 0.04717562478603218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7094,
      "step": 1378
    },
    {
      "epoch": 0.04720985963711058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7527,
      "step": 1379
    },
    {
      "epoch": 0.047244094488188976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7187,
      "step": 1380
    },
    {
      "epoch": 0.04727832933926737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7285,
      "step": 1381
    },
    {
      "epoch": 0.047312564190345775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7341,
      "step": 1382
    },
    {
      "epoch": 0.04734679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7077,
      "step": 1383
    },
    {
      "epoch": 0.04738103389250257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7131,
      "step": 1384
    },
    {
      "epoch": 0.047415268743580964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.718,
      "step": 1385
    },
    {
      "epoch": 0.04744950359465936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7075,
      "step": 1386
    },
    {
      "epoch": 0.047483738445737764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7563,
      "step": 1387
    },
    {
      "epoch": 0.04751797329681616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7287,
      "step": 1388
    },
    {
      "epoch": 0.047552208147894556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6916,
      "step": 1389
    },
    {
      "epoch": 0.04758644299897295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.717,
      "step": 1390
    },
    {
      "epoch": 0.04762067785005135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7241,
      "step": 1391
    },
    {
      "epoch": 0.04765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7061,
      "step": 1392
    },
    {
      "epoch": 0.04768914755220815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7063,
      "step": 1393
    },
    {
      "epoch": 0.047723382403286545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6785,
      "step": 1394
    },
    {
      "epoch": 0.04775761725436494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7099,
      "step": 1395
    },
    {
      "epoch": 0.047791852105443344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7545,
      "step": 1396
    },
    {
      "epoch": 0.04782608695652174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7419,
      "step": 1397
    },
    {
      "epoch": 0.04786032180760014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7205,
      "step": 1398
    },
    {
      "epoch": 0.04789455665867853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7296,
      "step": 1399
    },
    {
      "epoch": 0.04792879150975693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7426,
      "step": 1400
    },
    {
      "epoch": 0.04796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6999,
      "step": 1401
    },
    {
      "epoch": 0.04799726121191373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7259,
      "step": 1402
    },
    {
      "epoch": 0.048031496062992125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6999,
      "step": 1403
    },
    {
      "epoch": 0.04806573091407052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7066,
      "step": 1404
    },
    {
      "epoch": 0.048099965765148925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.722,
      "step": 1405
    },
    {
      "epoch": 0.04813420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7001,
      "step": 1406
    },
    {
      "epoch": 0.04816843546730572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7134,
      "step": 1407
    },
    {
      "epoch": 0.048202670318384114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7079,
      "step": 1408
    },
    {
      "epoch": 0.04823690516946251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6673,
      "step": 1409
    },
    {
      "epoch": 0.04827114002054091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7026,
      "step": 1410
    },
    {
      "epoch": 0.04830537487161931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6902,
      "step": 1411
    },
    {
      "epoch": 0.048339609722697706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6846,
      "step": 1412
    },
    {
      "epoch": 0.0483738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6791,
      "step": 1413
    },
    {
      "epoch": 0.0484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6761,
      "step": 1414
    },
    {
      "epoch": 0.0484423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6989,
      "step": 1415
    },
    {
      "epoch": 0.0484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7195,
      "step": 1416
    },
    {
      "epoch": 0.048510783978089694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6764,
      "step": 1417
    },
    {
      "epoch": 0.04854501882916809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6856,
      "step": 1418
    },
    {
      "epoch": 0.048579253680246494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6807,
      "step": 1419
    },
    {
      "epoch": 0.04861348853132489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6792,
      "step": 1420
    },
    {
      "epoch": 0.048647723382403286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6761,
      "step": 1421
    },
    {
      "epoch": 0.04868195823348168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6821,
      "step": 1422
    },
    {
      "epoch": 0.04871619308456008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6794,
      "step": 1423
    },
    {
      "epoch": 0.04875042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.688,
      "step": 1424
    },
    {
      "epoch": 0.04878466278671688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7027,
      "step": 1425
    },
    {
      "epoch": 0.048818897637795275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6984,
      "step": 1426
    },
    {
      "epoch": 0.04885313248887367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7272,
      "step": 1427
    },
    {
      "epoch": 0.048887367339952074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7078,
      "step": 1428
    },
    {
      "epoch": 0.04892160219103047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.71,
      "step": 1429
    },
    {
      "epoch": 0.04895583704210887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6867,
      "step": 1430
    },
    {
      "epoch": 0.04899007189318726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6889,
      "step": 1431
    },
    {
      "epoch": 0.04902430674426566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6883,
      "step": 1432
    },
    {
      "epoch": 0.04905854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.717,
      "step": 1433
    },
    {
      "epoch": 0.04909277644642246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7013,
      "step": 1434
    },
    {
      "epoch": 0.049127011297500855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6592,
      "step": 1435
    },
    {
      "epoch": 0.04916124614857925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6601,
      "step": 1436
    },
    {
      "epoch": 0.04919548099965765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6435,
      "step": 1437
    },
    {
      "epoch": 0.04922971585073605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6606,
      "step": 1438
    },
    {
      "epoch": 0.04926395070181445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7022,
      "step": 1439
    },
    {
      "epoch": 0.049298185552892844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6834,
      "step": 1440
    },
    {
      "epoch": 0.04933242040397124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.688,
      "step": 1441
    },
    {
      "epoch": 0.04936665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6526,
      "step": 1442
    },
    {
      "epoch": 0.04940089010612804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6884,
      "step": 1443
    },
    {
      "epoch": 0.049435124957206436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6862,
      "step": 1444
    },
    {
      "epoch": 0.04946935980828483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6811,
      "step": 1445
    },
    {
      "epoch": 0.04950359465936323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6765,
      "step": 1446
    },
    {
      "epoch": 0.04953782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6784,
      "step": 1447
    },
    {
      "epoch": 0.04957206436152003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6913,
      "step": 1448
    },
    {
      "epoch": 0.049606299212598425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6839,
      "step": 1449
    },
    {
      "epoch": 0.04964053406367682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7217,
      "step": 1450
    },
    {
      "epoch": 0.049674768914755224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6692,
      "step": 1451
    },
    {
      "epoch": 0.04970900376583362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6564,
      "step": 1452
    },
    {
      "epoch": 0.04974323861691202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6795,
      "step": 1453
    },
    {
      "epoch": 0.04977747346799041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6756,
      "step": 1454
    },
    {
      "epoch": 0.04981170831906881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6681,
      "step": 1455
    },
    {
      "epoch": 0.04984594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6807,
      "step": 1456
    },
    {
      "epoch": 0.04988017802122561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6467,
      "step": 1457
    },
    {
      "epoch": 0.049914412872304005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6399,
      "step": 1458
    },
    {
      "epoch": 0.0499486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6625,
      "step": 1459
    },
    {
      "epoch": 0.0499828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.662,
      "step": 1460
    },
    {
      "epoch": 0.0500171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6554,
      "step": 1461
    },
    {
      "epoch": 0.0500513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6668,
      "step": 1462
    },
    {
      "epoch": 0.050085587127695994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6775,
      "step": 1463
    },
    {
      "epoch": 0.05011982197877439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6924,
      "step": 1464
    },
    {
      "epoch": 0.05015405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6681,
      "step": 1465
    },
    {
      "epoch": 0.05018829168093119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6431,
      "step": 1466
    },
    {
      "epoch": 0.050222526532009586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6478,
      "step": 1467
    },
    {
      "epoch": 0.05025676138308798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6534,
      "step": 1468
    },
    {
      "epoch": 0.05029099623416638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6787,
      "step": 1469
    },
    {
      "epoch": 0.05032523108524478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6494,
      "step": 1470
    },
    {
      "epoch": 0.05035946593632318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6426,
      "step": 1471
    },
    {
      "epoch": 0.050393700787401574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6331,
      "step": 1472
    },
    {
      "epoch": 0.05042793563847997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.642,
      "step": 1473
    },
    {
      "epoch": 0.050462170489558374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6727,
      "step": 1474
    },
    {
      "epoch": 0.05049640534063677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6904,
      "step": 1475
    },
    {
      "epoch": 0.050530640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6533,
      "step": 1476
    },
    {
      "epoch": 0.05056487504279356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6147,
      "step": 1477
    },
    {
      "epoch": 0.05059910989387196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6366,
      "step": 1478
    },
    {
      "epoch": 0.05063334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6676,
      "step": 1479
    },
    {
      "epoch": 0.05066757959602876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6494,
      "step": 1480
    },
    {
      "epoch": 0.050701814447107155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6495,
      "step": 1481
    },
    {
      "epoch": 0.05073604929818555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6495,
      "step": 1482
    },
    {
      "epoch": 0.05077028414926395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6394,
      "step": 1483
    },
    {
      "epoch": 0.05080451900034235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6307,
      "step": 1484
    },
    {
      "epoch": 0.05083875385142075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6468,
      "step": 1485
    },
    {
      "epoch": 0.05087298870249914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6437,
      "step": 1486
    },
    {
      "epoch": 0.05090722355357754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6488,
      "step": 1487
    },
    {
      "epoch": 0.05094145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6388,
      "step": 1488
    },
    {
      "epoch": 0.05097569325573434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6657,
      "step": 1489
    },
    {
      "epoch": 0.051009928106812735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6476,
      "step": 1490
    },
    {
      "epoch": 0.05104416295789113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6213,
      "step": 1491
    },
    {
      "epoch": 0.05107839780896953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6405,
      "step": 1492
    },
    {
      "epoch": 0.05111263266004793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6593,
      "step": 1493
    },
    {
      "epoch": 0.05114686751112633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6344,
      "step": 1494
    },
    {
      "epoch": 0.051181102362204724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6341,
      "step": 1495
    },
    {
      "epoch": 0.05121533721328312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6273,
      "step": 1496
    },
    {
      "epoch": 0.05124957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6324,
      "step": 1497
    },
    {
      "epoch": 0.05128380691543992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6463,
      "step": 1498
    },
    {
      "epoch": 0.051318041766518316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6462,
      "step": 1499
    },
    {
      "epoch": 0.05135227661759671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6417,
      "step": 1500
    },
    {
      "epoch": 0.05138651146867511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6357,
      "step": 1501
    },
    {
      "epoch": 0.05142074631975351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6708,
      "step": 1502
    },
    {
      "epoch": 0.05145498117083191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6573,
      "step": 1503
    },
    {
      "epoch": 0.051489216021910304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6364,
      "step": 1504
    },
    {
      "epoch": 0.0515234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6418,
      "step": 1505
    },
    {
      "epoch": 0.0515576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6509,
      "step": 1506
    },
    {
      "epoch": 0.0515919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6633,
      "step": 1507
    },
    {
      "epoch": 0.051626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6711,
      "step": 1508
    },
    {
      "epoch": 0.05166039027730229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6333,
      "step": 1509
    },
    {
      "epoch": 0.05169462512838069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6319,
      "step": 1510
    },
    {
      "epoch": 0.05172885997945909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6616,
      "step": 1511
    },
    {
      "epoch": 0.05176309483053749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6457,
      "step": 1512
    },
    {
      "epoch": 0.051797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6441,
      "step": 1513
    },
    {
      "epoch": 0.05183156453269428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6043,
      "step": 1514
    },
    {
      "epoch": 0.05186579938377268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6411,
      "step": 1515
    },
    {
      "epoch": 0.05190003423485108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6435,
      "step": 1516
    },
    {
      "epoch": 0.05193426908592948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6573,
      "step": 1517
    },
    {
      "epoch": 0.05196850393700787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6256,
      "step": 1518
    },
    {
      "epoch": 0.05200273878808627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6207,
      "step": 1519
    },
    {
      "epoch": 0.05203697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6164,
      "step": 1520
    },
    {
      "epoch": 0.05207120849024307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6209,
      "step": 1521
    },
    {
      "epoch": 0.052105443341321465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6347,
      "step": 1522
    },
    {
      "epoch": 0.05213967819239986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6033,
      "step": 1523
    },
    {
      "epoch": 0.05217391304347826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.653,
      "step": 1524
    },
    {
      "epoch": 0.05220814789455666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.646,
      "step": 1525
    },
    {
      "epoch": 0.05224238274563506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.618,
      "step": 1526
    },
    {
      "epoch": 0.052276617596713454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6225,
      "step": 1527
    },
    {
      "epoch": 0.05231085244779185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5982,
      "step": 1528
    },
    {
      "epoch": 0.052345087298870246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6501,
      "step": 1529
    },
    {
      "epoch": 0.05237932214994865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.624,
      "step": 1530
    },
    {
      "epoch": 0.052413557001027046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6352,
      "step": 1531
    },
    {
      "epoch": 0.05244779185210544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6264,
      "step": 1532
    },
    {
      "epoch": 0.05248202670318384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5983,
      "step": 1533
    },
    {
      "epoch": 0.05251626155426224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6,
      "step": 1534
    },
    {
      "epoch": 0.05255049640534064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6323,
      "step": 1535
    },
    {
      "epoch": 0.052584731256419034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6175,
      "step": 1536
    },
    {
      "epoch": 0.05261896610749743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6192,
      "step": 1537
    },
    {
      "epoch": 0.05265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6318,
      "step": 1538
    },
    {
      "epoch": 0.05268743580965423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6076,
      "step": 1539
    },
    {
      "epoch": 0.05272167066073263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6182,
      "step": 1540
    },
    {
      "epoch": 0.05275590551181102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6362,
      "step": 1541
    },
    {
      "epoch": 0.05279014036288942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6162,
      "step": 1542
    },
    {
      "epoch": 0.05282437521396782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6263,
      "step": 1543
    },
    {
      "epoch": 0.05285861006504622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6093,
      "step": 1544
    },
    {
      "epoch": 0.052892844916124615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6214,
      "step": 1545
    },
    {
      "epoch": 0.05292707976720301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6232,
      "step": 1546
    },
    {
      "epoch": 0.05296131461828141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5945,
      "step": 1547
    },
    {
      "epoch": 0.05299554946935981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5994,
      "step": 1548
    },
    {
      "epoch": 0.05302978432043821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6021,
      "step": 1549
    },
    {
      "epoch": 0.0530640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5932,
      "step": 1550
    },
    {
      "epoch": 0.053098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6021,
      "step": 1551
    },
    {
      "epoch": 0.053132488873673396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6363,
      "step": 1552
    },
    {
      "epoch": 0.0531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6186,
      "step": 1553
    },
    {
      "epoch": 0.053200958575830196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5972,
      "step": 1554
    },
    {
      "epoch": 0.05323519342690859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.641,
      "step": 1555
    },
    {
      "epoch": 0.05326942827798699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6417,
      "step": 1556
    },
    {
      "epoch": 0.05330366312906539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6246,
      "step": 1557
    },
    {
      "epoch": 0.05333789798014379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6104,
      "step": 1558
    },
    {
      "epoch": 0.053372132831222184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6063,
      "step": 1559
    },
    {
      "epoch": 0.05340636768230058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6166,
      "step": 1560
    },
    {
      "epoch": 0.05344060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5945,
      "step": 1561
    },
    {
      "epoch": 0.05347483738445738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6198,
      "step": 1562
    },
    {
      "epoch": 0.053509072235535776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6281,
      "step": 1563
    },
    {
      "epoch": 0.05354330708661417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.594,
      "step": 1564
    },
    {
      "epoch": 0.05357754193769257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.596,
      "step": 1565
    },
    {
      "epoch": 0.05361177678877097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6077,
      "step": 1566
    },
    {
      "epoch": 0.05364601163984937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6026,
      "step": 1567
    },
    {
      "epoch": 0.053680246490927765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6078,
      "step": 1568
    },
    {
      "epoch": 0.05371448134200616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5964,
      "step": 1569
    },
    {
      "epoch": 0.05374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5956,
      "step": 1570
    },
    {
      "epoch": 0.05378295104416296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6158,
      "step": 1571
    },
    {
      "epoch": 0.05381718589524136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6226,
      "step": 1572
    },
    {
      "epoch": 0.05385142074631975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6219,
      "step": 1573
    },
    {
      "epoch": 0.05388565559739815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6175,
      "step": 1574
    },
    {
      "epoch": 0.053919890448476546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5784,
      "step": 1575
    },
    {
      "epoch": 0.05395412529955495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5907,
      "step": 1576
    },
    {
      "epoch": 0.053988360150633345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5799,
      "step": 1577
    },
    {
      "epoch": 0.05402259500171174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6114,
      "step": 1578
    },
    {
      "epoch": 0.05405682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5965,
      "step": 1579
    },
    {
      "epoch": 0.05409106470386854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5842,
      "step": 1580
    },
    {
      "epoch": 0.05412529955494694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5888,
      "step": 1581
    },
    {
      "epoch": 0.054159534406025334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6056,
      "step": 1582
    },
    {
      "epoch": 0.05419376925710373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.583,
      "step": 1583
    },
    {
      "epoch": 0.054228004108182126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5763,
      "step": 1584
    },
    {
      "epoch": 0.05426223895926053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5896,
      "step": 1585
    },
    {
      "epoch": 0.054296473810338926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6093,
      "step": 1586
    },
    {
      "epoch": 0.05433070866141732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5832,
      "step": 1587
    },
    {
      "epoch": 0.05436494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5723,
      "step": 1588
    },
    {
      "epoch": 0.05439917836357412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5993,
      "step": 1589
    },
    {
      "epoch": 0.05443341321465252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5822,
      "step": 1590
    },
    {
      "epoch": 0.054467648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5938,
      "step": 1591
    },
    {
      "epoch": 0.05450188291680931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5981,
      "step": 1592
    },
    {
      "epoch": 0.05453611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5796,
      "step": 1593
    },
    {
      "epoch": 0.05457035261896611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5459,
      "step": 1594
    },
    {
      "epoch": 0.054604587470044506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5943,
      "step": 1595
    },
    {
      "epoch": 0.0546388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5767,
      "step": 1596
    },
    {
      "epoch": 0.0546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6,
      "step": 1597
    },
    {
      "epoch": 0.0547072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6043,
      "step": 1598
    },
    {
      "epoch": 0.0547415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6039,
      "step": 1599
    },
    {
      "epoch": 0.054775761725436495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5523,
      "step": 1600
    },
    {
      "epoch": 0.05480999657651489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5658,
      "step": 1601
    },
    {
      "epoch": 0.05484423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5911,
      "step": 1602
    },
    {
      "epoch": 0.05487846627867169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5879,
      "step": 1603
    },
    {
      "epoch": 0.05491270112975009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5454,
      "step": 1604
    },
    {
      "epoch": 0.05494693598082848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5922,
      "step": 1605
    },
    {
      "epoch": 0.05498117083190688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5784,
      "step": 1606
    },
    {
      "epoch": 0.055015405682985276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5609,
      "step": 1607
    },
    {
      "epoch": 0.05504964053406368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6006,
      "step": 1608
    },
    {
      "epoch": 0.055083875385142075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6006,
      "step": 1609
    },
    {
      "epoch": 0.05511811023622047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6028,
      "step": 1610
    },
    {
      "epoch": 0.05515234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5712,
      "step": 1611
    },
    {
      "epoch": 0.05518657993837727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5979,
      "step": 1612
    },
    {
      "epoch": 0.05522081478945567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5954,
      "step": 1613
    },
    {
      "epoch": 0.055255049640534064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5624,
      "step": 1614
    },
    {
      "epoch": 0.05528928449161246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5824,
      "step": 1615
    },
    {
      "epoch": 0.055323519342690856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5615,
      "step": 1616
    },
    {
      "epoch": 0.05535775419376926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5657,
      "step": 1617
    },
    {
      "epoch": 0.055391989044847656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5574,
      "step": 1618
    },
    {
      "epoch": 0.05542622389592605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5929,
      "step": 1619
    },
    {
      "epoch": 0.05546045874700445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5955,
      "step": 1620
    },
    {
      "epoch": 0.05549469359808285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5979,
      "step": 1621
    },
    {
      "epoch": 0.05552892844916125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5685,
      "step": 1622
    },
    {
      "epoch": 0.055563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5862,
      "step": 1623
    },
    {
      "epoch": 0.05559739815131804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.561,
      "step": 1624
    },
    {
      "epoch": 0.05563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.556,
      "step": 1625
    },
    {
      "epoch": 0.05566586785347484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5792,
      "step": 1626
    },
    {
      "epoch": 0.05570010270455324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5613,
      "step": 1627
    },
    {
      "epoch": 0.05573433755563163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5894,
      "step": 1628
    },
    {
      "epoch": 0.05576857240671003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5579,
      "step": 1629
    },
    {
      "epoch": 0.055802807257788425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5504,
      "step": 1630
    },
    {
      "epoch": 0.05583704210886683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5477,
      "step": 1631
    },
    {
      "epoch": 0.055871276959945225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5764,
      "step": 1632
    },
    {
      "epoch": 0.05590551181102362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5713,
      "step": 1633
    },
    {
      "epoch": 0.05593974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5798,
      "step": 1634
    },
    {
      "epoch": 0.05597398151318042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5765,
      "step": 1635
    },
    {
      "epoch": 0.05600821636425882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5707,
      "step": 1636
    },
    {
      "epoch": 0.05604245121533721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5741,
      "step": 1637
    },
    {
      "epoch": 0.05607668606641561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5631,
      "step": 1638
    },
    {
      "epoch": 0.056110920917494006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5703,
      "step": 1639
    },
    {
      "epoch": 0.05614515576857241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5673,
      "step": 1640
    },
    {
      "epoch": 0.056179390619650806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.566,
      "step": 1641
    },
    {
      "epoch": 0.0562136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5208,
      "step": 1642
    },
    {
      "epoch": 0.0562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5503,
      "step": 1643
    },
    {
      "epoch": 0.056282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5602,
      "step": 1644
    },
    {
      "epoch": 0.0563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5477,
      "step": 1645
    },
    {
      "epoch": 0.056350564875042794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5454,
      "step": 1646
    },
    {
      "epoch": 0.05638479972612119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5272,
      "step": 1647
    },
    {
      "epoch": 0.05641903457719959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.559,
      "step": 1648
    },
    {
      "epoch": 0.05645326942827799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5591,
      "step": 1649
    },
    {
      "epoch": 0.056487504279356386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5698,
      "step": 1650
    },
    {
      "epoch": 0.05652173913043478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5609,
      "step": 1651
    },
    {
      "epoch": 0.05655597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5731,
      "step": 1652
    },
    {
      "epoch": 0.056590208832591575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5745,
      "step": 1653
    },
    {
      "epoch": 0.05662444368366998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5908,
      "step": 1654
    },
    {
      "epoch": 0.056658678534748375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5566,
      "step": 1655
    },
    {
      "epoch": 0.05669291338582677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5615,
      "step": 1656
    },
    {
      "epoch": 0.05672714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.573,
      "step": 1657
    },
    {
      "epoch": 0.05676138308798357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5509,
      "step": 1658
    },
    {
      "epoch": 0.05679561793906197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5354,
      "step": 1659
    },
    {
      "epoch": 0.05682985279014036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.536,
      "step": 1660
    },
    {
      "epoch": 0.05686408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5537,
      "step": 1661
    },
    {
      "epoch": 0.056898322492297156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5665,
      "step": 1662
    },
    {
      "epoch": 0.05693255734337556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5268,
      "step": 1663
    },
    {
      "epoch": 0.056966792194453955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5498,
      "step": 1664
    },
    {
      "epoch": 0.05700102704553235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5371,
      "step": 1665
    },
    {
      "epoch": 0.05703526189661075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.561,
      "step": 1666
    },
    {
      "epoch": 0.05706949674768915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.547,
      "step": 1667
    },
    {
      "epoch": 0.05710373159876755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5558,
      "step": 1668
    },
    {
      "epoch": 0.057137966449845944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5406,
      "step": 1669
    },
    {
      "epoch": 0.05717220130092434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5668,
      "step": 1670
    },
    {
      "epoch": 0.057206436152002736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5594,
      "step": 1671
    },
    {
      "epoch": 0.05724067100308114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5567,
      "step": 1672
    },
    {
      "epoch": 0.057274905854159536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5338,
      "step": 1673
    },
    {
      "epoch": 0.05730914070523793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5378,
      "step": 1674
    },
    {
      "epoch": 0.05734337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5642,
      "step": 1675
    },
    {
      "epoch": 0.057377610407394725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5387,
      "step": 1676
    },
    {
      "epoch": 0.05741184525847313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5366,
      "step": 1677
    },
    {
      "epoch": 0.057446080109551524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5175,
      "step": 1678
    },
    {
      "epoch": 0.05748031496062992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5511,
      "step": 1679
    },
    {
      "epoch": 0.05751454981170832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5327,
      "step": 1680
    },
    {
      "epoch": 0.05754878466278672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5619,
      "step": 1681
    },
    {
      "epoch": 0.057583019513865116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5372,
      "step": 1682
    },
    {
      "epoch": 0.05761725436494351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5304,
      "step": 1683
    },
    {
      "epoch": 0.05765148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5163,
      "step": 1684
    },
    {
      "epoch": 0.057685724067100305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5334,
      "step": 1685
    },
    {
      "epoch": 0.05771995891817871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5593,
      "step": 1686
    },
    {
      "epoch": 0.057754193769257105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5755,
      "step": 1687
    },
    {
      "epoch": 0.0577884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5773,
      "step": 1688
    },
    {
      "epoch": 0.0578226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5671,
      "step": 1689
    },
    {
      "epoch": 0.0578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.523,
      "step": 1690
    },
    {
      "epoch": 0.0578911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.57,
      "step": 1691
    },
    {
      "epoch": 0.05792536802464909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5236,
      "step": 1692
    },
    {
      "epoch": 0.05795960287572749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.561,
      "step": 1693
    },
    {
      "epoch": 0.057993837726805886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5534,
      "step": 1694
    },
    {
      "epoch": 0.05802807257788429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5111,
      "step": 1695
    },
    {
      "epoch": 0.058062307428962685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5451,
      "step": 1696
    },
    {
      "epoch": 0.05809654228004108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5293,
      "step": 1697
    },
    {
      "epoch": 0.05813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5172,
      "step": 1698
    },
    {
      "epoch": 0.058165011982197874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5125,
      "step": 1699
    },
    {
      "epoch": 0.05819924683327628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5139,
      "step": 1700
    },
    {
      "epoch": 0.058233481684354674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5355,
      "step": 1701
    },
    {
      "epoch": 0.05826771653543307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5017,
      "step": 1702
    },
    {
      "epoch": 0.058301951386511466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5057,
      "step": 1703
    },
    {
      "epoch": 0.05833618623758987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5024,
      "step": 1704
    },
    {
      "epoch": 0.058370421088668266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5435,
      "step": 1705
    },
    {
      "epoch": 0.05840465593974666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5483,
      "step": 1706
    },
    {
      "epoch": 0.05843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4895,
      "step": 1707
    },
    {
      "epoch": 0.058473125641903455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5257,
      "step": 1708
    },
    {
      "epoch": 0.05850736049298186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.526,
      "step": 1709
    },
    {
      "epoch": 0.058541595344060254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5155,
      "step": 1710
    },
    {
      "epoch": 0.05857583019513865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5293,
      "step": 1711
    },
    {
      "epoch": 0.05861006504621705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5096,
      "step": 1712
    },
    {
      "epoch": 0.05864429989729545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5299,
      "step": 1713
    },
    {
      "epoch": 0.058678534748373846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5123,
      "step": 1714
    },
    {
      "epoch": 0.05871276959945224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5075,
      "step": 1715
    },
    {
      "epoch": 0.05874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5263,
      "step": 1716
    },
    {
      "epoch": 0.058781239301609035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5167,
      "step": 1717
    },
    {
      "epoch": 0.05881547415268744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5114,
      "step": 1718
    },
    {
      "epoch": 0.058849709003765835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5286,
      "step": 1719
    },
    {
      "epoch": 0.05888394385484423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5085,
      "step": 1720
    },
    {
      "epoch": 0.05891817870592263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5253,
      "step": 1721
    },
    {
      "epoch": 0.058952413557001024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5212,
      "step": 1722
    },
    {
      "epoch": 0.05898664840807943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5062,
      "step": 1723
    },
    {
      "epoch": 0.05902088325915782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4973,
      "step": 1724
    },
    {
      "epoch": 0.05905511811023622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5322,
      "step": 1725
    },
    {
      "epoch": 0.059089352961314616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5456,
      "step": 1726
    },
    {
      "epoch": 0.05912358781239302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.501,
      "step": 1727
    },
    {
      "epoch": 0.059157822663471415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5295,
      "step": 1728
    },
    {
      "epoch": 0.05919205751454981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5345,
      "step": 1729
    },
    {
      "epoch": 0.05922629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.504,
      "step": 1730
    },
    {
      "epoch": 0.059260527216706604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5069,
      "step": 1731
    },
    {
      "epoch": 0.05929476206778501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.471,
      "step": 1732
    },
    {
      "epoch": 0.059328996918863404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5173,
      "step": 1733
    },
    {
      "epoch": 0.0593632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4982,
      "step": 1734
    },
    {
      "epoch": 0.0593974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5443,
      "step": 1735
    },
    {
      "epoch": 0.0594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5213,
      "step": 1736
    },
    {
      "epoch": 0.059465936323176996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5397,
      "step": 1737
    },
    {
      "epoch": 0.05950017117425539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4943,
      "step": 1738
    },
    {
      "epoch": 0.05953440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5198,
      "step": 1739
    },
    {
      "epoch": 0.059568640876412185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5045,
      "step": 1740
    },
    {
      "epoch": 0.05960287572749059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4559,
      "step": 1741
    },
    {
      "epoch": 0.059637110578568985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5267,
      "step": 1742
    },
    {
      "epoch": 0.05967134542964738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5108,
      "step": 1743
    },
    {
      "epoch": 0.05970558028072578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5278,
      "step": 1744
    },
    {
      "epoch": 0.05973981513180417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4986,
      "step": 1745
    },
    {
      "epoch": 0.05977404998288258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4926,
      "step": 1746
    },
    {
      "epoch": 0.05980828483396097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5137,
      "step": 1747
    },
    {
      "epoch": 0.05984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5026,
      "step": 1748
    },
    {
      "epoch": 0.059876754536117766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4937,
      "step": 1749
    },
    {
      "epoch": 0.05991098938719617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5054,
      "step": 1750
    },
    {
      "epoch": 0.059945224238274565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5182,
      "step": 1751
    },
    {
      "epoch": 0.05997945908935296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4755,
      "step": 1752
    },
    {
      "epoch": 0.06001369394043136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4678,
      "step": 1753
    },
    {
      "epoch": 0.060047928791509754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5129,
      "step": 1754
    },
    {
      "epoch": 0.06008216364258816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4845,
      "step": 1755
    },
    {
      "epoch": 0.060116398493666554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5112,
      "step": 1756
    },
    {
      "epoch": 0.06015063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5085,
      "step": 1757
    },
    {
      "epoch": 0.060184868195823346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5234,
      "step": 1758
    },
    {
      "epoch": 0.06021910304690175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5047,
      "step": 1759
    },
    {
      "epoch": 0.060253337897980146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5196,
      "step": 1760
    },
    {
      "epoch": 0.06028757274905854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5147,
      "step": 1761
    },
    {
      "epoch": 0.06032180760013694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5108,
      "step": 1762
    },
    {
      "epoch": 0.060356042451215335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5197,
      "step": 1763
    },
    {
      "epoch": 0.06039027730229374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4831,
      "step": 1764
    },
    {
      "epoch": 0.060424512153372134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5011,
      "step": 1765
    },
    {
      "epoch": 0.06045874700445053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4996,
      "step": 1766
    },
    {
      "epoch": 0.06049298185552893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4991,
      "step": 1767
    },
    {
      "epoch": 0.06052721670660732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.482,
      "step": 1768
    },
    {
      "epoch": 0.060561451557685726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4631,
      "step": 1769
    },
    {
      "epoch": 0.06059568640876412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4967,
      "step": 1770
    },
    {
      "epoch": 0.06062992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4611,
      "step": 1771
    },
    {
      "epoch": 0.060664156110920915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.474,
      "step": 1772
    },
    {
      "epoch": 0.06069839096199932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5041,
      "step": 1773
    },
    {
      "epoch": 0.060732625813077715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4933,
      "step": 1774
    },
    {
      "epoch": 0.06076686066415611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4814,
      "step": 1775
    },
    {
      "epoch": 0.06080109551523451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.482,
      "step": 1776
    },
    {
      "epoch": 0.060835330366312904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4795,
      "step": 1777
    },
    {
      "epoch": 0.06086956521739131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4659,
      "step": 1778
    },
    {
      "epoch": 0.0609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5121,
      "step": 1779
    },
    {
      "epoch": 0.0609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5175,
      "step": 1780
    },
    {
      "epoch": 0.060972269770626496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5054,
      "step": 1781
    },
    {
      "epoch": 0.0610065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4889,
      "step": 1782
    },
    {
      "epoch": 0.061040739472783295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4842,
      "step": 1783
    },
    {
      "epoch": 0.06107497432386169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4709,
      "step": 1784
    },
    {
      "epoch": 0.06110920917494009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4792,
      "step": 1785
    },
    {
      "epoch": 0.061143444026018484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4958,
      "step": 1786
    },
    {
      "epoch": 0.06117767887709689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4737,
      "step": 1787
    },
    {
      "epoch": 0.061211913728175284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4928,
      "step": 1788
    },
    {
      "epoch": 0.06124614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4941,
      "step": 1789
    },
    {
      "epoch": 0.061280383430332076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4842,
      "step": 1790
    },
    {
      "epoch": 0.06131461828141047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4825,
      "step": 1791
    },
    {
      "epoch": 0.061348853132488876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.464,
      "step": 1792
    },
    {
      "epoch": 0.06138308798356727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5104,
      "step": 1793
    },
    {
      "epoch": 0.06141732283464567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4911,
      "step": 1794
    },
    {
      "epoch": 0.061451557685724065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5325,
      "step": 1795
    },
    {
      "epoch": 0.06148579253680247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4788,
      "step": 1796
    },
    {
      "epoch": 0.061520027387880864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5253,
      "step": 1797
    },
    {
      "epoch": 0.06155426223895926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5323,
      "step": 1798
    },
    {
      "epoch": 0.06158849709003766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4863,
      "step": 1799
    },
    {
      "epoch": 0.06162273194111605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5031,
      "step": 1800
    },
    {
      "epoch": 0.061656966792194456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4939,
      "step": 1801
    },
    {
      "epoch": 0.06169120164327285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4937,
      "step": 1802
    },
    {
      "epoch": 0.06172543649435125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4873,
      "step": 1803
    },
    {
      "epoch": 0.061759671345429645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4731,
      "step": 1804
    },
    {
      "epoch": 0.06179390619650805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4452,
      "step": 1805
    },
    {
      "epoch": 0.061828141047586445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4928,
      "step": 1806
    },
    {
      "epoch": 0.06186237589866484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4806,
      "step": 1807
    },
    {
      "epoch": 0.06189661074974324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4977,
      "step": 1808
    },
    {
      "epoch": 0.061930845600821634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4528,
      "step": 1809
    },
    {
      "epoch": 0.06196508045190004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5188,
      "step": 1810
    },
    {
      "epoch": 0.06199931530297843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4792,
      "step": 1811
    },
    {
      "epoch": 0.06203355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4982,
      "step": 1812
    },
    {
      "epoch": 0.062067785005135226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4855,
      "step": 1813
    },
    {
      "epoch": 0.06210201985621362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5069,
      "step": 1814
    },
    {
      "epoch": 0.062136254707292025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4519,
      "step": 1815
    },
    {
      "epoch": 0.06217048955837042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4394,
      "step": 1816
    },
    {
      "epoch": 0.06220472440944882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4543,
      "step": 1817
    },
    {
      "epoch": 0.062238959260527214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.464,
      "step": 1818
    },
    {
      "epoch": 0.06227319411160562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4765,
      "step": 1819
    },
    {
      "epoch": 0.062307428962684014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4864,
      "step": 1820
    },
    {
      "epoch": 0.06234166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4435,
      "step": 1821
    },
    {
      "epoch": 0.062375898664840806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4875,
      "step": 1822
    },
    {
      "epoch": 0.0624101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4458,
      "step": 1823
    },
    {
      "epoch": 0.062444368366997606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4458,
      "step": 1824
    },
    {
      "epoch": 0.062478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4867,
      "step": 1825
    },
    {
      "epoch": 0.0625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4661,
      "step": 1826
    },
    {
      "epoch": 0.0625470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4739,
      "step": 1827
    },
    {
      "epoch": 0.0625813077713112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4715,
      "step": 1828
    },
    {
      "epoch": 0.0626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4728,
      "step": 1829
    },
    {
      "epoch": 0.06264977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4673,
      "step": 1830
    },
    {
      "epoch": 0.06268401232454639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4934,
      "step": 1831
    },
    {
      "epoch": 0.06271824717562478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4789,
      "step": 1832
    },
    {
      "epoch": 0.06275248202670318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.493,
      "step": 1833
    },
    {
      "epoch": 0.06278671687778158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4511,
      "step": 1834
    },
    {
      "epoch": 0.06282095172885999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4592,
      "step": 1835
    },
    {
      "epoch": 0.06285518657993838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4643,
      "step": 1836
    },
    {
      "epoch": 0.06288942143101678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4704,
      "step": 1837
    },
    {
      "epoch": 0.06292365628209518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4867,
      "step": 1838
    },
    {
      "epoch": 0.06295789113317357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4563,
      "step": 1839
    },
    {
      "epoch": 0.06299212598425197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4849,
      "step": 1840
    },
    {
      "epoch": 0.06302636083533036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4863,
      "step": 1841
    },
    {
      "epoch": 0.06306059568640876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4881,
      "step": 1842
    },
    {
      "epoch": 0.06309483053748716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.475,
      "step": 1843
    },
    {
      "epoch": 0.06312906538856555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4641,
      "step": 1844
    },
    {
      "epoch": 0.06316330023964396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4573,
      "step": 1845
    },
    {
      "epoch": 0.06319753509072236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4458,
      "step": 1846
    },
    {
      "epoch": 0.06323176994180076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4607,
      "step": 1847
    },
    {
      "epoch": 0.06326600479287915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4448,
      "step": 1848
    },
    {
      "epoch": 0.06330023964395755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.448,
      "step": 1849
    },
    {
      "epoch": 0.06333447449503594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4729,
      "step": 1850
    },
    {
      "epoch": 0.06336870934611434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4408,
      "step": 1851
    },
    {
      "epoch": 0.06340294419719274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4961,
      "step": 1852
    },
    {
      "epoch": 0.06343717904827113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4653,
      "step": 1853
    },
    {
      "epoch": 0.06347141389934954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4638,
      "step": 1854
    },
    {
      "epoch": 0.06350564875042794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4448,
      "step": 1855
    },
    {
      "epoch": 0.06353988360150634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4423,
      "step": 1856
    },
    {
      "epoch": 0.06357411845258473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4617,
      "step": 1857
    },
    {
      "epoch": 0.06360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4459,
      "step": 1858
    },
    {
      "epoch": 0.06364258815474153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4634,
      "step": 1859
    },
    {
      "epoch": 0.06367682300581992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4873,
      "step": 1860
    },
    {
      "epoch": 0.06371105785689832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4369,
      "step": 1861
    },
    {
      "epoch": 0.06374529270797671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4639,
      "step": 1862
    },
    {
      "epoch": 0.06377952755905512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4578,
      "step": 1863
    },
    {
      "epoch": 0.06381376241013352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4817,
      "step": 1864
    },
    {
      "epoch": 0.06384799726121192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.459,
      "step": 1865
    },
    {
      "epoch": 0.06388223211229031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4409,
      "step": 1866
    },
    {
      "epoch": 0.06391646696336871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4459,
      "step": 1867
    },
    {
      "epoch": 0.0639507018144471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4335,
      "step": 1868
    },
    {
      "epoch": 0.0639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4711,
      "step": 1869
    },
    {
      "epoch": 0.0640191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4559,
      "step": 1870
    },
    {
      "epoch": 0.0640534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4565,
      "step": 1871
    },
    {
      "epoch": 0.0640876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4579,
      "step": 1872
    },
    {
      "epoch": 0.0641218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4077,
      "step": 1873
    },
    {
      "epoch": 0.0641561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4519,
      "step": 1874
    },
    {
      "epoch": 0.0641903457719959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4518,
      "step": 1875
    },
    {
      "epoch": 0.06422458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4241,
      "step": 1876
    },
    {
      "epoch": 0.06425881547415269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.45,
      "step": 1877
    },
    {
      "epoch": 0.06429305032523108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4348,
      "step": 1878
    },
    {
      "epoch": 0.06432728517630948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4541,
      "step": 1879
    },
    {
      "epoch": 0.06436152002738788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4623,
      "step": 1880
    },
    {
      "epoch": 0.06439575487846629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4526,
      "step": 1881
    },
    {
      "epoch": 0.06442998972954468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4628,
      "step": 1882
    },
    {
      "epoch": 0.06446422458062308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4443,
      "step": 1883
    },
    {
      "epoch": 0.06449845943170147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4362,
      "step": 1884
    },
    {
      "epoch": 0.06453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4581,
      "step": 1885
    },
    {
      "epoch": 0.06456692913385827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4521,
      "step": 1886
    },
    {
      "epoch": 0.06460116398493666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4465,
      "step": 1887
    },
    {
      "epoch": 0.06463539883601506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4447,
      "step": 1888
    },
    {
      "epoch": 0.06466963368709346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4556,
      "step": 1889
    },
    {
      "epoch": 0.06470386853817185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4091,
      "step": 1890
    },
    {
      "epoch": 0.06473810338925026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4519,
      "step": 1891
    },
    {
      "epoch": 0.06477233824032866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4451,
      "step": 1892
    },
    {
      "epoch": 0.06480657309140705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4321,
      "step": 1893
    },
    {
      "epoch": 0.06484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4403,
      "step": 1894
    },
    {
      "epoch": 0.06487504279356385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.449,
      "step": 1895
    },
    {
      "epoch": 0.06490927764464224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4741,
      "step": 1896
    },
    {
      "epoch": 0.06494351249572064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4854,
      "step": 1897
    },
    {
      "epoch": 0.06497774734679904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4257,
      "step": 1898
    },
    {
      "epoch": 0.06501198219787743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4472,
      "step": 1899
    },
    {
      "epoch": 0.06504621704895584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4629,
      "step": 1900
    },
    {
      "epoch": 0.06508045190003424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4554,
      "step": 1901
    },
    {
      "epoch": 0.06511468675111264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4155,
      "step": 1902
    },
    {
      "epoch": 0.06514892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4481,
      "step": 1903
    },
    {
      "epoch": 0.06518315645326943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4268,
      "step": 1904
    },
    {
      "epoch": 0.06521739130434782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4512,
      "step": 1905
    },
    {
      "epoch": 0.06525162615542622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4358,
      "step": 1906
    },
    {
      "epoch": 0.06528586100650462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4354,
      "step": 1907
    },
    {
      "epoch": 0.06532009585758301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4165,
      "step": 1908
    },
    {
      "epoch": 0.06535433070866142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4398,
      "step": 1909
    },
    {
      "epoch": 0.06538856555973982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.428,
      "step": 1910
    },
    {
      "epoch": 0.06542280041081822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.432,
      "step": 1911
    },
    {
      "epoch": 0.06545703526189661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4139,
      "step": 1912
    },
    {
      "epoch": 0.06549127011297501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4309,
      "step": 1913
    },
    {
      "epoch": 0.0655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.432,
      "step": 1914
    },
    {
      "epoch": 0.0655597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4606,
      "step": 1915
    },
    {
      "epoch": 0.0655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4342,
      "step": 1916
    },
    {
      "epoch": 0.0656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3964,
      "step": 1917
    },
    {
      "epoch": 0.065662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4365,
      "step": 1918
    },
    {
      "epoch": 0.0656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4231,
      "step": 1919
    },
    {
      "epoch": 0.0657309140705238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4243,
      "step": 1920
    },
    {
      "epoch": 0.06576514892160219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4284,
      "step": 1921
    },
    {
      "epoch": 0.06579938377268059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4059,
      "step": 1922
    },
    {
      "epoch": 0.06583361862375899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.423,
      "step": 1923
    },
    {
      "epoch": 0.06586785347483738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3991,
      "step": 1924
    },
    {
      "epoch": 0.06590208832591578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3999,
      "step": 1925
    },
    {
      "epoch": 0.06593632317699417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4225,
      "step": 1926
    },
    {
      "epoch": 0.06597055802807258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4343,
      "step": 1927
    },
    {
      "epoch": 0.06600479287915098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.464,
      "step": 1928
    },
    {
      "epoch": 0.06603902773022938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4455,
      "step": 1929
    },
    {
      "epoch": 0.06607326258130777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.409,
      "step": 1930
    },
    {
      "epoch": 0.06610749743238617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4353,
      "step": 1931
    },
    {
      "epoch": 0.06614173228346457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.402,
      "step": 1932
    },
    {
      "epoch": 0.06617596713454296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4112,
      "step": 1933
    },
    {
      "epoch": 0.06621020198562136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4212,
      "step": 1934
    },
    {
      "epoch": 0.06624443683669975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4186,
      "step": 1935
    },
    {
      "epoch": 0.06627867168777817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.445,
      "step": 1936
    },
    {
      "epoch": 0.06631290653885656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4566,
      "step": 1937
    },
    {
      "epoch": 0.06634714138993496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4082,
      "step": 1938
    },
    {
      "epoch": 0.06638137624101335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3776,
      "step": 1939
    },
    {
      "epoch": 0.06641561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4239,
      "step": 1940
    },
    {
      "epoch": 0.06644984594317015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4056,
      "step": 1941
    },
    {
      "epoch": 0.06648408079424854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3793,
      "step": 1942
    },
    {
      "epoch": 0.06651831564532694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4333,
      "step": 1943
    },
    {
      "epoch": 0.06655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4185,
      "step": 1944
    },
    {
      "epoch": 0.06658678534748373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4034,
      "step": 1945
    },
    {
      "epoch": 0.06662102019856214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4528,
      "step": 1946
    },
    {
      "epoch": 0.06665525504964054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3828,
      "step": 1947
    },
    {
      "epoch": 0.06668948990071893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4332,
      "step": 1948
    },
    {
      "epoch": 0.06672372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4432,
      "step": 1949
    },
    {
      "epoch": 0.06675795960287573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4143,
      "step": 1950
    },
    {
      "epoch": 0.06679219445395412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4215,
      "step": 1951
    },
    {
      "epoch": 0.06682642930503252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4172,
      "step": 1952
    },
    {
      "epoch": 0.06686066415611092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4231,
      "step": 1953
    },
    {
      "epoch": 0.06689489900718931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4168,
      "step": 1954
    },
    {
      "epoch": 0.06692913385826772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3883,
      "step": 1955
    },
    {
      "epoch": 0.06696336870934612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4106,
      "step": 1956
    },
    {
      "epoch": 0.06699760356042452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4129,
      "step": 1957
    },
    {
      "epoch": 0.06703183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4309,
      "step": 1958
    },
    {
      "epoch": 0.06706607326258131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4143,
      "step": 1959
    },
    {
      "epoch": 0.0671003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.41,
      "step": 1960
    },
    {
      "epoch": 0.0671345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4009,
      "step": 1961
    },
    {
      "epoch": 0.0671687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4016,
      "step": 1962
    },
    {
      "epoch": 0.06720301266689489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.434,
      "step": 1963
    },
    {
      "epoch": 0.0672372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4002,
      "step": 1964
    },
    {
      "epoch": 0.0672714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4137,
      "step": 1965
    },
    {
      "epoch": 0.0673057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4035,
      "step": 1966
    },
    {
      "epoch": 0.06733995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3769,
      "step": 1967
    },
    {
      "epoch": 0.06737418692228689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4151,
      "step": 1968
    },
    {
      "epoch": 0.06740842177336528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3803,
      "step": 1969
    },
    {
      "epoch": 0.06744265662444368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3967,
      "step": 1970
    },
    {
      "epoch": 0.06747689147552208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.413,
      "step": 1971
    },
    {
      "epoch": 0.06751112632660047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4046,
      "step": 1972
    },
    {
      "epoch": 0.06754536117767888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.392,
      "step": 1973
    },
    {
      "epoch": 0.06757959602875728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3975,
      "step": 1974
    },
    {
      "epoch": 0.06761383087983568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3979,
      "step": 1975
    },
    {
      "epoch": 0.06764806573091407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4137,
      "step": 1976
    },
    {
      "epoch": 0.06768230058199247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4169,
      "step": 1977
    },
    {
      "epoch": 0.06771653543307087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4166,
      "step": 1978
    },
    {
      "epoch": 0.06775077028414926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3807,
      "step": 1979
    },
    {
      "epoch": 0.06778500513522766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4465,
      "step": 1980
    },
    {
      "epoch": 0.06781923998630605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3875,
      "step": 1981
    },
    {
      "epoch": 0.06785347483738446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4102,
      "step": 1982
    },
    {
      "epoch": 0.06788770968846286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3868,
      "step": 1983
    },
    {
      "epoch": 0.06792194453954126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4091,
      "step": 1984
    },
    {
      "epoch": 0.06795617939061965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4004,
      "step": 1985
    },
    {
      "epoch": 0.06799041424169805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4046,
      "step": 1986
    },
    {
      "epoch": 0.06802464909277645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3889,
      "step": 1987
    },
    {
      "epoch": 0.06805888394385484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3978,
      "step": 1988
    },
    {
      "epoch": 0.06809311879493324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3691,
      "step": 1989
    },
    {
      "epoch": 0.06812735364601163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3779,
      "step": 1990
    },
    {
      "epoch": 0.06816158849709003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.36,
      "step": 1991
    },
    {
      "epoch": 0.06819582334816844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4167,
      "step": 1992
    },
    {
      "epoch": 0.06823005819924684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.389,
      "step": 1993
    },
    {
      "epoch": 0.06826429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4171,
      "step": 1994
    },
    {
      "epoch": 0.06829852790140363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.393,
      "step": 1995
    },
    {
      "epoch": 0.06833276275248203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.417,
      "step": 1996
    },
    {
      "epoch": 0.06836699760356042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3925,
      "step": 1997
    },
    {
      "epoch": 0.06840123245463882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.416,
      "step": 1998
    },
    {
      "epoch": 0.06843546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3885,
      "step": 1999
    },
    {
      "epoch": 0.06846970215679561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.403,
      "step": 2000
    },
    {
      "epoch": 0.06850393700787402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4074,
      "step": 2001
    },
    {
      "epoch": 0.06853817185895242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4123,
      "step": 2002
    },
    {
      "epoch": 0.06857240671003081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3942,
      "step": 2003
    },
    {
      "epoch": 0.06860664156110921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3748,
      "step": 2004
    },
    {
      "epoch": 0.0686408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3858,
      "step": 2005
    },
    {
      "epoch": 0.068675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4102,
      "step": 2006
    },
    {
      "epoch": 0.0687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3913,
      "step": 2007
    },
    {
      "epoch": 0.0687435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3851,
      "step": 2008
    },
    {
      "epoch": 0.06877781581650119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4157,
      "step": 2009
    },
    {
      "epoch": 0.0688120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3976,
      "step": 2010
    },
    {
      "epoch": 0.068846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3869,
      "step": 2011
    },
    {
      "epoch": 0.0688805203697364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3944,
      "step": 2012
    },
    {
      "epoch": 0.06891475522081479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3869,
      "step": 2013
    },
    {
      "epoch": 0.06894899007189319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4172,
      "step": 2014
    },
    {
      "epoch": 0.06898322492297158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4013,
      "step": 2015
    },
    {
      "epoch": 0.06901745977404998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4285,
      "step": 2016
    },
    {
      "epoch": 0.06905169462512838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.374,
      "step": 2017
    },
    {
      "epoch": 0.06908592947620677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3691,
      "step": 2018
    },
    {
      "epoch": 0.06912016432728518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3671,
      "step": 2019
    },
    {
      "epoch": 0.06915439917836358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3945,
      "step": 2020
    },
    {
      "epoch": 0.06918863402944198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3795,
      "step": 2021
    },
    {
      "epoch": 0.06922286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3896,
      "step": 2022
    },
    {
      "epoch": 0.06925710373159877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3821,
      "step": 2023
    },
    {
      "epoch": 0.06929133858267716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4053,
      "step": 2024
    },
    {
      "epoch": 0.06932557343375556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3832,
      "step": 2025
    },
    {
      "epoch": 0.06935980828483396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4082,
      "step": 2026
    },
    {
      "epoch": 0.06939404313591235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.375,
      "step": 2027
    },
    {
      "epoch": 0.06942827798699076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3665,
      "step": 2028
    },
    {
      "epoch": 0.06946251283806916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3749,
      "step": 2029
    },
    {
      "epoch": 0.06949674768914756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3549,
      "step": 2030
    },
    {
      "epoch": 0.06953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3898,
      "step": 2031
    },
    {
      "epoch": 0.06956521739130435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3874,
      "step": 2032
    },
    {
      "epoch": 0.06959945224238274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3744,
      "step": 2033
    },
    {
      "epoch": 0.06963368709346114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3983,
      "step": 2034
    },
    {
      "epoch": 0.06966792194453954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3614,
      "step": 2035
    },
    {
      "epoch": 0.06970215679561793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3749,
      "step": 2036
    },
    {
      "epoch": 0.06973639164669633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3514,
      "step": 2037
    },
    {
      "epoch": 0.06977062649777474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3795,
      "step": 2038
    },
    {
      "epoch": 0.06980486134885314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3563,
      "step": 2039
    },
    {
      "epoch": 0.06983909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3377,
      "step": 2040
    },
    {
      "epoch": 0.06987333105100993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.358,
      "step": 2041
    },
    {
      "epoch": 0.06990756590208833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3481,
      "step": 2042
    },
    {
      "epoch": 0.06994180075316672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3692,
      "step": 2043
    },
    {
      "epoch": 0.06997603560424512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3699,
      "step": 2044
    },
    {
      "epoch": 0.07001027045532351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.33,
      "step": 2045
    },
    {
      "epoch": 0.07004450530640191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3534,
      "step": 2046
    },
    {
      "epoch": 0.07007874015748032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3965,
      "step": 2047
    },
    {
      "epoch": 0.07011297500855872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.363,
      "step": 2048
    },
    {
      "epoch": 0.07014720985963711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3823,
      "step": 2049
    },
    {
      "epoch": 0.07018144471071551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3574,
      "step": 2050
    },
    {
      "epoch": 0.0702156795617939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.387,
      "step": 2051
    },
    {
      "epoch": 0.0702499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3721,
      "step": 2052
    },
    {
      "epoch": 0.0702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3754,
      "step": 2053
    },
    {
      "epoch": 0.0703183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3607,
      "step": 2054
    },
    {
      "epoch": 0.07035261896610749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3826,
      "step": 2055
    },
    {
      "epoch": 0.0703868538171859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3504,
      "step": 2056
    },
    {
      "epoch": 0.0704210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3846,
      "step": 2057
    },
    {
      "epoch": 0.0704553235193427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3747,
      "step": 2058
    },
    {
      "epoch": 0.07048955837042109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3682,
      "step": 2059
    },
    {
      "epoch": 0.07052379322149949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3818,
      "step": 2060
    },
    {
      "epoch": 0.07055802807257788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3769,
      "step": 2061
    },
    {
      "epoch": 0.07059226292365628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3662,
      "step": 2062
    },
    {
      "epoch": 0.07062649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.376,
      "step": 2063
    },
    {
      "epoch": 0.07066073262581307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3862,
      "step": 2064
    },
    {
      "epoch": 0.07069496747689148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3885,
      "step": 2065
    },
    {
      "epoch": 0.07072920232796988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3445,
      "step": 2066
    },
    {
      "epoch": 0.07076343717904827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3762,
      "step": 2067
    },
    {
      "epoch": 0.07079767203012667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4246,
      "step": 2068
    },
    {
      "epoch": 0.07083190688120507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3715,
      "step": 2069
    },
    {
      "epoch": 0.07086614173228346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3334,
      "step": 2070
    },
    {
      "epoch": 0.07090037658336186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3469,
      "step": 2071
    },
    {
      "epoch": 0.07093461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.351,
      "step": 2072
    },
    {
      "epoch": 0.07096884628551865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3536,
      "step": 2073
    },
    {
      "epoch": 0.07100308113659706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.379,
      "step": 2074
    },
    {
      "epoch": 0.07103731598767546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3611,
      "step": 2075
    },
    {
      "epoch": 0.07107155083875386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3674,
      "step": 2076
    },
    {
      "epoch": 0.07110578568983225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.412,
      "step": 2077
    },
    {
      "epoch": 0.07114002054091065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3779,
      "step": 2078
    },
    {
      "epoch": 0.07117425539198904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3517,
      "step": 2079
    },
    {
      "epoch": 0.07120849024306744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3458,
      "step": 2080
    },
    {
      "epoch": 0.07124272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.366,
      "step": 2081
    },
    {
      "epoch": 0.07127695994522423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3474,
      "step": 2082
    },
    {
      "epoch": 0.07131119479630263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3785,
      "step": 2083
    },
    {
      "epoch": 0.07134542964738104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3918,
      "step": 2084
    },
    {
      "epoch": 0.07137966449845944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3167,
      "step": 2085
    },
    {
      "epoch": 0.07141389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3755,
      "step": 2086
    },
    {
      "epoch": 0.07144813420061623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3469,
      "step": 2087
    },
    {
      "epoch": 0.07148236905169462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3465,
      "step": 2088
    },
    {
      "epoch": 0.07151660390277302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3454,
      "step": 2089
    },
    {
      "epoch": 0.07155083875385142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3437,
      "step": 2090
    },
    {
      "epoch": 0.07158507360492981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4004,
      "step": 2091
    },
    {
      "epoch": 0.07161930845600821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3728,
      "step": 2092
    },
    {
      "epoch": 0.07165354330708662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3479,
      "step": 2093
    },
    {
      "epoch": 0.07168777815816502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3384,
      "step": 2094
    },
    {
      "epoch": 0.07172201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.355,
      "step": 2095
    },
    {
      "epoch": 0.07175624786032181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3585,
      "step": 2096
    },
    {
      "epoch": 0.0717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3181,
      "step": 2097
    },
    {
      "epoch": 0.0718247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3323,
      "step": 2098
    },
    {
      "epoch": 0.071858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3699,
      "step": 2099
    },
    {
      "epoch": 0.0718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3421,
      "step": 2100
    },
    {
      "epoch": 0.07192742211571379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3499,
      "step": 2101
    },
    {
      "epoch": 0.0719616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3639,
      "step": 2102
    },
    {
      "epoch": 0.0719958918178706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3545,
      "step": 2103
    },
    {
      "epoch": 0.072030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.347,
      "step": 2104
    },
    {
      "epoch": 0.07206436152002739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3821,
      "step": 2105
    },
    {
      "epoch": 0.07209859637110579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3514,
      "step": 2106
    },
    {
      "epoch": 0.07213283122218418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3601,
      "step": 2107
    },
    {
      "epoch": 0.07216706607326258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.326,
      "step": 2108
    },
    {
      "epoch": 0.07220130092434097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.321,
      "step": 2109
    },
    {
      "epoch": 0.07223553577541937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3559,
      "step": 2110
    },
    {
      "epoch": 0.07226977062649778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3517,
      "step": 2111
    },
    {
      "epoch": 0.07230400547757618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3724,
      "step": 2112
    },
    {
      "epoch": 0.07233824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3587,
      "step": 2113
    },
    {
      "epoch": 0.07237247517973297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3576,
      "step": 2114
    },
    {
      "epoch": 0.07240671003081137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3333,
      "step": 2115
    },
    {
      "epoch": 0.07244094488188976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.364,
      "step": 2116
    },
    {
      "epoch": 0.07247517973296816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3421,
      "step": 2117
    },
    {
      "epoch": 0.07250941458404656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3357,
      "step": 2118
    },
    {
      "epoch": 0.07254364943512495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.327,
      "step": 2119
    },
    {
      "epoch": 0.07257788428620336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3486,
      "step": 2120
    },
    {
      "epoch": 0.07261211913728176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.348,
      "step": 2121
    },
    {
      "epoch": 0.07264635398836015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3868,
      "step": 2122
    },
    {
      "epoch": 0.07268058883943855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.35,
      "step": 2123
    },
    {
      "epoch": 0.07271482369051695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3395,
      "step": 2124
    },
    {
      "epoch": 0.07274905854159534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3556,
      "step": 2125
    },
    {
      "epoch": 0.07278329339267374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3392,
      "step": 2126
    },
    {
      "epoch": 0.07281752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3443,
      "step": 2127
    },
    {
      "epoch": 0.07285176309483053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3517,
      "step": 2128
    },
    {
      "epoch": 0.07288599794590893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3625,
      "step": 2129
    },
    {
      "epoch": 0.07292023279698734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3702,
      "step": 2130
    },
    {
      "epoch": 0.07295446764806574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3432,
      "step": 2131
    },
    {
      "epoch": 0.07298870249914413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.349,
      "step": 2132
    },
    {
      "epoch": 0.07302293735022253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3079,
      "step": 2133
    },
    {
      "epoch": 0.07305717220130092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3486,
      "step": 2134
    },
    {
      "epoch": 0.07309140705237932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3121,
      "step": 2135
    },
    {
      "epoch": 0.07312564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3163,
      "step": 2136
    },
    {
      "epoch": 0.07315987675453611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3572,
      "step": 2137
    },
    {
      "epoch": 0.07319411160561451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3095,
      "step": 2138
    },
    {
      "epoch": 0.07322834645669292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3094,
      "step": 2139
    },
    {
      "epoch": 0.07326258130777132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3359,
      "step": 2140
    },
    {
      "epoch": 0.07329681615884971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3258,
      "step": 2141
    },
    {
      "epoch": 0.07333105100992811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3439,
      "step": 2142
    },
    {
      "epoch": 0.0733652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.332,
      "step": 2143
    },
    {
      "epoch": 0.0733995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3454,
      "step": 2144
    },
    {
      "epoch": 0.0734337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3384,
      "step": 2145
    },
    {
      "epoch": 0.0734679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3228,
      "step": 2146
    },
    {
      "epoch": 0.07350222526532009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3293,
      "step": 2147
    },
    {
      "epoch": 0.0735364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3673,
      "step": 2148
    },
    {
      "epoch": 0.0735706949674769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3214,
      "step": 2149
    },
    {
      "epoch": 0.07360492981855529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3688,
      "step": 2150
    },
    {
      "epoch": 0.07363916466963369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3377,
      "step": 2151
    },
    {
      "epoch": 0.07367339952071209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3413,
      "step": 2152
    },
    {
      "epoch": 0.07370763437179048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3308,
      "step": 2153
    },
    {
      "epoch": 0.07374186922286888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3256,
      "step": 2154
    },
    {
      "epoch": 0.07377610407394727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3579,
      "step": 2155
    },
    {
      "epoch": 0.07381033892502567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3166,
      "step": 2156
    },
    {
      "epoch": 0.07384457377610408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3311,
      "step": 2157
    },
    {
      "epoch": 0.07387880862718248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2882,
      "step": 2158
    },
    {
      "epoch": 0.07391304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2924,
      "step": 2159
    },
    {
      "epoch": 0.07394727832933927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3254,
      "step": 2160
    },
    {
      "epoch": 0.07398151318041767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3253,
      "step": 2161
    },
    {
      "epoch": 0.07401574803149606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3427,
      "step": 2162
    },
    {
      "epoch": 0.07404998288257446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3545,
      "step": 2163
    },
    {
      "epoch": 0.07408421773365285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3555,
      "step": 2164
    },
    {
      "epoch": 0.07411845258473125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3032,
      "step": 2165
    },
    {
      "epoch": 0.07415268743580966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.357,
      "step": 2166
    },
    {
      "epoch": 0.07418692228688806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3687,
      "step": 2167
    },
    {
      "epoch": 0.07422115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.354,
      "step": 2168
    },
    {
      "epoch": 0.07425539198904485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3285,
      "step": 2169
    },
    {
      "epoch": 0.07428962684012325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3017,
      "step": 2170
    },
    {
      "epoch": 0.07432386169120164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3347,
      "step": 2171
    },
    {
      "epoch": 0.07435809654228004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.326,
      "step": 2172
    },
    {
      "epoch": 0.07439233139335844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3063,
      "step": 2173
    },
    {
      "epoch": 0.07442656624443683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.305,
      "step": 2174
    },
    {
      "epoch": 0.07446080109551523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2989,
      "step": 2175
    },
    {
      "epoch": 0.07449503594659364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3281,
      "step": 2176
    },
    {
      "epoch": 0.07452927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3661,
      "step": 2177
    },
    {
      "epoch": 0.07456350564875043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3337,
      "step": 2178
    },
    {
      "epoch": 0.07459774049982883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3267,
      "step": 2179
    },
    {
      "epoch": 0.07463197535090722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.347,
      "step": 2180
    },
    {
      "epoch": 0.07466621020198562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3305,
      "step": 2181
    },
    {
      "epoch": 0.07470044505306402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3241,
      "step": 2182
    },
    {
      "epoch": 0.07473467990414241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3388,
      "step": 2183
    },
    {
      "epoch": 0.07476891475522081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2999,
      "step": 2184
    },
    {
      "epoch": 0.07480314960629922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3151,
      "step": 2185
    },
    {
      "epoch": 0.07483738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3491,
      "step": 2186
    },
    {
      "epoch": 0.07487161930845601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.305,
      "step": 2187
    },
    {
      "epoch": 0.07490585415953441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3274,
      "step": 2188
    },
    {
      "epoch": 0.0749400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3262,
      "step": 2189
    },
    {
      "epoch": 0.0749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3165,
      "step": 2190
    },
    {
      "epoch": 0.0750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.308,
      "step": 2191
    },
    {
      "epoch": 0.07504279356384799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3057,
      "step": 2192
    },
    {
      "epoch": 0.07507702841492639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3069,
      "step": 2193
    },
    {
      "epoch": 0.0751112632660048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.315,
      "step": 2194
    },
    {
      "epoch": 0.0751454981170832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3169,
      "step": 2195
    },
    {
      "epoch": 0.07517973296816159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3267,
      "step": 2196
    },
    {
      "epoch": 0.07521396781923999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3338,
      "step": 2197
    },
    {
      "epoch": 0.07524820267031838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.315,
      "step": 2198
    },
    {
      "epoch": 0.07528243752139678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2926,
      "step": 2199
    },
    {
      "epoch": 0.07531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.33,
      "step": 2200
    },
    {
      "epoch": 0.07535090722355357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3323,
      "step": 2201
    },
    {
      "epoch": 0.07538514207463197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3191,
      "step": 2202
    },
    {
      "epoch": 0.07541937692571038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3155,
      "step": 2203
    },
    {
      "epoch": 0.07545361177678878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3049,
      "step": 2204
    },
    {
      "epoch": 0.07548784662786717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3037,
      "step": 2205
    },
    {
      "epoch": 0.07552208147894557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3198,
      "step": 2206
    },
    {
      "epoch": 0.07555631633002396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3204,
      "step": 2207
    },
    {
      "epoch": 0.07559055118110236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3223,
      "step": 2208
    },
    {
      "epoch": 0.07562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2887,
      "step": 2209
    },
    {
      "epoch": 0.07565902088325915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3251,
      "step": 2210
    },
    {
      "epoch": 0.07569325573433755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3045,
      "step": 2211
    },
    {
      "epoch": 0.07572749058541596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3201,
      "step": 2212
    },
    {
      "epoch": 0.07576172543649436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.307,
      "step": 2213
    },
    {
      "epoch": 0.07579596028757275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3109,
      "step": 2214
    },
    {
      "epoch": 0.07583019513865115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2996,
      "step": 2215
    },
    {
      "epoch": 0.07586442998972955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2966,
      "step": 2216
    },
    {
      "epoch": 0.07589866484080794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3223,
      "step": 2217
    },
    {
      "epoch": 0.07593289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2986,
      "step": 2218
    },
    {
      "epoch": 0.07596713454296473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3027,
      "step": 2219
    },
    {
      "epoch": 0.07600136939404313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3151,
      "step": 2220
    },
    {
      "epoch": 0.07603560424512153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3095,
      "step": 2221
    },
    {
      "epoch": 0.07606983909619994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3234,
      "step": 2222
    },
    {
      "epoch": 0.07610407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2987,
      "step": 2223
    },
    {
      "epoch": 0.07613830879835673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3142,
      "step": 2224
    },
    {
      "epoch": 0.07617254364943513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3425,
      "step": 2225
    },
    {
      "epoch": 0.07620677850051352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3108,
      "step": 2226
    },
    {
      "epoch": 0.07624101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2995,
      "step": 2227
    },
    {
      "epoch": 0.07627524820267031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3149,
      "step": 2228
    },
    {
      "epoch": 0.07630948305374871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3288,
      "step": 2229
    },
    {
      "epoch": 0.07634371790482711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3329,
      "step": 2230
    },
    {
      "epoch": 0.07637795275590552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3265,
      "step": 2231
    },
    {
      "epoch": 0.07641218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3218,
      "step": 2232
    },
    {
      "epoch": 0.07644642245806231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3076,
      "step": 2233
    },
    {
      "epoch": 0.0764806573091407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2805,
      "step": 2234
    },
    {
      "epoch": 0.0765148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3376,
      "step": 2235
    },
    {
      "epoch": 0.0765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2852,
      "step": 2236
    },
    {
      "epoch": 0.0765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2858,
      "step": 2237
    },
    {
      "epoch": 0.07661759671345429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.281,
      "step": 2238
    },
    {
      "epoch": 0.07665183156453269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3026,
      "step": 2239
    },
    {
      "epoch": 0.0766860664156111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.306,
      "step": 2240
    },
    {
      "epoch": 0.0767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2933,
      "step": 2241
    },
    {
      "epoch": 0.07675453611776789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2921,
      "step": 2242
    },
    {
      "epoch": 0.07678877096884629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3056,
      "step": 2243
    },
    {
      "epoch": 0.07682300581992468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3079,
      "step": 2244
    },
    {
      "epoch": 0.07685724067100308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3253,
      "step": 2245
    },
    {
      "epoch": 0.07689147552208148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3206,
      "step": 2246
    },
    {
      "epoch": 0.07692571037315987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2844,
      "step": 2247
    },
    {
      "epoch": 0.07695994522423827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2976,
      "step": 2248
    },
    {
      "epoch": 0.07699418007531668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.285,
      "step": 2249
    },
    {
      "epoch": 0.07702841492639508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.309,
      "step": 2250
    },
    {
      "epoch": 0.07706264977747347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3159,
      "step": 2251
    },
    {
      "epoch": 0.07709688462855187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3514,
      "step": 2252
    },
    {
      "epoch": 0.07713111947963026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3619,
      "step": 2253
    },
    {
      "epoch": 0.07716535433070866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.312,
      "step": 2254
    },
    {
      "epoch": 0.07719958918178706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2983,
      "step": 2255
    },
    {
      "epoch": 0.07723382403286545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3258,
      "step": 2256
    },
    {
      "epoch": 0.07726805888394385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.335,
      "step": 2257
    },
    {
      "epoch": 0.07730229373502226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3143,
      "step": 2258
    },
    {
      "epoch": 0.07733652858610066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2996,
      "step": 2259
    },
    {
      "epoch": 0.07737076343717905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2796,
      "step": 2260
    },
    {
      "epoch": 0.07740499828825745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2971,
      "step": 2261
    },
    {
      "epoch": 0.07743923313933584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2871,
      "step": 2262
    },
    {
      "epoch": 0.07747346799041424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.302,
      "step": 2263
    },
    {
      "epoch": 0.07750770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2751,
      "step": 2264
    },
    {
      "epoch": 0.07754193769257103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.292,
      "step": 2265
    },
    {
      "epoch": 0.07757617254364943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2686,
      "step": 2266
    },
    {
      "epoch": 0.07761040739472783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3037,
      "step": 2267
    },
    {
      "epoch": 0.07764464224580624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.307,
      "step": 2268
    },
    {
      "epoch": 0.07767887709688463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3248,
      "step": 2269
    },
    {
      "epoch": 0.07771311194796303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2764,
      "step": 2270
    },
    {
      "epoch": 0.07774734679904143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3284,
      "step": 2271
    },
    {
      "epoch": 0.07778158165011982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.291,
      "step": 2272
    },
    {
      "epoch": 0.07781581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3051,
      "step": 2273
    },
    {
      "epoch": 0.07785005135227661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2927,
      "step": 2274
    },
    {
      "epoch": 0.07788428620335501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2992,
      "step": 2275
    },
    {
      "epoch": 0.0779185210544334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3009,
      "step": 2276
    },
    {
      "epoch": 0.07795275590551182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2935,
      "step": 2277
    },
    {
      "epoch": 0.07798699075659021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2744,
      "step": 2278
    },
    {
      "epoch": 0.07802122560766861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2755,
      "step": 2279
    },
    {
      "epoch": 0.078055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2776,
      "step": 2280
    },
    {
      "epoch": 0.0780896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2761,
      "step": 2281
    },
    {
      "epoch": 0.0781239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3113,
      "step": 2282
    },
    {
      "epoch": 0.0781581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2816,
      "step": 2283
    },
    {
      "epoch": 0.07819239986306059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3005,
      "step": 2284
    },
    {
      "epoch": 0.07822663471413899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2856,
      "step": 2285
    },
    {
      "epoch": 0.0782608695652174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3046,
      "step": 2286
    },
    {
      "epoch": 0.0782951044162958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.304,
      "step": 2287
    },
    {
      "epoch": 0.07832933926737419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2938,
      "step": 2288
    },
    {
      "epoch": 0.07836357411845259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2821,
      "step": 2289
    },
    {
      "epoch": 0.07839780896953098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3032,
      "step": 2290
    },
    {
      "epoch": 0.07843204382060938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.29,
      "step": 2291
    },
    {
      "epoch": 0.07846627867168778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2537,
      "step": 2292
    },
    {
      "epoch": 0.07850051352276617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.305,
      "step": 2293
    },
    {
      "epoch": 0.07853474837384457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2669,
      "step": 2294
    },
    {
      "epoch": 0.07856898322492298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3013,
      "step": 2295
    },
    {
      "epoch": 0.07860321807600137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2933,
      "step": 2296
    },
    {
      "epoch": 0.07863745292707977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2895,
      "step": 2297
    },
    {
      "epoch": 0.07867168777815817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2662,
      "step": 2298
    },
    {
      "epoch": 0.07870592262923656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2474,
      "step": 2299
    },
    {
      "epoch": 0.07874015748031496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2706,
      "step": 2300
    },
    {
      "epoch": 0.07877439233139336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2541,
      "step": 2301
    },
    {
      "epoch": 0.07880862718247175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2726,
      "step": 2302
    },
    {
      "epoch": 0.07884286203355015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3088,
      "step": 2303
    },
    {
      "epoch": 0.07887709688462856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3044,
      "step": 2304
    },
    {
      "epoch": 0.07891133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2662,
      "step": 2305
    },
    {
      "epoch": 0.07894556658678535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.272,
      "step": 2306
    },
    {
      "epoch": 0.07897980143786375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2926,
      "step": 2307
    },
    {
      "epoch": 0.07901403628894214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3165,
      "step": 2308
    },
    {
      "epoch": 0.07904827114002054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2811,
      "step": 2309
    },
    {
      "epoch": 0.07908250599109894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2681,
      "step": 2310
    },
    {
      "epoch": 0.07911674084217733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2663,
      "step": 2311
    },
    {
      "epoch": 0.07915097569325573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3021,
      "step": 2312
    },
    {
      "epoch": 0.07918521054433413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2975,
      "step": 2313
    },
    {
      "epoch": 0.07921944539541254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2822,
      "step": 2314
    },
    {
      "epoch": 0.07925368024649093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2708,
      "step": 2315
    },
    {
      "epoch": 0.07928791509756933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2794,
      "step": 2316
    },
    {
      "epoch": 0.07932214994864772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2756,
      "step": 2317
    },
    {
      "epoch": 0.07935638479972612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2974,
      "step": 2318
    },
    {
      "epoch": 0.07939061965080452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2772,
      "step": 2319
    },
    {
      "epoch": 0.07942485450188291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3019,
      "step": 2320
    },
    {
      "epoch": 0.07945908935296131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2717,
      "step": 2321
    },
    {
      "epoch": 0.0794933242040397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2738,
      "step": 2322
    },
    {
      "epoch": 0.07952755905511812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2548,
      "step": 2323
    },
    {
      "epoch": 0.07956179390619651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.284,
      "step": 2324
    },
    {
      "epoch": 0.07959602875727491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2549,
      "step": 2325
    },
    {
      "epoch": 0.0796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2692,
      "step": 2326
    },
    {
      "epoch": 0.0796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2931,
      "step": 2327
    },
    {
      "epoch": 0.0796987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2521,
      "step": 2328
    },
    {
      "epoch": 0.0797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2671,
      "step": 2329
    },
    {
      "epoch": 0.07976720301266689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2729,
      "step": 2330
    },
    {
      "epoch": 0.07980143786374529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2643,
      "step": 2331
    },
    {
      "epoch": 0.0798356727148237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2928,
      "step": 2332
    },
    {
      "epoch": 0.07986990756590209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2775,
      "step": 2333
    },
    {
      "epoch": 0.07990414241698049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2723,
      "step": 2334
    },
    {
      "epoch": 0.07993837726805889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2672,
      "step": 2335
    },
    {
      "epoch": 0.07997261211913728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2664,
      "step": 2336
    },
    {
      "epoch": 0.08000684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2721,
      "step": 2337
    },
    {
      "epoch": 0.08004108182129407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2692,
      "step": 2338
    },
    {
      "epoch": 0.08007531667237247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2428,
      "step": 2339
    },
    {
      "epoch": 0.08010955152345087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2745,
      "step": 2340
    },
    {
      "epoch": 0.08014378637452928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2784,
      "step": 2341
    },
    {
      "epoch": 0.08017802122560767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2542,
      "step": 2342
    },
    {
      "epoch": 0.08021225607668607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2823,
      "step": 2343
    },
    {
      "epoch": 0.08024649092776447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2579,
      "step": 2344
    },
    {
      "epoch": 0.08028072577884286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2773,
      "step": 2345
    },
    {
      "epoch": 0.08031496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2581,
      "step": 2346
    },
    {
      "epoch": 0.08034919548099966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2855,
      "step": 2347
    },
    {
      "epoch": 0.08038343033207805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2681,
      "step": 2348
    },
    {
      "epoch": 0.08041766518315645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2533,
      "step": 2349
    },
    {
      "epoch": 0.08045190003423486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2669,
      "step": 2350
    },
    {
      "epoch": 0.08048613488531325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2539,
      "step": 2351
    },
    {
      "epoch": 0.08052036973639165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.271,
      "step": 2352
    },
    {
      "epoch": 0.08055460458747005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2398,
      "step": 2353
    },
    {
      "epoch": 0.08058883943854844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2708,
      "step": 2354
    },
    {
      "epoch": 0.08062307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2416,
      "step": 2355
    },
    {
      "epoch": 0.08065730914070524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2527,
      "step": 2356
    },
    {
      "epoch": 0.08069154399178363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2534,
      "step": 2357
    },
    {
      "epoch": 0.08072577884286203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2587,
      "step": 2358
    },
    {
      "epoch": 0.08076001369394042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2424,
      "step": 2359
    },
    {
      "epoch": 0.08079424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2771,
      "step": 2360
    },
    {
      "epoch": 0.08082848339609723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.246,
      "step": 2361
    },
    {
      "epoch": 0.08086271824717563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.232,
      "step": 2362
    },
    {
      "epoch": 0.08089695309825402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2776,
      "step": 2363
    },
    {
      "epoch": 0.08093118794933242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2543,
      "step": 2364
    },
    {
      "epoch": 0.08096542280041082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.279,
      "step": 2365
    },
    {
      "epoch": 0.08099965765148921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2568,
      "step": 2366
    },
    {
      "epoch": 0.08103389250256761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.287,
      "step": 2367
    },
    {
      "epoch": 0.081068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2565,
      "step": 2368
    },
    {
      "epoch": 0.08110236220472442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2682,
      "step": 2369
    },
    {
      "epoch": 0.08113659705580281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2662,
      "step": 2370
    },
    {
      "epoch": 0.08117083190688121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2676,
      "step": 2371
    },
    {
      "epoch": 0.0812050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2573,
      "step": 2372
    },
    {
      "epoch": 0.081239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2818,
      "step": 2373
    },
    {
      "epoch": 0.0812735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2503,
      "step": 2374
    },
    {
      "epoch": 0.0813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.263,
      "step": 2375
    },
    {
      "epoch": 0.08134200616227319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.263,
      "step": 2376
    },
    {
      "epoch": 0.08137624101335159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2592,
      "step": 2377
    },
    {
      "epoch": 0.08141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2279,
      "step": 2378
    },
    {
      "epoch": 0.08144471071550839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2449,
      "step": 2379
    },
    {
      "epoch": 0.08147894556658679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2687,
      "step": 2380
    },
    {
      "epoch": 0.08151318041766518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2579,
      "step": 2381
    },
    {
      "epoch": 0.08154741526874358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2447,
      "step": 2382
    },
    {
      "epoch": 0.08158165011982198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2373,
      "step": 2383
    },
    {
      "epoch": 0.08161588497090037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2339,
      "step": 2384
    },
    {
      "epoch": 0.08165011982197877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2339,
      "step": 2385
    },
    {
      "epoch": 0.08168435467305717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2313,
      "step": 2386
    },
    {
      "epoch": 0.08171858952413558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.257,
      "step": 2387
    },
    {
      "epoch": 0.08175282437521397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2701,
      "step": 2388
    },
    {
      "epoch": 0.08178705922629237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.254,
      "step": 2389
    },
    {
      "epoch": 0.08182129407737077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2756,
      "step": 2390
    },
    {
      "epoch": 0.08185552892844916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2669,
      "step": 2391
    },
    {
      "epoch": 0.08188976377952756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2411,
      "step": 2392
    },
    {
      "epoch": 0.08192399863060595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2522,
      "step": 2393
    },
    {
      "epoch": 0.08195823348168435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2455,
      "step": 2394
    },
    {
      "epoch": 0.08199246833276275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2644,
      "step": 2395
    },
    {
      "epoch": 0.08202670318384116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2941,
      "step": 2396
    },
    {
      "epoch": 0.08206093803491955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2578,
      "step": 2397
    },
    {
      "epoch": 0.08209517288599795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2488,
      "step": 2398
    },
    {
      "epoch": 0.08212940773707635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2456,
      "step": 2399
    },
    {
      "epoch": 0.08216364258815474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.258,
      "step": 2400
    },
    {
      "epoch": 0.08219787743923314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2479,
      "step": 2401
    },
    {
      "epoch": 0.08223211229031153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2526,
      "step": 2402
    },
    {
      "epoch": 0.08226634714138993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2533,
      "step": 2403
    },
    {
      "epoch": 0.08230058199246833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2369,
      "step": 2404
    },
    {
      "epoch": 0.08233481684354672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2523,
      "step": 2405
    },
    {
      "epoch": 0.08236905169462513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2317,
      "step": 2406
    },
    {
      "epoch": 0.08240328654570353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2462,
      "step": 2407
    },
    {
      "epoch": 0.08243752139678193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2433,
      "step": 2408
    },
    {
      "epoch": 0.08247175624786032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.258,
      "step": 2409
    },
    {
      "epoch": 0.08250599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2319,
      "step": 2410
    },
    {
      "epoch": 0.08254022595001712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2373,
      "step": 2411
    },
    {
      "epoch": 0.08257446080109551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2483,
      "step": 2412
    },
    {
      "epoch": 0.08260869565217391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2459,
      "step": 2413
    },
    {
      "epoch": 0.0826429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.249,
      "step": 2414
    },
    {
      "epoch": 0.08267716535433071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2385,
      "step": 2415
    },
    {
      "epoch": 0.08271140020540911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2326,
      "step": 2416
    },
    {
      "epoch": 0.08274563505648751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2584,
      "step": 2417
    },
    {
      "epoch": 0.0827798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2184,
      "step": 2418
    },
    {
      "epoch": 0.0828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2532,
      "step": 2419
    },
    {
      "epoch": 0.0828483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2574,
      "step": 2420
    },
    {
      "epoch": 0.08288257446080109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2847,
      "step": 2421
    },
    {
      "epoch": 0.08291680931187949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2321,
      "step": 2422
    },
    {
      "epoch": 0.08295104416295788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2414,
      "step": 2423
    },
    {
      "epoch": 0.0829852790140363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2571,
      "step": 2424
    },
    {
      "epoch": 0.08301951386511469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2128,
      "step": 2425
    },
    {
      "epoch": 0.08305374871619309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2334,
      "step": 2426
    },
    {
      "epoch": 0.08308798356727148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2331,
      "step": 2427
    },
    {
      "epoch": 0.08312221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2612,
      "step": 2428
    },
    {
      "epoch": 0.08315645326942828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2297,
      "step": 2429
    },
    {
      "epoch": 0.08319068812050667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2301,
      "step": 2430
    },
    {
      "epoch": 0.08322492297158507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2113,
      "step": 2431
    },
    {
      "epoch": 0.08325915782266347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2419,
      "step": 2432
    },
    {
      "epoch": 0.08329339267374188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2514,
      "step": 2433
    },
    {
      "epoch": 0.08332762752482027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2273,
      "step": 2434
    },
    {
      "epoch": 0.08336186237589867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2556,
      "step": 2435
    },
    {
      "epoch": 0.08339609722697706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2373,
      "step": 2436
    },
    {
      "epoch": 0.08343033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2358,
      "step": 2437
    },
    {
      "epoch": 0.08346456692913386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.197,
      "step": 2438
    },
    {
      "epoch": 0.08349880178021225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2358,
      "step": 2439
    },
    {
      "epoch": 0.08353303663129065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2143,
      "step": 2440
    },
    {
      "epoch": 0.08356727148236905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2553,
      "step": 2441
    },
    {
      "epoch": 0.08360150633344746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2334,
      "step": 2442
    },
    {
      "epoch": 0.08363574118452585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.233,
      "step": 2443
    },
    {
      "epoch": 0.08366997603560425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2541,
      "step": 2444
    },
    {
      "epoch": 0.08370421088668265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2271,
      "step": 2445
    },
    {
      "epoch": 0.08373844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.211,
      "step": 2446
    },
    {
      "epoch": 0.08377268058883944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2353,
      "step": 2447
    },
    {
      "epoch": 0.08380691543991783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2242,
      "step": 2448
    },
    {
      "epoch": 0.08384115029099623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2205,
      "step": 2449
    },
    {
      "epoch": 0.08387538514207463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2212,
      "step": 2450
    },
    {
      "epoch": 0.08390961999315302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2264,
      "step": 2451
    },
    {
      "epoch": 0.08394385484423143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.245,
      "step": 2452
    },
    {
      "epoch": 0.08397808969530983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2021,
      "step": 2453
    },
    {
      "epoch": 0.08401232454638823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2361,
      "step": 2454
    },
    {
      "epoch": 0.08404655939746662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2705,
      "step": 2455
    },
    {
      "epoch": 0.08408079424854502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2948,
      "step": 2456
    },
    {
      "epoch": 0.08411502909962341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2534,
      "step": 2457
    },
    {
      "epoch": 0.08414926395070181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2309,
      "step": 2458
    },
    {
      "epoch": 0.08418349880178021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2473,
      "step": 2459
    },
    {
      "epoch": 0.0842177336528586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2553,
      "step": 2460
    },
    {
      "epoch": 0.08425196850393701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2343,
      "step": 2461
    },
    {
      "epoch": 0.08428620335501541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2462,
      "step": 2462
    },
    {
      "epoch": 0.0843204382060938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2593,
      "step": 2463
    },
    {
      "epoch": 0.0843546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2312,
      "step": 2464
    },
    {
      "epoch": 0.0843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2274,
      "step": 2465
    },
    {
      "epoch": 0.084423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2542,
      "step": 2466
    },
    {
      "epoch": 0.08445737761040739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2561,
      "step": 2467
    },
    {
      "epoch": 0.08449161246148579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2328,
      "step": 2468
    },
    {
      "epoch": 0.08452584731256418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2111,
      "step": 2469
    },
    {
      "epoch": 0.0845600821636426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2378,
      "step": 2470
    },
    {
      "epoch": 0.08459431701472099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2301,
      "step": 2471
    },
    {
      "epoch": 0.08462855186579939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2453,
      "step": 2472
    },
    {
      "epoch": 0.08466278671687778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2069,
      "step": 2473
    },
    {
      "epoch": 0.08469702156795618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2086,
      "step": 2474
    },
    {
      "epoch": 0.08473125641903458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2176,
      "step": 2475
    },
    {
      "epoch": 0.08476549127011297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1958,
      "step": 2476
    },
    {
      "epoch": 0.08479972612119137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2104,
      "step": 2477
    },
    {
      "epoch": 0.08483396097226976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2357,
      "step": 2478
    },
    {
      "epoch": 0.08486819582334817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2333,
      "step": 2479
    },
    {
      "epoch": 0.08490243067442657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2215,
      "step": 2480
    },
    {
      "epoch": 0.08493666552550497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2063,
      "step": 2481
    },
    {
      "epoch": 0.08497090037658336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.213,
      "step": 2482
    },
    {
      "epoch": 0.08500513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2346,
      "step": 2483
    },
    {
      "epoch": 0.08503937007874016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2332,
      "step": 2484
    },
    {
      "epoch": 0.08507360492981855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2318,
      "step": 2485
    },
    {
      "epoch": 0.08510783978089695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2268,
      "step": 2486
    },
    {
      "epoch": 0.08514207463197535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2129,
      "step": 2487
    },
    {
      "epoch": 0.08517630948305376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2331,
      "step": 2488
    },
    {
      "epoch": 0.08521054433413215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2299,
      "step": 2489
    },
    {
      "epoch": 0.08524477918521055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2293,
      "step": 2490
    },
    {
      "epoch": 0.08527901403628894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2005,
      "step": 2491
    },
    {
      "epoch": 0.08531324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2384,
      "step": 2492
    },
    {
      "epoch": 0.08534748373844574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2303,
      "step": 2493
    },
    {
      "epoch": 0.08538171858952413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2216,
      "step": 2494
    },
    {
      "epoch": 0.08541595344060253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2033,
      "step": 2495
    },
    {
      "epoch": 0.08545018829168093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2226,
      "step": 2496
    },
    {
      "epoch": 0.08548442314275932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2085,
      "step": 2497
    },
    {
      "epoch": 0.08551865799383773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2061,
      "step": 2498
    },
    {
      "epoch": 0.08555289284491613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2023,
      "step": 2499
    },
    {
      "epoch": 0.08558712769599452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1985,
      "step": 2500
    },
    {
      "epoch": 0.08562136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1859,
      "step": 2501
    },
    {
      "epoch": 0.08565559739815132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2161,
      "step": 2502
    },
    {
      "epoch": 0.08568983224922971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.185,
      "step": 2503
    },
    {
      "epoch": 0.08572406710030811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1776,
      "step": 2504
    },
    {
      "epoch": 0.0857583019513865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2248,
      "step": 2505
    },
    {
      "epoch": 0.0857925368024649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2331,
      "step": 2506
    },
    {
      "epoch": 0.08582677165354331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2862,
      "step": 2507
    },
    {
      "epoch": 0.08586100650462171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.184,
      "step": 2508
    },
    {
      "epoch": 0.0858952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2359,
      "step": 2509
    },
    {
      "epoch": 0.0859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1836,
      "step": 2510
    },
    {
      "epoch": 0.0859637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2133,
      "step": 2511
    },
    {
      "epoch": 0.0859979459089353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.202,
      "step": 2512
    },
    {
      "epoch": 0.08603218076001369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2058,
      "step": 2513
    },
    {
      "epoch": 0.08606641561109209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2158,
      "step": 2514
    },
    {
      "epoch": 0.08610065046217048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2143,
      "step": 2515
    },
    {
      "epoch": 0.0861348853132489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1998,
      "step": 2516
    },
    {
      "epoch": 0.08616912016432729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2116,
      "step": 2517
    },
    {
      "epoch": 0.08620335501540569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1972,
      "step": 2518
    },
    {
      "epoch": 0.08623758986648408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2483,
      "step": 2519
    },
    {
      "epoch": 0.08627182471756248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2196,
      "step": 2520
    },
    {
      "epoch": 0.08630605956864087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2079,
      "step": 2521
    },
    {
      "epoch": 0.08634029441971927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2204,
      "step": 2522
    },
    {
      "epoch": 0.08637452927079767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2293,
      "step": 2523
    },
    {
      "epoch": 0.08640876412187606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2077,
      "step": 2524
    },
    {
      "epoch": 0.08644299897295447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2123,
      "step": 2525
    },
    {
      "epoch": 0.08647723382403287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2214,
      "step": 2526
    },
    {
      "epoch": 0.08651146867511127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2216,
      "step": 2527
    },
    {
      "epoch": 0.08654570352618966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2001,
      "step": 2528
    },
    {
      "epoch": 0.08657993837726806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1956,
      "step": 2529
    },
    {
      "epoch": 0.08661417322834646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2203,
      "step": 2530
    },
    {
      "epoch": 0.08664840807942485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2163,
      "step": 2531
    },
    {
      "epoch": 0.08668264293050325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2009,
      "step": 2532
    },
    {
      "epoch": 0.08671687778158164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1876,
      "step": 2533
    },
    {
      "epoch": 0.08675111263266005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2256,
      "step": 2534
    },
    {
      "epoch": 0.08678534748373845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2138,
      "step": 2535
    },
    {
      "epoch": 0.08681958233481685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1644,
      "step": 2536
    },
    {
      "epoch": 0.08685381718589524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2135,
      "step": 2537
    },
    {
      "epoch": 0.08688805203697364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1976,
      "step": 2538
    },
    {
      "epoch": 0.08692228688805204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2196,
      "step": 2539
    },
    {
      "epoch": 0.08695652173913043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2138,
      "step": 2540
    },
    {
      "epoch": 0.08699075659020883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2088,
      "step": 2541
    },
    {
      "epoch": 0.08702499144128722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2068,
      "step": 2542
    },
    {
      "epoch": 0.08705922629236564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2233,
      "step": 2543
    },
    {
      "epoch": 0.08709346114344403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2199,
      "step": 2544
    },
    {
      "epoch": 0.08712769599452243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2023,
      "step": 2545
    },
    {
      "epoch": 0.08716193084560082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2142,
      "step": 2546
    },
    {
      "epoch": 0.08719616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.184,
      "step": 2547
    },
    {
      "epoch": 0.08723040054775762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2119,
      "step": 2548
    },
    {
      "epoch": 0.08726463539883601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2105,
      "step": 2549
    },
    {
      "epoch": 0.08729887024991441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2395,
      "step": 2550
    },
    {
      "epoch": 0.0873331051009928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1849,
      "step": 2551
    },
    {
      "epoch": 0.0873673399520712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1952,
      "step": 2552
    },
    {
      "epoch": 0.08740157480314961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2,
      "step": 2553
    },
    {
      "epoch": 0.08743580965422801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1892,
      "step": 2554
    },
    {
      "epoch": 0.0874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2156,
      "step": 2555
    },
    {
      "epoch": 0.0875042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1978,
      "step": 2556
    },
    {
      "epoch": 0.0875385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1868,
      "step": 2557
    },
    {
      "epoch": 0.0875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1905,
      "step": 2558
    },
    {
      "epoch": 0.08760698390961999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.212,
      "step": 2559
    },
    {
      "epoch": 0.08764121876069839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2224,
      "step": 2560
    },
    {
      "epoch": 0.08767545361177678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1878,
      "step": 2561
    },
    {
      "epoch": 0.08770968846285519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1893,
      "step": 2562
    },
    {
      "epoch": 0.08774392331393359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2043,
      "step": 2563
    },
    {
      "epoch": 0.08777815816501199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2117,
      "step": 2564
    },
    {
      "epoch": 0.08781239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.196,
      "step": 2565
    },
    {
      "epoch": 0.08784662786716878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2132,
      "step": 2566
    },
    {
      "epoch": 0.08788086271824717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2117,
      "step": 2567
    },
    {
      "epoch": 0.08791509756932557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2053,
      "step": 2568
    },
    {
      "epoch": 0.08794933242040397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.19,
      "step": 2569
    },
    {
      "epoch": 0.08798356727148236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2045,
      "step": 2570
    },
    {
      "epoch": 0.08801780212256077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1959,
      "step": 2571
    },
    {
      "epoch": 0.08805203697363917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2208,
      "step": 2572
    },
    {
      "epoch": 0.08808627182471757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2035,
      "step": 2573
    },
    {
      "epoch": 0.08812050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1821,
      "step": 2574
    },
    {
      "epoch": 0.08815474152687436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1777,
      "step": 2575
    },
    {
      "epoch": 0.08818897637795275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1793,
      "step": 2576
    },
    {
      "epoch": 0.08822321122903115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1915,
      "step": 2577
    },
    {
      "epoch": 0.08825744608010955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1867,
      "step": 2578
    },
    {
      "epoch": 0.08829168093118794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1798,
      "step": 2579
    },
    {
      "epoch": 0.08832591578226635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.207,
      "step": 2580
    },
    {
      "epoch": 0.08836015063334475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1777,
      "step": 2581
    },
    {
      "epoch": 0.08839438548442315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1947,
      "step": 2582
    },
    {
      "epoch": 0.08842862033550154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2133,
      "step": 2583
    },
    {
      "epoch": 0.08846285518657994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1781,
      "step": 2584
    },
    {
      "epoch": 0.08849709003765834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2066,
      "step": 2585
    },
    {
      "epoch": 0.08853132488873673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1743,
      "step": 2586
    },
    {
      "epoch": 0.08856555973981513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1629,
      "step": 2587
    },
    {
      "epoch": 0.08859979459089352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1607,
      "step": 2588
    },
    {
      "epoch": 0.08863402944197193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1646,
      "step": 2589
    },
    {
      "epoch": 0.08866826429305033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1733,
      "step": 2590
    },
    {
      "epoch": 0.08870249914412873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1864,
      "step": 2591
    },
    {
      "epoch": 0.08873673399520712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.191,
      "step": 2592
    },
    {
      "epoch": 0.08877096884628552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1883,
      "step": 2593
    },
    {
      "epoch": 0.08880520369736392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1759,
      "step": 2594
    },
    {
      "epoch": 0.08883943854844231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1627,
      "step": 2595
    },
    {
      "epoch": 0.08887367339952071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1745,
      "step": 2596
    },
    {
      "epoch": 0.0889079082505991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.209,
      "step": 2597
    },
    {
      "epoch": 0.0889421431016775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1999,
      "step": 2598
    },
    {
      "epoch": 0.08897637795275591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2044,
      "step": 2599
    },
    {
      "epoch": 0.08901061280383431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2001,
      "step": 2600
    },
    {
      "epoch": 0.0890448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2152,
      "step": 2601
    },
    {
      "epoch": 0.0890790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1981,
      "step": 2602
    },
    {
      "epoch": 0.0891133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1801,
      "step": 2603
    },
    {
      "epoch": 0.08914755220814789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1954,
      "step": 2604
    },
    {
      "epoch": 0.08918178705922629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1621,
      "step": 2605
    },
    {
      "epoch": 0.08921602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1856,
      "step": 2606
    },
    {
      "epoch": 0.08925025676138308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1884,
      "step": 2607
    },
    {
      "epoch": 0.08928449161246149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1943,
      "step": 2608
    },
    {
      "epoch": 0.08931872646353989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1692,
      "step": 2609
    },
    {
      "epoch": 0.08935296131461828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1878,
      "step": 2610
    },
    {
      "epoch": 0.08938719616569668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2111,
      "step": 2611
    },
    {
      "epoch": 0.08942143101677508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2056,
      "step": 2612
    },
    {
      "epoch": 0.08945566586785347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.212,
      "step": 2613
    },
    {
      "epoch": 0.08948990071893187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1536,
      "step": 2614
    },
    {
      "epoch": 0.08952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1968,
      "step": 2615
    },
    {
      "epoch": 0.08955837042108866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1879,
      "step": 2616
    },
    {
      "epoch": 0.08959260527216707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1736,
      "step": 2617
    },
    {
      "epoch": 0.08962684012324547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1974,
      "step": 2618
    },
    {
      "epoch": 0.08966107497432386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.21,
      "step": 2619
    },
    {
      "epoch": 0.08969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1888,
      "step": 2620
    },
    {
      "epoch": 0.08972954467648066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1948,
      "step": 2621
    },
    {
      "epoch": 0.08976377952755905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1945,
      "step": 2622
    },
    {
      "epoch": 0.08979801437863745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1869,
      "step": 2623
    },
    {
      "epoch": 0.08983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1762,
      "step": 2624
    },
    {
      "epoch": 0.08986648408079424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2026,
      "step": 2625
    },
    {
      "epoch": 0.08990071893187265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1823,
      "step": 2626
    },
    {
      "epoch": 0.08993495378295105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1767,
      "step": 2627
    },
    {
      "epoch": 0.08996918863402945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1699,
      "step": 2628
    },
    {
      "epoch": 0.09000342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1772,
      "step": 2629
    },
    {
      "epoch": 0.09003765833618624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2088,
      "step": 2630
    },
    {
      "epoch": 0.09007189318726463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.173,
      "step": 2631
    },
    {
      "epoch": 0.09010612803834303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1804,
      "step": 2632
    },
    {
      "epoch": 0.09014036288942143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1664,
      "step": 2633
    },
    {
      "epoch": 0.09017459774049982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.196,
      "step": 2634
    },
    {
      "epoch": 0.09020883259157823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1824,
      "step": 2635
    },
    {
      "epoch": 0.09024306744265663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.176,
      "step": 2636
    },
    {
      "epoch": 0.09027730229373503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1716,
      "step": 2637
    },
    {
      "epoch": 0.09031153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1843,
      "step": 2638
    },
    {
      "epoch": 0.09034577199589182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1664,
      "step": 2639
    },
    {
      "epoch": 0.09038000684697022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1741,
      "step": 2640
    },
    {
      "epoch": 0.09041424169804861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1989,
      "step": 2641
    },
    {
      "epoch": 0.09044847654912701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1571,
      "step": 2642
    },
    {
      "epoch": 0.0904827114002054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.182,
      "step": 2643
    },
    {
      "epoch": 0.0905169462512838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1762,
      "step": 2644
    },
    {
      "epoch": 0.09055118110236221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1938,
      "step": 2645
    },
    {
      "epoch": 0.0905854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1609,
      "step": 2646
    },
    {
      "epoch": 0.090619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1923,
      "step": 2647
    },
    {
      "epoch": 0.0906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1865,
      "step": 2648
    },
    {
      "epoch": 0.0906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1703,
      "step": 2649
    },
    {
      "epoch": 0.09072235535775419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1772,
      "step": 2650
    },
    {
      "epoch": 0.09075659020883259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1738,
      "step": 2651
    },
    {
      "epoch": 0.09079082505991098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1992,
      "step": 2652
    },
    {
      "epoch": 0.09082505991098938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1784,
      "step": 2653
    },
    {
      "epoch": 0.09085929476206779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1647,
      "step": 2654
    },
    {
      "epoch": 0.09089352961314619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1881,
      "step": 2655
    },
    {
      "epoch": 0.09092776446422458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2056,
      "step": 2656
    },
    {
      "epoch": 0.09096199931530298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1898,
      "step": 2657
    },
    {
      "epoch": 0.09099623416638138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1559,
      "step": 2658
    },
    {
      "epoch": 0.09103046901745977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1803,
      "step": 2659
    },
    {
      "epoch": 0.09106470386853817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2066,
      "step": 2660
    },
    {
      "epoch": 0.09109893871961657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1774,
      "step": 2661
    },
    {
      "epoch": 0.09113317357069496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1479,
      "step": 2662
    },
    {
      "epoch": 0.09116740842177337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1896,
      "step": 2663
    },
    {
      "epoch": 0.09120164327285177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1859,
      "step": 2664
    },
    {
      "epoch": 0.09123587812393016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1727,
      "step": 2665
    },
    {
      "epoch": 0.09127011297500856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1703,
      "step": 2666
    },
    {
      "epoch": 0.09130434782608696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1615,
      "step": 2667
    },
    {
      "epoch": 0.09133858267716535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1335,
      "step": 2668
    },
    {
      "epoch": 0.09137281752824375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1799,
      "step": 2669
    },
    {
      "epoch": 0.09140705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1686,
      "step": 2670
    },
    {
      "epoch": 0.09144128723040054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1822,
      "step": 2671
    },
    {
      "epoch": 0.09147552208147895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1933,
      "step": 2672
    },
    {
      "epoch": 0.09150975693255735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1387,
      "step": 2673
    },
    {
      "epoch": 0.09154399178363574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1886,
      "step": 2674
    },
    {
      "epoch": 0.09157822663471414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1668,
      "step": 2675
    },
    {
      "epoch": 0.09161246148579254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.152,
      "step": 2676
    },
    {
      "epoch": 0.09164669633687093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1604,
      "step": 2677
    },
    {
      "epoch": 0.09168093118794933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1675,
      "step": 2678
    },
    {
      "epoch": 0.09171516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1574,
      "step": 2679
    },
    {
      "epoch": 0.09174940089010612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1443,
      "step": 2680
    },
    {
      "epoch": 0.09178363574118453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1586,
      "step": 2681
    },
    {
      "epoch": 0.09181787059226293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1702,
      "step": 2682
    },
    {
      "epoch": 0.09185210544334133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1706,
      "step": 2683
    },
    {
      "epoch": 0.09188634029441972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1769,
      "step": 2684
    },
    {
      "epoch": 0.09192057514549812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1613,
      "step": 2685
    },
    {
      "epoch": 0.09195480999657651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.15,
      "step": 2686
    },
    {
      "epoch": 0.09198904484765491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1337,
      "step": 2687
    },
    {
      "epoch": 0.0920232796987333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1689,
      "step": 2688
    },
    {
      "epoch": 0.0920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1805,
      "step": 2689
    },
    {
      "epoch": 0.0920917494008901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1649,
      "step": 2690
    },
    {
      "epoch": 0.09212598425196851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1523,
      "step": 2691
    },
    {
      "epoch": 0.0921602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1677,
      "step": 2692
    },
    {
      "epoch": 0.0921944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1526,
      "step": 2693
    },
    {
      "epoch": 0.0922286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.211,
      "step": 2694
    },
    {
      "epoch": 0.0922629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.167,
      "step": 2695
    },
    {
      "epoch": 0.09229715850736049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1525,
      "step": 2696
    },
    {
      "epoch": 0.09233139335843889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1643,
      "step": 2697
    },
    {
      "epoch": 0.09236562820951728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1659,
      "step": 2698
    },
    {
      "epoch": 0.09239986306059568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.142,
      "step": 2699
    },
    {
      "epoch": 0.09243409791167409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1697,
      "step": 2700
    },
    {
      "epoch": 0.09246833276275249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1647,
      "step": 2701
    },
    {
      "epoch": 0.09250256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1814,
      "step": 2702
    },
    {
      "epoch": 0.09253680246490928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1666,
      "step": 2703
    },
    {
      "epoch": 0.09257103731598768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1736,
      "step": 2704
    },
    {
      "epoch": 0.09260527216706607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1632,
      "step": 2705
    },
    {
      "epoch": 0.09263950701814447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1832,
      "step": 2706
    },
    {
      "epoch": 0.09267374186922286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1625,
      "step": 2707
    },
    {
      "epoch": 0.09270797672030126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1573,
      "step": 2708
    },
    {
      "epoch": 0.09274221157137967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1677,
      "step": 2709
    },
    {
      "epoch": 0.09277644642245807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1625,
      "step": 2710
    },
    {
      "epoch": 0.09281068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1802,
      "step": 2711
    },
    {
      "epoch": 0.09284491612461486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1543,
      "step": 2712
    },
    {
      "epoch": 0.09287915097569326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1523,
      "step": 2713
    },
    {
      "epoch": 0.09291338582677165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1367,
      "step": 2714
    },
    {
      "epoch": 0.09294762067785005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.149,
      "step": 2715
    },
    {
      "epoch": 0.09298185552892844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1619,
      "step": 2716
    },
    {
      "epoch": 0.09301609038000684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1426,
      "step": 2717
    },
    {
      "epoch": 0.09305032523108525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1574,
      "step": 2718
    },
    {
      "epoch": 0.09308456008216365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.139,
      "step": 2719
    },
    {
      "epoch": 0.09311879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1356,
      "step": 2720
    },
    {
      "epoch": 0.09315302978432044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1567,
      "step": 2721
    },
    {
      "epoch": 0.09318726463539884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1821,
      "step": 2722
    },
    {
      "epoch": 0.09322149948647723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1448,
      "step": 2723
    },
    {
      "epoch": 0.09325573433755563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.129,
      "step": 2724
    },
    {
      "epoch": 0.09328996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1921,
      "step": 2725
    },
    {
      "epoch": 0.09332420403971242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.184,
      "step": 2726
    },
    {
      "epoch": 0.09335843889079083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1594,
      "step": 2727
    },
    {
      "epoch": 0.09339267374186923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1504,
      "step": 2728
    },
    {
      "epoch": 0.09342690859294762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1543,
      "step": 2729
    },
    {
      "epoch": 0.09346114344402602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1382,
      "step": 2730
    },
    {
      "epoch": 0.09349537829510442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1436,
      "step": 2731
    },
    {
      "epoch": 0.09352961314618281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1476,
      "step": 2732
    },
    {
      "epoch": 0.09356384799726121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1434,
      "step": 2733
    },
    {
      "epoch": 0.0935980828483396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1494,
      "step": 2734
    },
    {
      "epoch": 0.093632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1506,
      "step": 2735
    },
    {
      "epoch": 0.0936665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1645,
      "step": 2736
    },
    {
      "epoch": 0.09370078740157481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1782,
      "step": 2737
    },
    {
      "epoch": 0.0937350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1464,
      "step": 2738
    },
    {
      "epoch": 0.0937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1674,
      "step": 2739
    },
    {
      "epoch": 0.09380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1355,
      "step": 2740
    },
    {
      "epoch": 0.0938377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1579,
      "step": 2741
    },
    {
      "epoch": 0.09387196165696679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1482,
      "step": 2742
    },
    {
      "epoch": 0.09390619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1482,
      "step": 2743
    },
    {
      "epoch": 0.09394043135912358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1432,
      "step": 2744
    },
    {
      "epoch": 0.09397466621020198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1366,
      "step": 2745
    },
    {
      "epoch": 0.09400890106128039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.154,
      "step": 2746
    },
    {
      "epoch": 0.09404313591235879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.149,
      "step": 2747
    },
    {
      "epoch": 0.09407737076343718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1324,
      "step": 2748
    },
    {
      "epoch": 0.09411160561451558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.139,
      "step": 2749
    },
    {
      "epoch": 0.09414584046559397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1334,
      "step": 2750
    },
    {
      "epoch": 0.09418007531667237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1739,
      "step": 2751
    },
    {
      "epoch": 0.09421431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1508,
      "step": 2752
    },
    {
      "epoch": 0.09424854501882916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1559,
      "step": 2753
    },
    {
      "epoch": 0.09428277986990756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1373,
      "step": 2754
    },
    {
      "epoch": 0.09431701472098597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1412,
      "step": 2755
    },
    {
      "epoch": 0.09435124957206437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1345,
      "step": 2756
    },
    {
      "epoch": 0.09438548442314276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1255,
      "step": 2757
    },
    {
      "epoch": 0.09441971927422116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1463,
      "step": 2758
    },
    {
      "epoch": 0.09445395412529956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.142,
      "step": 2759
    },
    {
      "epoch": 0.09448818897637795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1427,
      "step": 2760
    },
    {
      "epoch": 0.09452242382745635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1501,
      "step": 2761
    },
    {
      "epoch": 0.09455665867853474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1647,
      "step": 2762
    },
    {
      "epoch": 0.09459089352961314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1281,
      "step": 2763
    },
    {
      "epoch": 0.09462512838069155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1489,
      "step": 2764
    },
    {
      "epoch": 0.09465936323176995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1315,
      "step": 2765
    },
    {
      "epoch": 0.09469359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1098,
      "step": 2766
    },
    {
      "epoch": 0.09472783293392674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1501,
      "step": 2767
    },
    {
      "epoch": 0.09476206778500514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1408,
      "step": 2768
    },
    {
      "epoch": 0.09479630263608353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1533,
      "step": 2769
    },
    {
      "epoch": 0.09483053748716193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1386,
      "step": 2770
    },
    {
      "epoch": 0.09486477233824032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1404,
      "step": 2771
    },
    {
      "epoch": 0.09489900718931872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1267,
      "step": 2772
    },
    {
      "epoch": 0.09493324204039713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1608,
      "step": 2773
    },
    {
      "epoch": 0.09496747689147553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.15,
      "step": 2774
    },
    {
      "epoch": 0.09500171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1344,
      "step": 2775
    },
    {
      "epoch": 0.09503594659363232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1512,
      "step": 2776
    },
    {
      "epoch": 0.09507018144471072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1621,
      "step": 2777
    },
    {
      "epoch": 0.09510441629578911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1291,
      "step": 2778
    },
    {
      "epoch": 0.09513865114686751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.139,
      "step": 2779
    },
    {
      "epoch": 0.0951728859979459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1445,
      "step": 2780
    },
    {
      "epoch": 0.0952071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1341,
      "step": 2781
    },
    {
      "epoch": 0.0952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1645,
      "step": 2782
    },
    {
      "epoch": 0.09527559055118111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1216,
      "step": 2783
    },
    {
      "epoch": 0.0953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1334,
      "step": 2784
    },
    {
      "epoch": 0.0953440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1145,
      "step": 2785
    },
    {
      "epoch": 0.0953782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1188,
      "step": 2786
    },
    {
      "epoch": 0.0954125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1554,
      "step": 2787
    },
    {
      "epoch": 0.09544676480657309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1374,
      "step": 2788
    },
    {
      "epoch": 0.09548099965765149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1277,
      "step": 2789
    },
    {
      "epoch": 0.09551523450872988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1286,
      "step": 2790
    },
    {
      "epoch": 0.09554946935980828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1476,
      "step": 2791
    },
    {
      "epoch": 0.09558370421088669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1473,
      "step": 2792
    },
    {
      "epoch": 0.09561793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1306,
      "step": 2793
    },
    {
      "epoch": 0.09565217391304348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.154,
      "step": 2794
    },
    {
      "epoch": 0.09568640876412188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1221,
      "step": 2795
    },
    {
      "epoch": 0.09572064361520027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1256,
      "step": 2796
    },
    {
      "epoch": 0.09575487846627867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1028,
      "step": 2797
    },
    {
      "epoch": 0.09578911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1302,
      "step": 2798
    },
    {
      "epoch": 0.09582334816843546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1272,
      "step": 2799
    },
    {
      "epoch": 0.09585758301951386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1365,
      "step": 2800
    },
    {
      "epoch": 0.09589181787059227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1382,
      "step": 2801
    },
    {
      "epoch": 0.09592605272167067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1522,
      "step": 2802
    },
    {
      "epoch": 0.09596028757274906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1401,
      "step": 2803
    },
    {
      "epoch": 0.09599452242382746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1197,
      "step": 2804
    },
    {
      "epoch": 0.09602875727490585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1304,
      "step": 2805
    },
    {
      "epoch": 0.09606299212598425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.128,
      "step": 2806
    },
    {
      "epoch": 0.09609722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1265,
      "step": 2807
    },
    {
      "epoch": 0.09613146182814104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1586,
      "step": 2808
    },
    {
      "epoch": 0.09616569667921944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1532,
      "step": 2809
    },
    {
      "epoch": 0.09619993153029785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1394,
      "step": 2810
    },
    {
      "epoch": 0.09623416638137625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1327,
      "step": 2811
    },
    {
      "epoch": 0.09626840123245464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1285,
      "step": 2812
    },
    {
      "epoch": 0.09630263608353304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.139,
      "step": 2813
    },
    {
      "epoch": 0.09633687093461143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.121,
      "step": 2814
    },
    {
      "epoch": 0.09637110578568983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1371,
      "step": 2815
    },
    {
      "epoch": 0.09640534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1511,
      "step": 2816
    },
    {
      "epoch": 0.09643957548784662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1474,
      "step": 2817
    },
    {
      "epoch": 0.09647381033892502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1617,
      "step": 2818
    },
    {
      "epoch": 0.09650804519000343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1631,
      "step": 2819
    },
    {
      "epoch": 0.09654228004108183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1689,
      "step": 2820
    },
    {
      "epoch": 0.09657651489216022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1581,
      "step": 2821
    },
    {
      "epoch": 0.09661074974323862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1454,
      "step": 2822
    },
    {
      "epoch": 0.09664498459431702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.129,
      "step": 2823
    },
    {
      "epoch": 0.09667921944539541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1174,
      "step": 2824
    },
    {
      "epoch": 0.09671345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1349,
      "step": 2825
    },
    {
      "epoch": 0.0967476891475522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0636,
      "step": 2826
    },
    {
      "epoch": 0.0967819239986306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1224,
      "step": 2827
    },
    {
      "epoch": 0.096816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1195,
      "step": 2828
    },
    {
      "epoch": 0.09685039370078741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0798,
      "step": 2829
    },
    {
      "epoch": 0.0968846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1225,
      "step": 2830
    },
    {
      "epoch": 0.0969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1107,
      "step": 2831
    },
    {
      "epoch": 0.0969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1197,
      "step": 2832
    },
    {
      "epoch": 0.09698733310510099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1189,
      "step": 2833
    },
    {
      "epoch": 0.09702156795617939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1072,
      "step": 2834
    },
    {
      "epoch": 0.09705580280725778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1264,
      "step": 2835
    },
    {
      "epoch": 0.09709003765833618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1281,
      "step": 2836
    },
    {
      "epoch": 0.09712427250941458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1288,
      "step": 2837
    },
    {
      "epoch": 0.09715850736049299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1542,
      "step": 2838
    },
    {
      "epoch": 0.09719274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1196,
      "step": 2839
    },
    {
      "epoch": 0.09722697706264978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1191,
      "step": 2840
    },
    {
      "epoch": 0.09726121191372818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1207,
      "step": 2841
    },
    {
      "epoch": 0.09729544676480657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1368,
      "step": 2842
    },
    {
      "epoch": 0.09732968161588497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1394,
      "step": 2843
    },
    {
      "epoch": 0.09736391646696337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1413,
      "step": 2844
    },
    {
      "epoch": 0.09739815131804176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1303,
      "step": 2845
    },
    {
      "epoch": 0.09743238616912016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1308,
      "step": 2846
    },
    {
      "epoch": 0.09746662102019857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1233,
      "step": 2847
    },
    {
      "epoch": 0.09750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1147,
      "step": 2848
    },
    {
      "epoch": 0.09753509072235536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1193,
      "step": 2849
    },
    {
      "epoch": 0.09756932557343376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1247,
      "step": 2850
    },
    {
      "epoch": 0.09760356042451215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1265,
      "step": 2851
    },
    {
      "epoch": 0.09763779527559055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1239,
      "step": 2852
    },
    {
      "epoch": 0.09767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1338,
      "step": 2853
    },
    {
      "epoch": 0.09770626497774734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1131,
      "step": 2854
    },
    {
      "epoch": 0.09774049982882574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1473,
      "step": 2855
    },
    {
      "epoch": 0.09777473467990415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0927,
      "step": 2856
    },
    {
      "epoch": 0.09780896953098255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1581,
      "step": 2857
    },
    {
      "epoch": 0.09784320438206094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1271,
      "step": 2858
    },
    {
      "epoch": 0.09787743923313934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1322,
      "step": 2859
    },
    {
      "epoch": 0.09791167408421773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1059,
      "step": 2860
    },
    {
      "epoch": 0.09794590893529613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1104,
      "step": 2861
    },
    {
      "epoch": 0.09798014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1264,
      "step": 2862
    },
    {
      "epoch": 0.09801437863745292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1111,
      "step": 2863
    },
    {
      "epoch": 0.09804861348853132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1259,
      "step": 2864
    },
    {
      "epoch": 0.09808284833960973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1094,
      "step": 2865
    },
    {
      "epoch": 0.09811708319068813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.113,
      "step": 2866
    },
    {
      "epoch": 0.09815131804176652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.12,
      "step": 2867
    },
    {
      "epoch": 0.09818555289284492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1293,
      "step": 2868
    },
    {
      "epoch": 0.09821978774392331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1433,
      "step": 2869
    },
    {
      "epoch": 0.09825402259500171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1196,
      "step": 2870
    },
    {
      "epoch": 0.09828825744608011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1427,
      "step": 2871
    },
    {
      "epoch": 0.0983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1278,
      "step": 2872
    },
    {
      "epoch": 0.0983567271482369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1701,
      "step": 2873
    },
    {
      "epoch": 0.0983909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1306,
      "step": 2874
    },
    {
      "epoch": 0.0984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1004,
      "step": 2875
    },
    {
      "epoch": 0.0984594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0912,
      "step": 2876
    },
    {
      "epoch": 0.0984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1088,
      "step": 2877
    },
    {
      "epoch": 0.0985279014036289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1332,
      "step": 2878
    },
    {
      "epoch": 0.09856213625470729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1272,
      "step": 2879
    },
    {
      "epoch": 0.09859637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0945,
      "step": 2880
    },
    {
      "epoch": 0.09863060595686408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1202,
      "step": 2881
    },
    {
      "epoch": 0.09866484080794248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1037,
      "step": 2882
    },
    {
      "epoch": 0.09869907565902088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.122,
      "step": 2883
    },
    {
      "epoch": 0.09873331051009929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1237,
      "step": 2884
    },
    {
      "epoch": 0.09876754536117768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1185,
      "step": 2885
    },
    {
      "epoch": 0.09880178021225608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.117,
      "step": 2886
    },
    {
      "epoch": 0.09883601506333448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1118,
      "step": 2887
    },
    {
      "epoch": 0.09887024991441287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1394,
      "step": 2888
    },
    {
      "epoch": 0.09890448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1212,
      "step": 2889
    },
    {
      "epoch": 0.09893871961656966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1227,
      "step": 2890
    },
    {
      "epoch": 0.09897295446764806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1206,
      "step": 2891
    },
    {
      "epoch": 0.09900718931872646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1368,
      "step": 2892
    },
    {
      "epoch": 0.09904142416980487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1172,
      "step": 2893
    },
    {
      "epoch": 0.09907565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0999,
      "step": 2894
    },
    {
      "epoch": 0.09910989387196166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1314,
      "step": 2895
    },
    {
      "epoch": 0.09914412872304006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1112,
      "step": 2896
    },
    {
      "epoch": 0.09917836357411845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1111,
      "step": 2897
    },
    {
      "epoch": 0.09921259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1283,
      "step": 2898
    },
    {
      "epoch": 0.09924683327627525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0915,
      "step": 2899
    },
    {
      "epoch": 0.09928106812735364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1177,
      "step": 2900
    },
    {
      "epoch": 0.09931530297843204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1283,
      "step": 2901
    },
    {
      "epoch": 0.09934953782951045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.142,
      "step": 2902
    },
    {
      "epoch": 0.09938377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1393,
      "step": 2903
    },
    {
      "epoch": 0.09941800753166724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1234,
      "step": 2904
    },
    {
      "epoch": 0.09945224238274564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1013,
      "step": 2905
    },
    {
      "epoch": 0.09948647723382403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1088,
      "step": 2906
    },
    {
      "epoch": 0.09952071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.104,
      "step": 2907
    },
    {
      "epoch": 0.09955494693598083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0942,
      "step": 2908
    },
    {
      "epoch": 0.09958918178705922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1012,
      "step": 2909
    },
    {
      "epoch": 0.09962341663813762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1159,
      "step": 2910
    },
    {
      "epoch": 0.09965765148921603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1121,
      "step": 2911
    },
    {
      "epoch": 0.09969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1223,
      "step": 2912
    },
    {
      "epoch": 0.09972612119137282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1034,
      "step": 2913
    },
    {
      "epoch": 0.09976035604245122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1233,
      "step": 2914
    },
    {
      "epoch": 0.09979459089352961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0888,
      "step": 2915
    },
    {
      "epoch": 0.09982882574460801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1114,
      "step": 2916
    },
    {
      "epoch": 0.0998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0982,
      "step": 2917
    },
    {
      "epoch": 0.0998972954467648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1022,
      "step": 2918
    },
    {
      "epoch": 0.0999315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1048,
      "step": 2919
    },
    {
      "epoch": 0.0999657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1305,
      "step": 2920
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1144,
      "step": 2921
    },
    {
      "epoch": 0.1000342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1036,
      "step": 2922
    },
    {
      "epoch": 0.1000684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1032,
      "step": 2923
    },
    {
      "epoch": 0.1001027045532352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1011,
      "step": 2924
    },
    {
      "epoch": 0.10013693940431359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1233,
      "step": 2925
    },
    {
      "epoch": 0.10017117425539199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0965,
      "step": 2926
    },
    {
      "epoch": 0.10020540910647038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0936,
      "step": 2927
    },
    {
      "epoch": 0.10023964395754878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.089,
      "step": 2928
    },
    {
      "epoch": 0.10027387880862718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1268,
      "step": 2929
    },
    {
      "epoch": 0.10030811365970559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1054,
      "step": 2930
    },
    {
      "epoch": 0.10034234851078398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1083,
      "step": 2931
    },
    {
      "epoch": 0.10037658336186238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1011,
      "step": 2932
    },
    {
      "epoch": 0.10041081821294078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.125,
      "step": 2933
    },
    {
      "epoch": 0.10044505306401917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0857,
      "step": 2934
    },
    {
      "epoch": 0.10047928791509757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.098,
      "step": 2935
    },
    {
      "epoch": 0.10051352276617596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1189,
      "step": 2936
    },
    {
      "epoch": 0.10054775761725436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0773,
      "step": 2937
    },
    {
      "epoch": 0.10058199246833276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1037,
      "step": 2938
    },
    {
      "epoch": 0.10061622731941117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1327,
      "step": 2939
    },
    {
      "epoch": 0.10065046217048956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.098,
      "step": 2940
    },
    {
      "epoch": 0.10068469702156796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0913,
      "step": 2941
    },
    {
      "epoch": 0.10071893187264636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1051,
      "step": 2942
    },
    {
      "epoch": 0.10075316672372475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0832,
      "step": 2943
    },
    {
      "epoch": 0.10078740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1075,
      "step": 2944
    },
    {
      "epoch": 0.10082163642588154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0942,
      "step": 2945
    },
    {
      "epoch": 0.10085587127695994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1057,
      "step": 2946
    },
    {
      "epoch": 0.10089010612803834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1217,
      "step": 2947
    },
    {
      "epoch": 0.10092434097911675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.112,
      "step": 2948
    },
    {
      "epoch": 0.10095857583019514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1031,
      "step": 2949
    },
    {
      "epoch": 0.10099281068127354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.105,
      "step": 2950
    },
    {
      "epoch": 0.10102704553235194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.088,
      "step": 2951
    },
    {
      "epoch": 0.10106128038343033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.083,
      "step": 2952
    },
    {
      "epoch": 0.10109551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1029,
      "step": 2953
    },
    {
      "epoch": 0.10112975008558713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1143,
      "step": 2954
    },
    {
      "epoch": 0.10116398493666552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0847,
      "step": 2955
    },
    {
      "epoch": 0.10119821978774392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1042,
      "step": 2956
    },
    {
      "epoch": 0.10123245463882233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0802,
      "step": 2957
    },
    {
      "epoch": 0.10126668948990072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1097,
      "step": 2958
    },
    {
      "epoch": 0.10130092434097912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0901,
      "step": 2959
    },
    {
      "epoch": 0.10133515919205752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1063,
      "step": 2960
    },
    {
      "epoch": 0.10136939404313591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0917,
      "step": 2961
    },
    {
      "epoch": 0.10140362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1002,
      "step": 2962
    },
    {
      "epoch": 0.1014378637452927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1046,
      "step": 2963
    },
    {
      "epoch": 0.1014720985963711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0623,
      "step": 2964
    },
    {
      "epoch": 0.1015063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1121,
      "step": 2965
    },
    {
      "epoch": 0.1015405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1018,
      "step": 2966
    },
    {
      "epoch": 0.1015748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0932,
      "step": 2967
    },
    {
      "epoch": 0.1016090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0928,
      "step": 2968
    },
    {
      "epoch": 0.1016432728517631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1161,
      "step": 2969
    },
    {
      "epoch": 0.1016775077028415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0699,
      "step": 2970
    },
    {
      "epoch": 0.10171174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0648,
      "step": 2971
    },
    {
      "epoch": 0.10174597740499829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0987,
      "step": 2972
    },
    {
      "epoch": 0.10178021225607668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0891,
      "step": 2973
    },
    {
      "epoch": 0.10181444710715508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0927,
      "step": 2974
    },
    {
      "epoch": 0.10184868195823348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0858,
      "step": 2975
    },
    {
      "epoch": 0.10188291680931189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0971,
      "step": 2976
    },
    {
      "epoch": 0.10191715166039028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1105,
      "step": 2977
    },
    {
      "epoch": 0.10195138651146868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0616,
      "step": 2978
    },
    {
      "epoch": 0.10198562136254707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1154,
      "step": 2979
    },
    {
      "epoch": 0.10201985621362547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.093,
      "step": 2980
    },
    {
      "epoch": 0.10205409106470387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0965,
      "step": 2981
    },
    {
      "epoch": 0.10208832591578226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1007,
      "step": 2982
    },
    {
      "epoch": 0.10212256076686066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0735,
      "step": 2983
    },
    {
      "epoch": 0.10215679561793906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0987,
      "step": 2984
    },
    {
      "epoch": 0.10219103046901747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0685,
      "step": 2985
    },
    {
      "epoch": 0.10222526532009586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0964,
      "step": 2986
    },
    {
      "epoch": 0.10225950017117426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1143,
      "step": 2987
    },
    {
      "epoch": 0.10229373502225265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0913,
      "step": 2988
    },
    {
      "epoch": 0.10232796987333105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1157,
      "step": 2989
    },
    {
      "epoch": 0.10236220472440945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1032,
      "step": 2990
    },
    {
      "epoch": 0.10239643957548784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0706,
      "step": 2991
    },
    {
      "epoch": 0.10243067442656624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0916,
      "step": 2992
    },
    {
      "epoch": 0.10246490927764464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0953,
      "step": 2993
    },
    {
      "epoch": 0.10249914412872305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1031,
      "step": 2994
    },
    {
      "epoch": 0.10253337897980144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0999,
      "step": 2995
    },
    {
      "epoch": 0.10256761383087984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0741,
      "step": 2996
    },
    {
      "epoch": 0.10260184868195824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1015,
      "step": 2997
    },
    {
      "epoch": 0.10263608353303663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0882,
      "step": 2998
    },
    {
      "epoch": 0.10267031838411503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0808,
      "step": 2999
    },
    {
      "epoch": 0.10270455323519342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1052,
      "step": 3000
    },
    {
      "epoch": 0.10273878808627182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0904,
      "step": 3001
    },
    {
      "epoch": 0.10277302293735022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0536,
      "step": 3002
    },
    {
      "epoch": 0.10280725778842863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0794,
      "step": 3003
    },
    {
      "epoch": 0.10284149263950702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.078,
      "step": 3004
    },
    {
      "epoch": 0.10287572749058542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0795,
      "step": 3005
    },
    {
      "epoch": 0.10290996234166382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0998,
      "step": 3006
    },
    {
      "epoch": 0.10294419719274221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1065,
      "step": 3007
    },
    {
      "epoch": 0.10297843204382061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0771,
      "step": 3008
    },
    {
      "epoch": 0.103012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0924,
      "step": 3009
    },
    {
      "epoch": 0.1030469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0741,
      "step": 3010
    },
    {
      "epoch": 0.1030811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1081,
      "step": 3011
    },
    {
      "epoch": 0.1031153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0728,
      "step": 3012
    },
    {
      "epoch": 0.1031496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0661,
      "step": 3013
    },
    {
      "epoch": 0.103183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0756,
      "step": 3014
    },
    {
      "epoch": 0.1032180760013694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0828,
      "step": 3015
    },
    {
      "epoch": 0.10325231085244779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1007,
      "step": 3016
    },
    {
      "epoch": 0.10328654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1244,
      "step": 3017
    },
    {
      "epoch": 0.10332078055460459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0754,
      "step": 3018
    },
    {
      "epoch": 0.10335501540568298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.107,
      "step": 3019
    },
    {
      "epoch": 0.10338925025676138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0927,
      "step": 3020
    },
    {
      "epoch": 0.10342348510783977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0857,
      "step": 3021
    },
    {
      "epoch": 0.10345771995891818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.098,
      "step": 3022
    },
    {
      "epoch": 0.10349195480999658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1108,
      "step": 3023
    },
    {
      "epoch": 0.10352618966107498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0672,
      "step": 3024
    },
    {
      "epoch": 0.10356042451215337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0995,
      "step": 3025
    },
    {
      "epoch": 0.10359465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0794,
      "step": 3026
    },
    {
      "epoch": 0.10362889421431017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0863,
      "step": 3027
    },
    {
      "epoch": 0.10366312906538856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0916,
      "step": 3028
    },
    {
      "epoch": 0.10369736391646696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0808,
      "step": 3029
    },
    {
      "epoch": 0.10373159876754535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0741,
      "step": 3030
    },
    {
      "epoch": 0.10376583361862377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0786,
      "step": 3031
    },
    {
      "epoch": 0.10380006846970216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1155,
      "step": 3032
    },
    {
      "epoch": 0.10383430332078056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1093,
      "step": 3033
    },
    {
      "epoch": 0.10386853817185895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1081,
      "step": 3034
    },
    {
      "epoch": 0.10390277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.105,
      "step": 3035
    },
    {
      "epoch": 0.10393700787401575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0844,
      "step": 3036
    },
    {
      "epoch": 0.10397124272509414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1268,
      "step": 3037
    },
    {
      "epoch": 0.10400547757617254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0945,
      "step": 3038
    },
    {
      "epoch": 0.10403971242725094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1045,
      "step": 3039
    },
    {
      "epoch": 0.10407394727832935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0711,
      "step": 3040
    },
    {
      "epoch": 0.10410818212940774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0913,
      "step": 3041
    },
    {
      "epoch": 0.10414241698048614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0672,
      "step": 3042
    },
    {
      "epoch": 0.10417665183156453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0515,
      "step": 3043
    },
    {
      "epoch": 0.10421088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0903,
      "step": 3044
    },
    {
      "epoch": 0.10424512153372133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0979,
      "step": 3045
    },
    {
      "epoch": 0.10427935638479972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0995,
      "step": 3046
    },
    {
      "epoch": 0.10431359123587812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0638,
      "step": 3047
    },
    {
      "epoch": 0.10434782608695652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0936,
      "step": 3048
    },
    {
      "epoch": 0.10438206093803493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0795,
      "step": 3049
    },
    {
      "epoch": 0.10441629578911332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0684,
      "step": 3050
    },
    {
      "epoch": 0.10445053064019172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0769,
      "step": 3051
    },
    {
      "epoch": 0.10448476549127012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0845,
      "step": 3052
    },
    {
      "epoch": 0.10451900034234851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0602,
      "step": 3053
    },
    {
      "epoch": 0.10455323519342691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0651,
      "step": 3054
    },
    {
      "epoch": 0.1045874700445053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0728,
      "step": 3055
    },
    {
      "epoch": 0.1046217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.062,
      "step": 3056
    },
    {
      "epoch": 0.1046559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0756,
      "step": 3057
    },
    {
      "epoch": 0.10469017459774049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.041,
      "step": 3058
    },
    {
      "epoch": 0.1047244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0645,
      "step": 3059
    },
    {
      "epoch": 0.1047586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0846,
      "step": 3060
    },
    {
      "epoch": 0.1047928791509757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0463,
      "step": 3061
    },
    {
      "epoch": 0.10482711400205409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0748,
      "step": 3062
    },
    {
      "epoch": 0.10486134885313249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0577,
      "step": 3063
    },
    {
      "epoch": 0.10489558370421088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1009,
      "step": 3064
    },
    {
      "epoch": 0.10492981855528928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0982,
      "step": 3065
    },
    {
      "epoch": 0.10496405340636768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0856,
      "step": 3066
    },
    {
      "epoch": 0.10499828825744607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0682,
      "step": 3067
    },
    {
      "epoch": 0.10503252310852448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0834,
      "step": 3068
    },
    {
      "epoch": 0.10506675795960288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0687,
      "step": 3069
    },
    {
      "epoch": 0.10510099281068128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0588,
      "step": 3070
    },
    {
      "epoch": 0.10513522766175967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0963,
      "step": 3071
    },
    {
      "epoch": 0.10516946251283807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.074,
      "step": 3072
    },
    {
      "epoch": 0.10520369736391647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0811,
      "step": 3073
    },
    {
      "epoch": 0.10523793221499486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0693,
      "step": 3074
    },
    {
      "epoch": 0.10527216706607326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0636,
      "step": 3075
    },
    {
      "epoch": 0.10530640191715165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0711,
      "step": 3076
    },
    {
      "epoch": 0.10534063676823006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0674,
      "step": 3077
    },
    {
      "epoch": 0.10537487161930846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0736,
      "step": 3078
    },
    {
      "epoch": 0.10540910647038686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0849,
      "step": 3079
    },
    {
      "epoch": 0.10544334132146525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0719,
      "step": 3080
    },
    {
      "epoch": 0.10547757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0725,
      "step": 3081
    },
    {
      "epoch": 0.10551181102362205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0779,
      "step": 3082
    },
    {
      "epoch": 0.10554604587470044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.057,
      "step": 3083
    },
    {
      "epoch": 0.10558028072577884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0687,
      "step": 3084
    },
    {
      "epoch": 0.10561451557685723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.071,
      "step": 3085
    },
    {
      "epoch": 0.10564875042793564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0734,
      "step": 3086
    },
    {
      "epoch": 0.10568298527901404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0844,
      "step": 3087
    },
    {
      "epoch": 0.10571722013009244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0761,
      "step": 3088
    },
    {
      "epoch": 0.10575145498117083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0572,
      "step": 3089
    },
    {
      "epoch": 0.10578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0933,
      "step": 3090
    },
    {
      "epoch": 0.10581992468332763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0728,
      "step": 3091
    },
    {
      "epoch": 0.10585415953440602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0455,
      "step": 3092
    },
    {
      "epoch": 0.10588839438548442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0777,
      "step": 3093
    },
    {
      "epoch": 0.10592262923656282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0558,
      "step": 3094
    },
    {
      "epoch": 0.10595686408764123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0722,
      "step": 3095
    },
    {
      "epoch": 0.10599109893871962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0659,
      "step": 3096
    },
    {
      "epoch": 0.10602533378979802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0793,
      "step": 3097
    },
    {
      "epoch": 0.10605956864087641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0465,
      "step": 3098
    },
    {
      "epoch": 0.10609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0471,
      "step": 3099
    },
    {
      "epoch": 0.1061280383430332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0419,
      "step": 3100
    },
    {
      "epoch": 0.1061622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0604,
      "step": 3101
    },
    {
      "epoch": 0.10619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0749,
      "step": 3102
    },
    {
      "epoch": 0.1062307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0614,
      "step": 3103
    },
    {
      "epoch": 0.10626497774734679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0636,
      "step": 3104
    },
    {
      "epoch": 0.1062992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0759,
      "step": 3105
    },
    {
      "epoch": 0.1063334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.073,
      "step": 3106
    },
    {
      "epoch": 0.106367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0584,
      "step": 3107
    },
    {
      "epoch": 0.10640191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0369,
      "step": 3108
    },
    {
      "epoch": 0.10643615200273879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0543,
      "step": 3109
    },
    {
      "epoch": 0.10647038685381718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0552,
      "step": 3110
    },
    {
      "epoch": 0.10650462170489558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0401,
      "step": 3111
    },
    {
      "epoch": 0.10653885655597398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.071,
      "step": 3112
    },
    {
      "epoch": 0.10657309140705237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0344,
      "step": 3113
    },
    {
      "epoch": 0.10660732625813078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0496,
      "step": 3114
    },
    {
      "epoch": 0.10664156110920918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0699,
      "step": 3115
    },
    {
      "epoch": 0.10667579596028758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0355,
      "step": 3116
    },
    {
      "epoch": 0.10671003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0576,
      "step": 3117
    },
    {
      "epoch": 0.10674426566244437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0822,
      "step": 3118
    },
    {
      "epoch": 0.10677850051352276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0423,
      "step": 3119
    },
    {
      "epoch": 0.10681273536460116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0896,
      "step": 3120
    },
    {
      "epoch": 0.10684697021567956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0678,
      "step": 3121
    },
    {
      "epoch": 0.10688120506675795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0549,
      "step": 3122
    },
    {
      "epoch": 0.10691543991783636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.036,
      "step": 3123
    },
    {
      "epoch": 0.10694967476891476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0356,
      "step": 3124
    },
    {
      "epoch": 0.10698390961999316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.06,
      "step": 3125
    },
    {
      "epoch": 0.10701814447107155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0752,
      "step": 3126
    },
    {
      "epoch": 0.10705237932214995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.053,
      "step": 3127
    },
    {
      "epoch": 0.10708661417322834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0588,
      "step": 3128
    },
    {
      "epoch": 0.10712084902430674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0571,
      "step": 3129
    },
    {
      "epoch": 0.10715508387538514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0535,
      "step": 3130
    },
    {
      "epoch": 0.10718931872646353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0346,
      "step": 3131
    },
    {
      "epoch": 0.10722355357754194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0683,
      "step": 3132
    },
    {
      "epoch": 0.10725778842862034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0678,
      "step": 3133
    },
    {
      "epoch": 0.10729202327969874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0474,
      "step": 3134
    },
    {
      "epoch": 0.10732625813077713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0563,
      "step": 3135
    },
    {
      "epoch": 0.10736049298185553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0457,
      "step": 3136
    },
    {
      "epoch": 0.10739472783293393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0861,
      "step": 3137
    },
    {
      "epoch": 0.10742896268401232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0264,
      "step": 3138
    },
    {
      "epoch": 0.10746319753509072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0676,
      "step": 3139
    },
    {
      "epoch": 0.10749743238616911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0383,
      "step": 3140
    },
    {
      "epoch": 0.10753166723724752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0558,
      "step": 3141
    },
    {
      "epoch": 0.10756590208832592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0421,
      "step": 3142
    },
    {
      "epoch": 0.10760013693940432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0527,
      "step": 3143
    },
    {
      "epoch": 0.10763437179048271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0673,
      "step": 3144
    },
    {
      "epoch": 0.10766860664156111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0354,
      "step": 3145
    },
    {
      "epoch": 0.1077028414926395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0651,
      "step": 3146
    },
    {
      "epoch": 0.1077370763437179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0266,
      "step": 3147
    },
    {
      "epoch": 0.1077713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.036,
      "step": 3148
    },
    {
      "epoch": 0.1078055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.052,
      "step": 3149
    },
    {
      "epoch": 0.10783978089695309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0449,
      "step": 3150
    },
    {
      "epoch": 0.1078740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0101,
      "step": 3151
    },
    {
      "epoch": 0.1079082505991099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0674,
      "step": 3152
    },
    {
      "epoch": 0.1079424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0604,
      "step": 3153
    },
    {
      "epoch": 0.10797672030126669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0621,
      "step": 3154
    },
    {
      "epoch": 0.10801095515234509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.046,
      "step": 3155
    },
    {
      "epoch": 0.10804519000342348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0281,
      "step": 3156
    },
    {
      "epoch": 0.10807942485450188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0311,
      "step": 3157
    },
    {
      "epoch": 0.10811365970558028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0686,
      "step": 3158
    },
    {
      "epoch": 0.10814789455665867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0476,
      "step": 3159
    },
    {
      "epoch": 0.10818212940773708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.069,
      "step": 3160
    },
    {
      "epoch": 0.10821636425881548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0464,
      "step": 3161
    },
    {
      "epoch": 0.10825059910989387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0606,
      "step": 3162
    },
    {
      "epoch": 0.10828483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0211,
      "step": 3163
    },
    {
      "epoch": 0.10831906881205067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0569,
      "step": 3164
    },
    {
      "epoch": 0.10835330366312906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0585,
      "step": 3165
    },
    {
      "epoch": 0.10838753851420746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0557,
      "step": 3166
    },
    {
      "epoch": 0.10842177336528586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.055,
      "step": 3167
    },
    {
      "epoch": 0.10845600821636425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0396,
      "step": 3168
    },
    {
      "epoch": 0.10849024306744266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.045,
      "step": 3169
    },
    {
      "epoch": 0.10852447791852106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0407,
      "step": 3170
    },
    {
      "epoch": 0.10855871276959946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0625,
      "step": 3171
    },
    {
      "epoch": 0.10859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0828,
      "step": 3172
    },
    {
      "epoch": 0.10862718247175625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0457,
      "step": 3173
    },
    {
      "epoch": 0.10866141732283464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0513,
      "step": 3174
    },
    {
      "epoch": 0.10869565217391304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0582,
      "step": 3175
    },
    {
      "epoch": 0.10872988702499144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0406,
      "step": 3176
    },
    {
      "epoch": 0.10876412187606983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0559,
      "step": 3177
    },
    {
      "epoch": 0.10879835672714824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0622,
      "step": 3178
    },
    {
      "epoch": 0.10883259157822664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0709,
      "step": 3179
    },
    {
      "epoch": 0.10886682642930504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0361,
      "step": 3180
    },
    {
      "epoch": 0.10890106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0546,
      "step": 3181
    },
    {
      "epoch": 0.10893529613146183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0566,
      "step": 3182
    },
    {
      "epoch": 0.10896953098254022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0389,
      "step": 3183
    },
    {
      "epoch": 0.10900376583361862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0611,
      "step": 3184
    },
    {
      "epoch": 0.10903800068469702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0369,
      "step": 3185
    },
    {
      "epoch": 0.10907223553577541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0296,
      "step": 3186
    },
    {
      "epoch": 0.10910647038685382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.051,
      "step": 3187
    },
    {
      "epoch": 0.10914070523793222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0346,
      "step": 3188
    },
    {
      "epoch": 0.10917494008901062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0574,
      "step": 3189
    },
    {
      "epoch": 0.10920917494008901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0295,
      "step": 3190
    },
    {
      "epoch": 0.10924340979116741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0401,
      "step": 3191
    },
    {
      "epoch": 0.1092776446422458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.043,
      "step": 3192
    },
    {
      "epoch": 0.1093118794933242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0549,
      "step": 3193
    },
    {
      "epoch": 0.1093461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0265,
      "step": 3194
    },
    {
      "epoch": 0.109380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0375,
      "step": 3195
    },
    {
      "epoch": 0.1094145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0226,
      "step": 3196
    },
    {
      "epoch": 0.1094488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0232,
      "step": 3197
    },
    {
      "epoch": 0.1094830537487162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0156,
      "step": 3198
    },
    {
      "epoch": 0.1095172885997946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0396,
      "step": 3199
    },
    {
      "epoch": 0.10955152345087299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.03,
      "step": 3200
    },
    {
      "epoch": 0.10958575830195139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0683,
      "step": 3201
    },
    {
      "epoch": 0.10961999315302978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0669,
      "step": 3202
    },
    {
      "epoch": 0.10965422800410818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0598,
      "step": 3203
    },
    {
      "epoch": 0.10968846285518657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0571,
      "step": 3204
    },
    {
      "epoch": 0.10972269770626497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0614,
      "step": 3205
    },
    {
      "epoch": 0.10975693255734338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0252,
      "step": 3206
    },
    {
      "epoch": 0.10979116740842178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0439,
      "step": 3207
    },
    {
      "epoch": 0.10982540225950017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0266,
      "step": 3208
    },
    {
      "epoch": 0.10985963711057857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0372,
      "step": 3209
    },
    {
      "epoch": 0.10989387196165697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0282,
      "step": 3210
    },
    {
      "epoch": 0.10992810681273536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0233,
      "step": 3211
    },
    {
      "epoch": 0.10996234166381376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0522,
      "step": 3212
    },
    {
      "epoch": 0.10999657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0084,
      "step": 3213
    },
    {
      "epoch": 0.11003081136597055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0496,
      "step": 3214
    },
    {
      "epoch": 0.11006504621704896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0365,
      "step": 3215
    },
    {
      "epoch": 0.11009928106812736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0393,
      "step": 3216
    },
    {
      "epoch": 0.11013351591920575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0532,
      "step": 3217
    },
    {
      "epoch": 0.11016775077028415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0018,
      "step": 3218
    },
    {
      "epoch": 0.11020198562136255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0514,
      "step": 3219
    },
    {
      "epoch": 0.11023622047244094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0288,
      "step": 3220
    },
    {
      "epoch": 0.11027045532351934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0382,
      "step": 3221
    },
    {
      "epoch": 0.11030469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.026,
      "step": 3222
    },
    {
      "epoch": 0.11033892502567613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0262,
      "step": 3223
    },
    {
      "epoch": 0.11037315987675454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0274,
      "step": 3224
    },
    {
      "epoch": 0.11040739472783294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0262,
      "step": 3225
    },
    {
      "epoch": 0.11044162957891134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.016,
      "step": 3226
    },
    {
      "epoch": 0.11047586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.033,
      "step": 3227
    },
    {
      "epoch": 0.11051009928106813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0327,
      "step": 3228
    },
    {
      "epoch": 0.11054433413214652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0037,
      "step": 3229
    },
    {
      "epoch": 0.11057856898322492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.031,
      "step": 3230
    },
    {
      "epoch": 0.11061280383430332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0533,
      "step": 3231
    },
    {
      "epoch": 0.11064703868538171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0212,
      "step": 3232
    },
    {
      "epoch": 0.11068127353646012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0405,
      "step": 3233
    },
    {
      "epoch": 0.11071550838753852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0454,
      "step": 3234
    },
    {
      "epoch": 0.11074974323861692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0254,
      "step": 3235
    },
    {
      "epoch": 0.11078397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0276,
      "step": 3236
    },
    {
      "epoch": 0.11081821294077371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0401,
      "step": 3237
    },
    {
      "epoch": 0.1108524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0475,
      "step": 3238
    },
    {
      "epoch": 0.1108866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.051,
      "step": 3239
    },
    {
      "epoch": 0.1109209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0457,
      "step": 3240
    },
    {
      "epoch": 0.1109551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0739,
      "step": 3241
    },
    {
      "epoch": 0.1109893871961657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0638,
      "step": 3242
    },
    {
      "epoch": 0.1110236220472441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0059,
      "step": 3243
    },
    {
      "epoch": 0.1110578568983225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0369,
      "step": 3244
    },
    {
      "epoch": 0.11109209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0179,
      "step": 3245
    },
    {
      "epoch": 0.11112632660047929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0099,
      "step": 3246
    },
    {
      "epoch": 0.11116056145155769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0416,
      "step": 3247
    },
    {
      "epoch": 0.11119479630263608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0208,
      "step": 3248
    },
    {
      "epoch": 0.11122903115371448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0543,
      "step": 3249
    },
    {
      "epoch": 0.11126326600479287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0055,
      "step": 3250
    },
    {
      "epoch": 0.11129750085587127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0385,
      "step": 3251
    },
    {
      "epoch": 0.11133173570694968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0304,
      "step": 3252
    },
    {
      "epoch": 0.11136597055802808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0135,
      "step": 3253
    },
    {
      "epoch": 0.11140020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.026,
      "step": 3254
    },
    {
      "epoch": 0.11143444026018487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0203,
      "step": 3255
    },
    {
      "epoch": 0.11146867511126327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0073,
      "step": 3256
    },
    {
      "epoch": 0.11150290996234166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.049,
      "step": 3257
    },
    {
      "epoch": 0.11153714481342006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.037,
      "step": 3258
    },
    {
      "epoch": 0.11157137966449845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.007,
      "step": 3259
    },
    {
      "epoch": 0.11160561451557685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0384,
      "step": 3260
    },
    {
      "epoch": 0.11163984936665526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0458,
      "step": 3261
    },
    {
      "epoch": 0.11167408421773366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0181,
      "step": 3262
    },
    {
      "epoch": 0.11170831906881205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0244,
      "step": 3263
    },
    {
      "epoch": 0.11174255391989045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0462,
      "step": 3264
    },
    {
      "epoch": 0.11177678877096885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0328,
      "step": 3265
    },
    {
      "epoch": 0.11181102362204724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0383,
      "step": 3266
    },
    {
      "epoch": 0.11184525847312564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0294,
      "step": 3267
    },
    {
      "epoch": 0.11187949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0378,
      "step": 3268
    },
    {
      "epoch": 0.11191372817528243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0284,
      "step": 3269
    },
    {
      "epoch": 0.11194796302636084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0181,
      "step": 3270
    },
    {
      "epoch": 0.11198219787743924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0234,
      "step": 3271
    },
    {
      "epoch": 0.11201643272851763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0151,
      "step": 3272
    },
    {
      "epoch": 0.11205066757959603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0348,
      "step": 3273
    },
    {
      "epoch": 0.11208490243067443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0328,
      "step": 3274
    },
    {
      "epoch": 0.11211913728175282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0442,
      "step": 3275
    },
    {
      "epoch": 0.11215337213283122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0033,
      "step": 3276
    },
    {
      "epoch": 0.11218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0323,
      "step": 3277
    },
    {
      "epoch": 0.11222184183498801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.027,
      "step": 3278
    },
    {
      "epoch": 0.11225607668606642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0021,
      "step": 3279
    },
    {
      "epoch": 0.11229031153714482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0115,
      "step": 3280
    },
    {
      "epoch": 0.11232454638822321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0207,
      "step": 3281
    },
    {
      "epoch": 0.11235878123930161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.03,
      "step": 3282
    },
    {
      "epoch": 0.11239301609038001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.001,
      "step": 3283
    },
    {
      "epoch": 0.1124272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0118,
      "step": 3284
    },
    {
      "epoch": 0.1124614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0347,
      "step": 3285
    },
    {
      "epoch": 0.1124957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0111,
      "step": 3286
    },
    {
      "epoch": 0.11252995549469359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0226,
      "step": 3287
    },
    {
      "epoch": 0.112564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0201,
      "step": 3288
    },
    {
      "epoch": 0.1125984251968504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0174,
      "step": 3289
    },
    {
      "epoch": 0.1126326600479288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.018,
      "step": 3290
    },
    {
      "epoch": 0.11266689489900719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0333,
      "step": 3291
    },
    {
      "epoch": 0.11270112975008559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0171,
      "step": 3292
    },
    {
      "epoch": 0.11273536460116398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0227,
      "step": 3293
    },
    {
      "epoch": 0.11276959945224238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0197,
      "step": 3294
    },
    {
      "epoch": 0.11280383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0008,
      "step": 3295
    },
    {
      "epoch": 0.11283806915439917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0207,
      "step": 3296
    },
    {
      "epoch": 0.11287230400547757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0158,
      "step": 3297
    },
    {
      "epoch": 0.11290653885655598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0261,
      "step": 3298
    },
    {
      "epoch": 0.11294077370763438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0034,
      "step": 3299
    },
    {
      "epoch": 0.11297500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0034,
      "step": 3300
    },
    {
      "epoch": 0.11300924340979117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9938,
      "step": 3301
    },
    {
      "epoch": 0.11304347826086956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0142,
      "step": 3302
    },
    {
      "epoch": 0.11307771311194796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.006,
      "step": 3303
    },
    {
      "epoch": 0.11311194796302636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0267,
      "step": 3304
    },
    {
      "epoch": 0.11314618281410475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0004,
      "step": 3305
    },
    {
      "epoch": 0.11318041766518315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0192,
      "step": 3306
    },
    {
      "epoch": 0.11321465251626156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.027,
      "step": 3307
    },
    {
      "epoch": 0.11324888736733996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0126,
      "step": 3308
    },
    {
      "epoch": 0.11328312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0403,
      "step": 3309
    },
    {
      "epoch": 0.11331735706949675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.032,
      "step": 3310
    },
    {
      "epoch": 0.11335159192057515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.025,
      "step": 3311
    },
    {
      "epoch": 0.11338582677165354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.017,
      "step": 3312
    },
    {
      "epoch": 0.11342006162273194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0114,
      "step": 3313
    },
    {
      "epoch": 0.11345429647381033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0342,
      "step": 3314
    },
    {
      "epoch": 0.11348853132488873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0184,
      "step": 3315
    },
    {
      "epoch": 0.11352276617596714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0449,
      "step": 3316
    },
    {
      "epoch": 0.11355700102704554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0195,
      "step": 3317
    },
    {
      "epoch": 0.11359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0051,
      "step": 3318
    },
    {
      "epoch": 0.11362547072920233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0214,
      "step": 3319
    },
    {
      "epoch": 0.11365970558028073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0014,
      "step": 3320
    },
    {
      "epoch": 0.11369394043135912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9692,
      "step": 3321
    },
    {
      "epoch": 0.11372817528243752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0185,
      "step": 3322
    },
    {
      "epoch": 0.11376241013351591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.012,
      "step": 3323
    },
    {
      "epoch": 0.11379664498459431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0058,
      "step": 3324
    },
    {
      "epoch": 0.11383087983567272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0263,
      "step": 3325
    },
    {
      "epoch": 0.11386511468675112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9921,
      "step": 3326
    },
    {
      "epoch": 0.11389934953782951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0073,
      "step": 3327
    },
    {
      "epoch": 0.11393358438890791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0021,
      "step": 3328
    },
    {
      "epoch": 0.1139678192399863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0146,
      "step": 3329
    },
    {
      "epoch": 0.1140020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0571,
      "step": 3330
    },
    {
      "epoch": 0.1140362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0248,
      "step": 3331
    },
    {
      "epoch": 0.1140705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0317,
      "step": 3332
    },
    {
      "epoch": 0.11410475864429989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0175,
      "step": 3333
    },
    {
      "epoch": 0.1141389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0165,
      "step": 3334
    },
    {
      "epoch": 0.1141732283464567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9988,
      "step": 3335
    },
    {
      "epoch": 0.1142074631975351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0181,
      "step": 3336
    },
    {
      "epoch": 0.11424169804861349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0392,
      "step": 3337
    },
    {
      "epoch": 0.11427593289969189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.011,
      "step": 3338
    },
    {
      "epoch": 0.11431016775077028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0217,
      "step": 3339
    },
    {
      "epoch": 0.11434440260184868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9788,
      "step": 3340
    },
    {
      "epoch": 0.11437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0142,
      "step": 3341
    },
    {
      "epoch": 0.11441287230400547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0045,
      "step": 3342
    },
    {
      "epoch": 0.11444710715508387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.01,
      "step": 3343
    },
    {
      "epoch": 0.11448134200616228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0223,
      "step": 3344
    },
    {
      "epoch": 0.11451557685724068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0295,
      "step": 3345
    },
    {
      "epoch": 0.11454981170831907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0168,
      "step": 3346
    },
    {
      "epoch": 0.11458404655939747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9983,
      "step": 3347
    },
    {
      "epoch": 0.11461828141047586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0209,
      "step": 3348
    },
    {
      "epoch": 0.11465251626155426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0093,
      "step": 3349
    },
    {
      "epoch": 0.11468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9987,
      "step": 3350
    },
    {
      "epoch": 0.11472098596371105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0216,
      "step": 3351
    },
    {
      "epoch": 0.11475522081478945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9919,
      "step": 3352
    },
    {
      "epoch": 0.11478945566586786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9982,
      "step": 3353
    },
    {
      "epoch": 0.11482369051694626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9916,
      "step": 3354
    },
    {
      "epoch": 0.11485792536802465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0092,
      "step": 3355
    },
    {
      "epoch": 0.11489216021910305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0261,
      "step": 3356
    },
    {
      "epoch": 0.11492639507018144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0271,
      "step": 3357
    },
    {
      "epoch": 0.11496062992125984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9904,
      "step": 3358
    },
    {
      "epoch": 0.11499486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0391,
      "step": 3359
    },
    {
      "epoch": 0.11502909962341663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9909,
      "step": 3360
    },
    {
      "epoch": 0.11506333447449503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0105,
      "step": 3361
    },
    {
      "epoch": 0.11509756932557344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0091,
      "step": 3362
    },
    {
      "epoch": 0.11513180417665184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0101,
      "step": 3363
    },
    {
      "epoch": 0.11516603902773023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0198,
      "step": 3364
    },
    {
      "epoch": 0.11520027387880863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0061,
      "step": 3365
    },
    {
      "epoch": 0.11523450872988703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0172,
      "step": 3366
    },
    {
      "epoch": 0.11526874358096542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0228,
      "step": 3367
    },
    {
      "epoch": 0.11530297843204382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0113,
      "step": 3368
    },
    {
      "epoch": 0.11533721328312221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9886,
      "step": 3369
    },
    {
      "epoch": 0.11537144813420061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9926,
      "step": 3370
    },
    {
      "epoch": 0.11540568298527902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9973,
      "step": 3371
    },
    {
      "epoch": 0.11543991783635742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9844,
      "step": 3372
    },
    {
      "epoch": 0.11547415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0129,
      "step": 3373
    },
    {
      "epoch": 0.11550838753851421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9752,
      "step": 3374
    },
    {
      "epoch": 0.1155426223895926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0066,
      "step": 3375
    },
    {
      "epoch": 0.115576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9616,
      "step": 3376
    },
    {
      "epoch": 0.1156110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0272,
      "step": 3377
    },
    {
      "epoch": 0.1156453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9872,
      "step": 3378
    },
    {
      "epoch": 0.11567956179390619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0243,
      "step": 3379
    },
    {
      "epoch": 0.1157137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9767,
      "step": 3380
    },
    {
      "epoch": 0.115748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9897,
      "step": 3381
    },
    {
      "epoch": 0.1157822663471414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0,
      "step": 3382
    },
    {
      "epoch": 0.11581650119821979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0023,
      "step": 3383
    },
    {
      "epoch": 0.11585073604929819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0213,
      "step": 3384
    },
    {
      "epoch": 0.11588497090037658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9963,
      "step": 3385
    },
    {
      "epoch": 0.11591920575145498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9901,
      "step": 3386
    },
    {
      "epoch": 0.11595344060253338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9957,
      "step": 3387
    },
    {
      "epoch": 0.11598767545361177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9996,
      "step": 3388
    },
    {
      "epoch": 0.11602191030469017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0138,
      "step": 3389
    },
    {
      "epoch": 0.11605614515576858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0026,
      "step": 3390
    },
    {
      "epoch": 0.11609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.997,
      "step": 3391
    },
    {
      "epoch": 0.11612461485792537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0173,
      "step": 3392
    },
    {
      "epoch": 0.11615884970900377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9991,
      "step": 3393
    },
    {
      "epoch": 0.11619308456008216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0243,
      "step": 3394
    },
    {
      "epoch": 0.11622731941116056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0204,
      "step": 3395
    },
    {
      "epoch": 0.11626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0104,
      "step": 3396
    },
    {
      "epoch": 0.11629578911331735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0026,
      "step": 3397
    },
    {
      "epoch": 0.11633002396439575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0169,
      "step": 3398
    },
    {
      "epoch": 0.11636425881547416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0176,
      "step": 3399
    },
    {
      "epoch": 0.11639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9902,
      "step": 3400
    },
    {
      "epoch": 0.11643272851763095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.983,
      "step": 3401
    },
    {
      "epoch": 0.11646696336870935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9819,
      "step": 3402
    },
    {
      "epoch": 0.11650119821978774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9988,
      "step": 3403
    },
    {
      "epoch": 0.11653543307086614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9836,
      "step": 3404
    },
    {
      "epoch": 0.11656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9991,
      "step": 3405
    },
    {
      "epoch": 0.11660390277302293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9923,
      "step": 3406
    },
    {
      "epoch": 0.11663813762410133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9922,
      "step": 3407
    },
    {
      "epoch": 0.11667237247517974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.967,
      "step": 3408
    },
    {
      "epoch": 0.11670660732625814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9858,
      "step": 3409
    },
    {
      "epoch": 0.11674084217733653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9937,
      "step": 3410
    },
    {
      "epoch": 0.11677507702841493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0271,
      "step": 3411
    },
    {
      "epoch": 0.11680931187949332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9829,
      "step": 3412
    },
    {
      "epoch": 0.11684354673057172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9781,
      "step": 3413
    },
    {
      "epoch": 0.11687778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.027,
      "step": 3414
    },
    {
      "epoch": 0.11691201643272851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9957,
      "step": 3415
    },
    {
      "epoch": 0.11694625128380691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0063,
      "step": 3416
    },
    {
      "epoch": 0.11698048613488532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0002,
      "step": 3417
    },
    {
      "epoch": 0.11701472098596372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9985,
      "step": 3418
    },
    {
      "epoch": 0.11704895583704211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9733,
      "step": 3419
    },
    {
      "epoch": 0.11708319068812051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9751,
      "step": 3420
    },
    {
      "epoch": 0.1171174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9957,
      "step": 3421
    },
    {
      "epoch": 0.1171516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.011,
      "step": 3422
    },
    {
      "epoch": 0.1171858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9975,
      "step": 3423
    },
    {
      "epoch": 0.1172201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9679,
      "step": 3424
    },
    {
      "epoch": 0.11725436494351249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9828,
      "step": 3425
    },
    {
      "epoch": 0.1172885997945909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9902,
      "step": 3426
    },
    {
      "epoch": 0.1173228346456693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9947,
      "step": 3427
    },
    {
      "epoch": 0.11735706949674769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0193,
      "step": 3428
    },
    {
      "epoch": 0.11739130434782609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0078,
      "step": 3429
    },
    {
      "epoch": 0.11742553919890449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9605,
      "step": 3430
    },
    {
      "epoch": 0.11745977404998288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9873,
      "step": 3431
    },
    {
      "epoch": 0.11749400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9744,
      "step": 3432
    },
    {
      "epoch": 0.11752824375213967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9993,
      "step": 3433
    },
    {
      "epoch": 0.11756247860321807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0061,
      "step": 3434
    },
    {
      "epoch": 0.11759671345429647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0056,
      "step": 3435
    },
    {
      "epoch": 0.11763094830537488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9958,
      "step": 3436
    },
    {
      "epoch": 0.11766518315645327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9954,
      "step": 3437
    },
    {
      "epoch": 0.11769941800753167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.966,
      "step": 3438
    },
    {
      "epoch": 0.11773365285861007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9894,
      "step": 3439
    },
    {
      "epoch": 0.11776788770968846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.009,
      "step": 3440
    },
    {
      "epoch": 0.11780212256076686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0049,
      "step": 3441
    },
    {
      "epoch": 0.11783635741184526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9762,
      "step": 3442
    },
    {
      "epoch": 0.11787059226292365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9891,
      "step": 3443
    },
    {
      "epoch": 0.11790482711400205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9906,
      "step": 3444
    },
    {
      "epoch": 0.11793906196508046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9791,
      "step": 3445
    },
    {
      "epoch": 0.11797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9999,
      "step": 3446
    },
    {
      "epoch": 0.11800753166723725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.011,
      "step": 3447
    },
    {
      "epoch": 0.11804176651831565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9913,
      "step": 3448
    },
    {
      "epoch": 0.11807600136939404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9999,
      "step": 3449
    },
    {
      "epoch": 0.11811023622047244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0107,
      "step": 3450
    },
    {
      "epoch": 0.11814447107155084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9905,
      "step": 3451
    },
    {
      "epoch": 0.11817870592262923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9926,
      "step": 3452
    },
    {
      "epoch": 0.11821294077370763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9962,
      "step": 3453
    },
    {
      "epoch": 0.11824717562478604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9872,
      "step": 3454
    },
    {
      "epoch": 0.11828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9875,
      "step": 3455
    },
    {
      "epoch": 0.11831564532694283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9874,
      "step": 3456
    },
    {
      "epoch": 0.11834988017802123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9951,
      "step": 3457
    },
    {
      "epoch": 0.11838411502909962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0144,
      "step": 3458
    },
    {
      "epoch": 0.11841834988017802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9991,
      "step": 3459
    },
    {
      "epoch": 0.11845258473125642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9643,
      "step": 3460
    },
    {
      "epoch": 0.11848681958233481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9707,
      "step": 3461
    },
    {
      "epoch": 0.11852105443341321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9985,
      "step": 3462
    },
    {
      "epoch": 0.11855528928449162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9841,
      "step": 3463
    },
    {
      "epoch": 0.11858952413557002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0016,
      "step": 3464
    },
    {
      "epoch": 0.11862375898664841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.972,
      "step": 3465
    },
    {
      "epoch": 0.11865799383772681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0203,
      "step": 3466
    },
    {
      "epoch": 0.1186922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9952,
      "step": 3467
    },
    {
      "epoch": 0.1187264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9713,
      "step": 3468
    },
    {
      "epoch": 0.118760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9583,
      "step": 3469
    },
    {
      "epoch": 0.1187949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9933,
      "step": 3470
    },
    {
      "epoch": 0.11882916809311879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.971,
      "step": 3471
    },
    {
      "epoch": 0.1188634029441972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9694,
      "step": 3472
    },
    {
      "epoch": 0.1188976377952756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9734,
      "step": 3473
    },
    {
      "epoch": 0.11893187264635399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0049,
      "step": 3474
    },
    {
      "epoch": 0.11896610749743239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9645,
      "step": 3475
    },
    {
      "epoch": 0.11900034234851078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9718,
      "step": 3476
    },
    {
      "epoch": 0.11903457719958918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0294,
      "step": 3477
    },
    {
      "epoch": 0.11906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9686,
      "step": 3478
    },
    {
      "epoch": 0.11910304690174597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9993,
      "step": 3479
    },
    {
      "epoch": 0.11913728175282437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9907,
      "step": 3480
    },
    {
      "epoch": 0.11917151660390277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9772,
      "step": 3481
    },
    {
      "epoch": 0.11920575145498118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0086,
      "step": 3482
    },
    {
      "epoch": 0.11923998630605957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9715,
      "step": 3483
    },
    {
      "epoch": 0.11927422115713797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9601,
      "step": 3484
    },
    {
      "epoch": 0.11930845600821637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.001,
      "step": 3485
    },
    {
      "epoch": 0.11934269085929476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9919,
      "step": 3486
    },
    {
      "epoch": 0.11937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9573,
      "step": 3487
    },
    {
      "epoch": 0.11941116056145155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9993,
      "step": 3488
    },
    {
      "epoch": 0.11944539541252995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9963,
      "step": 3489
    },
    {
      "epoch": 0.11947963026360835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0106,
      "step": 3490
    },
    {
      "epoch": 0.11951386511468676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0027,
      "step": 3491
    },
    {
      "epoch": 0.11954809996576515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9787,
      "step": 3492
    },
    {
      "epoch": 0.11958233481684355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9684,
      "step": 3493
    },
    {
      "epoch": 0.11961656966792195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9602,
      "step": 3494
    },
    {
      "epoch": 0.11965080451900034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9885,
      "step": 3495
    },
    {
      "epoch": 0.11968503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9983,
      "step": 3496
    },
    {
      "epoch": 0.11971927422115713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9665,
      "step": 3497
    },
    {
      "epoch": 0.11975350907223553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9731,
      "step": 3498
    },
    {
      "epoch": 0.11978774392331393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9543,
      "step": 3499
    },
    {
      "epoch": 0.11982197877439234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9761,
      "step": 3500
    },
    {
      "epoch": 0.11985621362547073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0087,
      "step": 3501
    },
    {
      "epoch": 0.11989044847654913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9764,
      "step": 3502
    },
    {
      "epoch": 0.11992468332762753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9806,
      "step": 3503
    },
    {
      "epoch": 0.11995891817870592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9779,
      "step": 3504
    },
    {
      "epoch": 0.11999315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9922,
      "step": 3505
    },
    {
      "epoch": 0.12002738788086272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9775,
      "step": 3506
    },
    {
      "epoch": 0.12006162273194111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9628,
      "step": 3507
    },
    {
      "epoch": 0.12009585758301951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9833,
      "step": 3508
    },
    {
      "epoch": 0.12013009243409792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9885,
      "step": 3509
    },
    {
      "epoch": 0.12016432728517631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0112,
      "step": 3510
    },
    {
      "epoch": 0.12019856213625471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.988,
      "step": 3511
    },
    {
      "epoch": 0.12023279698733311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9944,
      "step": 3512
    },
    {
      "epoch": 0.1202670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.975,
      "step": 3513
    },
    {
      "epoch": 0.1203012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9885,
      "step": 3514
    },
    {
      "epoch": 0.1203355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.988,
      "step": 3515
    },
    {
      "epoch": 0.12036973639164669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9797,
      "step": 3516
    },
    {
      "epoch": 0.12040397124272509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.983,
      "step": 3517
    },
    {
      "epoch": 0.1204382060938035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9614,
      "step": 3518
    },
    {
      "epoch": 0.1204724409448819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9856,
      "step": 3519
    },
    {
      "epoch": 0.12050667579596029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9898,
      "step": 3520
    },
    {
      "epoch": 0.12054091064703869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9925,
      "step": 3521
    },
    {
      "epoch": 0.12057514549811708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9503,
      "step": 3522
    },
    {
      "epoch": 0.12060938034919548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9738,
      "step": 3523
    },
    {
      "epoch": 0.12064361520027388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9794,
      "step": 3524
    },
    {
      "epoch": 0.12067785005135227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9811,
      "step": 3525
    },
    {
      "epoch": 0.12071208490243067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.014,
      "step": 3526
    },
    {
      "epoch": 0.12074631975350907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9767,
      "step": 3527
    },
    {
      "epoch": 0.12078055460458748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9546,
      "step": 3528
    },
    {
      "epoch": 0.12081478945566587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9597,
      "step": 3529
    },
    {
      "epoch": 0.12084902430674427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9759,
      "step": 3530
    },
    {
      "epoch": 0.12088325915782266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9949,
      "step": 3531
    },
    {
      "epoch": 0.12091749400890106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9804,
      "step": 3532
    },
    {
      "epoch": 0.12095172885997946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9725,
      "step": 3533
    },
    {
      "epoch": 0.12098596371105785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9777,
      "step": 3534
    },
    {
      "epoch": 0.12102019856213625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9634,
      "step": 3535
    },
    {
      "epoch": 0.12105443341321465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9665,
      "step": 3536
    },
    {
      "epoch": 0.12108866826429306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.964,
      "step": 3537
    },
    {
      "epoch": 0.12112290311537145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9559,
      "step": 3538
    },
    {
      "epoch": 0.12115713796644985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9582,
      "step": 3539
    },
    {
      "epoch": 0.12119137281752825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9621,
      "step": 3540
    },
    {
      "epoch": 0.12122560766860664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9548,
      "step": 3541
    },
    {
      "epoch": 0.12125984251968504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9834,
      "step": 3542
    },
    {
      "epoch": 0.12129407737076343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.977,
      "step": 3543
    },
    {
      "epoch": 0.12132831222184183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9676,
      "step": 3544
    },
    {
      "epoch": 0.12136254707292023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9696,
      "step": 3545
    },
    {
      "epoch": 0.12139678192399864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9715,
      "step": 3546
    },
    {
      "epoch": 0.12143101677507703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9451,
      "step": 3547
    },
    {
      "epoch": 0.12146525162615543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9755,
      "step": 3548
    },
    {
      "epoch": 0.12149948647723383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9828,
      "step": 3549
    },
    {
      "epoch": 0.12153372132831222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9765,
      "step": 3550
    },
    {
      "epoch": 0.12156795617939062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9396,
      "step": 3551
    },
    {
      "epoch": 0.12160219103046901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9701,
      "step": 3552
    },
    {
      "epoch": 0.12163642588154741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9538,
      "step": 3553
    },
    {
      "epoch": 0.12167066073262581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9738,
      "step": 3554
    },
    {
      "epoch": 0.12170489558370422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9842,
      "step": 3555
    },
    {
      "epoch": 0.12173913043478261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.968,
      "step": 3556
    },
    {
      "epoch": 0.12177336528586101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9562,
      "step": 3557
    },
    {
      "epoch": 0.1218076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9547,
      "step": 3558
    },
    {
      "epoch": 0.1218418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.984,
      "step": 3559
    },
    {
      "epoch": 0.1218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9701,
      "step": 3560
    },
    {
      "epoch": 0.1219103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9697,
      "step": 3561
    },
    {
      "epoch": 0.12194453954125299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9637,
      "step": 3562
    },
    {
      "epoch": 0.12197877439233139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9841,
      "step": 3563
    },
    {
      "epoch": 0.1220130092434098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9773,
      "step": 3564
    },
    {
      "epoch": 0.1220472440944882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9608,
      "step": 3565
    },
    {
      "epoch": 0.12208147894556659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9724,
      "step": 3566
    },
    {
      "epoch": 0.12211571379664499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9542,
      "step": 3567
    },
    {
      "epoch": 0.12214994864772338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9758,
      "step": 3568
    },
    {
      "epoch": 0.12218418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.958,
      "step": 3569
    },
    {
      "epoch": 0.12221841834988018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9425,
      "step": 3570
    },
    {
      "epoch": 0.12225265320095857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9648,
      "step": 3571
    },
    {
      "epoch": 0.12228688805203697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9698,
      "step": 3572
    },
    {
      "epoch": 0.12232112290311536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9651,
      "step": 3573
    },
    {
      "epoch": 0.12235535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9812,
      "step": 3574
    },
    {
      "epoch": 0.12238959260527217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9507,
      "step": 3575
    },
    {
      "epoch": 0.12242382745635057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.956,
      "step": 3576
    },
    {
      "epoch": 0.12245806230742896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9596,
      "step": 3577
    },
    {
      "epoch": 0.12249229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9615,
      "step": 3578
    },
    {
      "epoch": 0.12252653200958576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9736,
      "step": 3579
    },
    {
      "epoch": 0.12256076686066415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9778,
      "step": 3580
    },
    {
      "epoch": 0.12259500171174255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9569,
      "step": 3581
    },
    {
      "epoch": 0.12262923656282095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9608,
      "step": 3582
    },
    {
      "epoch": 0.12266347141389936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9838,
      "step": 3583
    },
    {
      "epoch": 0.12269770626497775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9994,
      "step": 3584
    },
    {
      "epoch": 0.12273194111605615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9587,
      "step": 3585
    },
    {
      "epoch": 0.12276617596713454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9616,
      "step": 3586
    },
    {
      "epoch": 0.12280041081821294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9722,
      "step": 3587
    },
    {
      "epoch": 0.12283464566929134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9748,
      "step": 3588
    },
    {
      "epoch": 0.12286888052036973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9523,
      "step": 3589
    },
    {
      "epoch": 0.12290311537144813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9746,
      "step": 3590
    },
    {
      "epoch": 0.12293735022252653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9515,
      "step": 3591
    },
    {
      "epoch": 0.12297158507360494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9513,
      "step": 3592
    },
    {
      "epoch": 0.12300581992468333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9755,
      "step": 3593
    },
    {
      "epoch": 0.12304005477576173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9605,
      "step": 3594
    },
    {
      "epoch": 0.12307428962684012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9745,
      "step": 3595
    },
    {
      "epoch": 0.12310852447791852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9676,
      "step": 3596
    },
    {
      "epoch": 0.12314275932899692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9481,
      "step": 3597
    },
    {
      "epoch": 0.12317699418007531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9499,
      "step": 3598
    },
    {
      "epoch": 0.12321122903115371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9707,
      "step": 3599
    },
    {
      "epoch": 0.1232454638822321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9673,
      "step": 3600
    },
    {
      "epoch": 0.12327969873331052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9705,
      "step": 3601
    },
    {
      "epoch": 0.12331393358438891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9567,
      "step": 3602
    },
    {
      "epoch": 0.12334816843546731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9294,
      "step": 3603
    },
    {
      "epoch": 0.1233824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.966,
      "step": 3604
    },
    {
      "epoch": 0.1234166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.961,
      "step": 3605
    },
    {
      "epoch": 0.1234508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9772,
      "step": 3606
    },
    {
      "epoch": 0.1234851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9544,
      "step": 3607
    },
    {
      "epoch": 0.12351934269085929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9581,
      "step": 3608
    },
    {
      "epoch": 0.12355357754193769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9704,
      "step": 3609
    },
    {
      "epoch": 0.1235878123930161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9366,
      "step": 3610
    },
    {
      "epoch": 0.1236220472440945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9439,
      "step": 3611
    },
    {
      "epoch": 0.12365628209517289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9643,
      "step": 3612
    },
    {
      "epoch": 0.12369051694625129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9776,
      "step": 3613
    },
    {
      "epoch": 0.12372475179732968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9615,
      "step": 3614
    },
    {
      "epoch": 0.12375898664840808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9923,
      "step": 3615
    },
    {
      "epoch": 0.12379322149948647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9653,
      "step": 3616
    },
    {
      "epoch": 0.12382745635056487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9539,
      "step": 3617
    },
    {
      "epoch": 0.12386169120164327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9649,
      "step": 3618
    },
    {
      "epoch": 0.12389592605272166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9411,
      "step": 3619
    },
    {
      "epoch": 0.12393016090380007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9396,
      "step": 3620
    },
    {
      "epoch": 0.12396439575487847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9859,
      "step": 3621
    },
    {
      "epoch": 0.12399863060595687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9579,
      "step": 3622
    },
    {
      "epoch": 0.12403286545703526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9425,
      "step": 3623
    },
    {
      "epoch": 0.12406710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9703,
      "step": 3624
    },
    {
      "epoch": 0.12410133515919206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9797,
      "step": 3625
    },
    {
      "epoch": 0.12413557001027045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9658,
      "step": 3626
    },
    {
      "epoch": 0.12416980486134885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9701,
      "step": 3627
    },
    {
      "epoch": 0.12420403971242724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9829,
      "step": 3628
    },
    {
      "epoch": 0.12423827456350565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9614,
      "step": 3629
    },
    {
      "epoch": 0.12427250941458405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9883,
      "step": 3630
    },
    {
      "epoch": 0.12430674426566245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9806,
      "step": 3631
    },
    {
      "epoch": 0.12434097911674084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9689,
      "step": 3632
    },
    {
      "epoch": 0.12437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9518,
      "step": 3633
    },
    {
      "epoch": 0.12440944881889764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9541,
      "step": 3634
    },
    {
      "epoch": 0.12444368366997603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9437,
      "step": 3635
    },
    {
      "epoch": 0.12447791852105443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9543,
      "step": 3636
    },
    {
      "epoch": 0.12451215337213282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9561,
      "step": 3637
    },
    {
      "epoch": 0.12454638822321124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.961,
      "step": 3638
    },
    {
      "epoch": 0.12458062307428963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9568,
      "step": 3639
    },
    {
      "epoch": 0.12461485792536803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9183,
      "step": 3640
    },
    {
      "epoch": 0.12464909277644642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.952,
      "step": 3641
    },
    {
      "epoch": 0.12468332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9431,
      "step": 3642
    },
    {
      "epoch": 0.12471756247860322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9576,
      "step": 3643
    },
    {
      "epoch": 0.12475179732968161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9557,
      "step": 3644
    },
    {
      "epoch": 0.12478603218076001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9628,
      "step": 3645
    },
    {
      "epoch": 0.1248202670318384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9369,
      "step": 3646
    },
    {
      "epoch": 0.12485450188291682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9579,
      "step": 3647
    },
    {
      "epoch": 0.12488873673399521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9561,
      "step": 3648
    },
    {
      "epoch": 0.12492297158507361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9659,
      "step": 3649
    },
    {
      "epoch": 0.124957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.96,
      "step": 3650
    },
    {
      "epoch": 0.1249914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9405,
      "step": 3651
    },
    {
      "epoch": 0.1250256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9501,
      "step": 3652
    },
    {
      "epoch": 0.1250599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9432,
      "step": 3653
    },
    {
      "epoch": 0.1250941458404656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.974,
      "step": 3654
    },
    {
      "epoch": 0.125128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9552,
      "step": 3655
    },
    {
      "epoch": 0.1251626155426224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9454,
      "step": 3656
    },
    {
      "epoch": 0.1251968503937008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9463,
      "step": 3657
    },
    {
      "epoch": 0.1252310852447792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9352,
      "step": 3658
    },
    {
      "epoch": 0.12526532009585759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9717,
      "step": 3659
    },
    {
      "epoch": 0.12529955494693598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9386,
      "step": 3660
    },
    {
      "epoch": 0.12533378979801438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9512,
      "step": 3661
    },
    {
      "epoch": 0.12536802464909277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9277,
      "step": 3662
    },
    {
      "epoch": 0.12540225950017117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9497,
      "step": 3663
    },
    {
      "epoch": 0.12543649435124957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9444,
      "step": 3664
    },
    {
      "epoch": 0.12547072920232796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9479,
      "step": 3665
    },
    {
      "epoch": 0.12550496405340636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9407,
      "step": 3666
    },
    {
      "epoch": 0.12553919890448476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9649,
      "step": 3667
    },
    {
      "epoch": 0.12557343375556315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9228,
      "step": 3668
    },
    {
      "epoch": 0.12560766860664155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9498,
      "step": 3669
    },
    {
      "epoch": 0.12564190345771997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9617,
      "step": 3670
    },
    {
      "epoch": 0.12567613830879837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9496,
      "step": 3671
    },
    {
      "epoch": 0.12571037315987676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.961,
      "step": 3672
    },
    {
      "epoch": 0.12574460801095516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9537,
      "step": 3673
    },
    {
      "epoch": 0.12577884286203356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9529,
      "step": 3674
    },
    {
      "epoch": 0.12581307771311195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9456,
      "step": 3675
    },
    {
      "epoch": 0.12584731256419035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9779,
      "step": 3676
    },
    {
      "epoch": 0.12588154741526875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9641,
      "step": 3677
    },
    {
      "epoch": 0.12591578226634714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9326,
      "step": 3678
    },
    {
      "epoch": 0.12595001711742554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9641,
      "step": 3679
    },
    {
      "epoch": 0.12598425196850394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9262,
      "step": 3680
    },
    {
      "epoch": 0.12601848681958233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9433,
      "step": 3681
    },
    {
      "epoch": 0.12605272167066073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9457,
      "step": 3682
    },
    {
      "epoch": 0.12608695652173912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9548,
      "step": 3683
    },
    {
      "epoch": 0.12612119137281752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9556,
      "step": 3684
    },
    {
      "epoch": 0.12615542622389592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9585,
      "step": 3685
    },
    {
      "epoch": 0.1261896610749743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9673,
      "step": 3686
    },
    {
      "epoch": 0.1262238959260527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9479,
      "step": 3687
    },
    {
      "epoch": 0.1262581307771311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9471,
      "step": 3688
    },
    {
      "epoch": 0.12629236562820953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9555,
      "step": 3689
    },
    {
      "epoch": 0.12632660047928793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9342,
      "step": 3690
    },
    {
      "epoch": 0.12636083533036632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.945,
      "step": 3691
    },
    {
      "epoch": 0.12639507018144472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9519,
      "step": 3692
    },
    {
      "epoch": 0.12642930503252311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9334,
      "step": 3693
    },
    {
      "epoch": 0.1264635398836015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9407,
      "step": 3694
    },
    {
      "epoch": 0.1264977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9451,
      "step": 3695
    },
    {
      "epoch": 0.1265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9607,
      "step": 3696
    },
    {
      "epoch": 0.1265662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9535,
      "step": 3697
    },
    {
      "epoch": 0.1266004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9549,
      "step": 3698
    },
    {
      "epoch": 0.1266347141389935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9595,
      "step": 3699
    },
    {
      "epoch": 0.1266689489900719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9315,
      "step": 3700
    },
    {
      "epoch": 0.12670318384115029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.916,
      "step": 3701
    },
    {
      "epoch": 0.12673741869222868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9652,
      "step": 3702
    },
    {
      "epoch": 0.12677165354330708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9352,
      "step": 3703
    },
    {
      "epoch": 0.12680588839438547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9326,
      "step": 3704
    },
    {
      "epoch": 0.12684012324546387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9363,
      "step": 3705
    },
    {
      "epoch": 0.12687435809654227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9547,
      "step": 3706
    },
    {
      "epoch": 0.1269085929476207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9529,
      "step": 3707
    },
    {
      "epoch": 0.1269428277986991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9432,
      "step": 3708
    },
    {
      "epoch": 0.12697706264977748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9409,
      "step": 3709
    },
    {
      "epoch": 0.12701129750085588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9487,
      "step": 3710
    },
    {
      "epoch": 0.12704553235193428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9439,
      "step": 3711
    },
    {
      "epoch": 0.12707976720301267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9397,
      "step": 3712
    },
    {
      "epoch": 0.12711400205409107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9596,
      "step": 3713
    },
    {
      "epoch": 0.12714823690516946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.935,
      "step": 3714
    },
    {
      "epoch": 0.12718247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9175,
      "step": 3715
    },
    {
      "epoch": 0.12721670660732626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9328,
      "step": 3716
    },
    {
      "epoch": 0.12725094145840465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9208,
      "step": 3717
    },
    {
      "epoch": 0.12728517630948305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9425,
      "step": 3718
    },
    {
      "epoch": 0.12731941116056145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9335,
      "step": 3719
    },
    {
      "epoch": 0.12735364601163984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9211,
      "step": 3720
    },
    {
      "epoch": 0.12738788086271824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.932,
      "step": 3721
    },
    {
      "epoch": 0.12742211571379664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9328,
      "step": 3722
    },
    {
      "epoch": 0.12745635056487503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9533,
      "step": 3723
    },
    {
      "epoch": 0.12749058541595343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9583,
      "step": 3724
    },
    {
      "epoch": 0.12752482026703185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9132,
      "step": 3725
    },
    {
      "epoch": 0.12755905511811025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9268,
      "step": 3726
    },
    {
      "epoch": 0.12759328996918864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9158,
      "step": 3727
    },
    {
      "epoch": 0.12762752482026704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9417,
      "step": 3728
    },
    {
      "epoch": 0.12766175967134544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9502,
      "step": 3729
    },
    {
      "epoch": 0.12769599452242383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9383,
      "step": 3730
    },
    {
      "epoch": 0.12773022937350223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9469,
      "step": 3731
    },
    {
      "epoch": 0.12776446422458063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9349,
      "step": 3732
    },
    {
      "epoch": 0.12779869907565902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9319,
      "step": 3733
    },
    {
      "epoch": 0.12783293392673742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9284,
      "step": 3734
    },
    {
      "epoch": 0.12786716877781582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9613,
      "step": 3735
    },
    {
      "epoch": 0.1279014036288942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9301,
      "step": 3736
    },
    {
      "epoch": 0.1279356384799726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9487,
      "step": 3737
    },
    {
      "epoch": 0.127969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9568,
      "step": 3738
    },
    {
      "epoch": 0.1280041081821294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9428,
      "step": 3739
    },
    {
      "epoch": 0.1280383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9293,
      "step": 3740
    },
    {
      "epoch": 0.1280725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9383,
      "step": 3741
    },
    {
      "epoch": 0.1281068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9369,
      "step": 3742
    },
    {
      "epoch": 0.12814104758644299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9565,
      "step": 3743
    },
    {
      "epoch": 0.1281752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9203,
      "step": 3744
    },
    {
      "epoch": 0.1282095172885998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9334,
      "step": 3745
    },
    {
      "epoch": 0.1282437521396782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.942,
      "step": 3746
    },
    {
      "epoch": 0.1282779869907566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9191,
      "step": 3747
    },
    {
      "epoch": 0.128312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9294,
      "step": 3748
    },
    {
      "epoch": 0.1283464566929134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9338,
      "step": 3749
    },
    {
      "epoch": 0.1283806915439918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9255,
      "step": 3750
    },
    {
      "epoch": 0.12841492639507018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9479,
      "step": 3751
    },
    {
      "epoch": 0.12844916124614858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9631,
      "step": 3752
    },
    {
      "epoch": 0.12848339609722698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9511,
      "step": 3753
    },
    {
      "epoch": 0.12851763094830537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.945,
      "step": 3754
    },
    {
      "epoch": 0.12855186579938377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9294,
      "step": 3755
    },
    {
      "epoch": 0.12858610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9075,
      "step": 3756
    },
    {
      "epoch": 0.12862033550154056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9267,
      "step": 3757
    },
    {
      "epoch": 0.12865457035261896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9361,
      "step": 3758
    },
    {
      "epoch": 0.12868880520369735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9248,
      "step": 3759
    },
    {
      "epoch": 0.12872304005477575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9413,
      "step": 3760
    },
    {
      "epoch": 0.12875727490585415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.925,
      "step": 3761
    },
    {
      "epoch": 0.12879150975693257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9354,
      "step": 3762
    },
    {
      "epoch": 0.12882574460801097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.934,
      "step": 3763
    },
    {
      "epoch": 0.12885997945908936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9246,
      "step": 3764
    },
    {
      "epoch": 0.12889421431016776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9701,
      "step": 3765
    },
    {
      "epoch": 0.12892844916124616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9187,
      "step": 3766
    },
    {
      "epoch": 0.12896268401232455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9052,
      "step": 3767
    },
    {
      "epoch": 0.12899691886340295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9194,
      "step": 3768
    },
    {
      "epoch": 0.12903115371448134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9322,
      "step": 3769
    },
    {
      "epoch": 0.12906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9445,
      "step": 3770
    },
    {
      "epoch": 0.12909962341663814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9108,
      "step": 3771
    },
    {
      "epoch": 0.12913385826771653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9669,
      "step": 3772
    },
    {
      "epoch": 0.12916809311879493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9323,
      "step": 3773
    },
    {
      "epoch": 0.12920232796987333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.947,
      "step": 3774
    },
    {
      "epoch": 0.12923656282095172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.943,
      "step": 3775
    },
    {
      "epoch": 0.12927079767203012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9168,
      "step": 3776
    },
    {
      "epoch": 0.12930503252310852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9453,
      "step": 3777
    },
    {
      "epoch": 0.1293392673741869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9334,
      "step": 3778
    },
    {
      "epoch": 0.1293735022252653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9445,
      "step": 3779
    },
    {
      "epoch": 0.1294077370763437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9441,
      "step": 3780
    },
    {
      "epoch": 0.12944197192742213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9114,
      "step": 3781
    },
    {
      "epoch": 0.12947620677850052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9505,
      "step": 3782
    },
    {
      "epoch": 0.12951044162957892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9443,
      "step": 3783
    },
    {
      "epoch": 0.12954467648065732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9698,
      "step": 3784
    },
    {
      "epoch": 0.1295789113317357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9831,
      "step": 3785
    },
    {
      "epoch": 0.1296131461828141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9472,
      "step": 3786
    },
    {
      "epoch": 0.1296473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9437,
      "step": 3787
    },
    {
      "epoch": 0.1296816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9287,
      "step": 3788
    },
    {
      "epoch": 0.1297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9357,
      "step": 3789
    },
    {
      "epoch": 0.1297500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9733,
      "step": 3790
    },
    {
      "epoch": 0.1297843204382061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9228,
      "step": 3791
    },
    {
      "epoch": 0.1298185552892845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.944,
      "step": 3792
    },
    {
      "epoch": 0.12985279014036288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9444,
      "step": 3793
    },
    {
      "epoch": 0.12988702499144128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9172,
      "step": 3794
    },
    {
      "epoch": 0.12992125984251968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9164,
      "step": 3795
    },
    {
      "epoch": 0.12995549469359807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8861,
      "step": 3796
    },
    {
      "epoch": 0.12998972954467647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9393,
      "step": 3797
    },
    {
      "epoch": 0.13002396439575487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.929,
      "step": 3798
    },
    {
      "epoch": 0.1300581992468333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9236,
      "step": 3799
    },
    {
      "epoch": 0.13009243409791169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9322,
      "step": 3800
    },
    {
      "epoch": 0.13012666894899008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9165,
      "step": 3801
    },
    {
      "epoch": 0.13016090380006848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9203,
      "step": 3802
    },
    {
      "epoch": 0.13019513865114687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9429,
      "step": 3803
    },
    {
      "epoch": 0.13022937350222527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8881,
      "step": 3804
    },
    {
      "epoch": 0.13026360835330367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9354,
      "step": 3805
    },
    {
      "epoch": 0.13029784320438206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9267,
      "step": 3806
    },
    {
      "epoch": 0.13033207805546046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9152,
      "step": 3807
    },
    {
      "epoch": 0.13036631290653886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9375,
      "step": 3808
    },
    {
      "epoch": 0.13040054775761725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9132,
      "step": 3809
    },
    {
      "epoch": 0.13043478260869565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9279,
      "step": 3810
    },
    {
      "epoch": 0.13046901745977404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9043,
      "step": 3811
    },
    {
      "epoch": 0.13050325231085244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9394,
      "step": 3812
    },
    {
      "epoch": 0.13053748716193084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9125,
      "step": 3813
    },
    {
      "epoch": 0.13057172201300923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8929,
      "step": 3814
    },
    {
      "epoch": 0.13060595686408763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9274,
      "step": 3815
    },
    {
      "epoch": 0.13064019171516603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9459,
      "step": 3816
    },
    {
      "epoch": 0.13067442656624445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9245,
      "step": 3817
    },
    {
      "epoch": 0.13070866141732285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9101,
      "step": 3818
    },
    {
      "epoch": 0.13074289626840124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9059,
      "step": 3819
    },
    {
      "epoch": 0.13077713111947964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9215,
      "step": 3820
    },
    {
      "epoch": 0.13081136597055804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9224,
      "step": 3821
    },
    {
      "epoch": 0.13084560082163643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9448,
      "step": 3822
    },
    {
      "epoch": 0.13087983567271483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9107,
      "step": 3823
    },
    {
      "epoch": 0.13091407052379322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9301,
      "step": 3824
    },
    {
      "epoch": 0.13094830537487162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9371,
      "step": 3825
    },
    {
      "epoch": 0.13098254022595002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9093,
      "step": 3826
    },
    {
      "epoch": 0.1310167750770284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9041,
      "step": 3827
    },
    {
      "epoch": 0.1310510099281068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9207,
      "step": 3828
    },
    {
      "epoch": 0.1310852447791852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9276,
      "step": 3829
    },
    {
      "epoch": 0.1311194796302636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9262,
      "step": 3830
    },
    {
      "epoch": 0.131153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9416,
      "step": 3831
    },
    {
      "epoch": 0.1311879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9217,
      "step": 3832
    },
    {
      "epoch": 0.1312221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.904,
      "step": 3833
    },
    {
      "epoch": 0.1312564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8968,
      "step": 3834
    },
    {
      "epoch": 0.13129065388565558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9074,
      "step": 3835
    },
    {
      "epoch": 0.131324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.915,
      "step": 3836
    },
    {
      "epoch": 0.1313591235878124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9107,
      "step": 3837
    },
    {
      "epoch": 0.1313933584388908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9254,
      "step": 3838
    },
    {
      "epoch": 0.1314275932899692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9218,
      "step": 3839
    },
    {
      "epoch": 0.1314618281410476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9014,
      "step": 3840
    },
    {
      "epoch": 0.131496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9417,
      "step": 3841
    },
    {
      "epoch": 0.13153029784320439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9223,
      "step": 3842
    },
    {
      "epoch": 0.13156453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9189,
      "step": 3843
    },
    {
      "epoch": 0.13159876754536118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9172,
      "step": 3844
    },
    {
      "epoch": 0.13163300239643957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.923,
      "step": 3845
    },
    {
      "epoch": 0.13166723724751797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9019,
      "step": 3846
    },
    {
      "epoch": 0.13170147209859637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9161,
      "step": 3847
    },
    {
      "epoch": 0.13173570694967476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9467,
      "step": 3848
    },
    {
      "epoch": 0.13176994180075316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9018,
      "step": 3849
    },
    {
      "epoch": 0.13180417665183156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9022,
      "step": 3850
    },
    {
      "epoch": 0.13183841150290995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8992,
      "step": 3851
    },
    {
      "epoch": 0.13187264635398835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9134,
      "step": 3852
    },
    {
      "epoch": 0.13190688120506674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9201,
      "step": 3853
    },
    {
      "epoch": 0.13194111605614517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9323,
      "step": 3854
    },
    {
      "epoch": 0.13197535090722357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9171,
      "step": 3855
    },
    {
      "epoch": 0.13200958575830196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9088,
      "step": 3856
    },
    {
      "epoch": 0.13204382060938036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9055,
      "step": 3857
    },
    {
      "epoch": 0.13207805546045875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9159,
      "step": 3858
    },
    {
      "epoch": 0.13211229031153715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9183,
      "step": 3859
    },
    {
      "epoch": 0.13214652516261555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9296,
      "step": 3860
    },
    {
      "epoch": 0.13218076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8843,
      "step": 3861
    },
    {
      "epoch": 0.13221499486477234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9232,
      "step": 3862
    },
    {
      "epoch": 0.13224922971585074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9115,
      "step": 3863
    },
    {
      "epoch": 0.13228346456692913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9173,
      "step": 3864
    },
    {
      "epoch": 0.13231769941800753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8947,
      "step": 3865
    },
    {
      "epoch": 0.13235193426908592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9083,
      "step": 3866
    },
    {
      "epoch": 0.13238616912016432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.894,
      "step": 3867
    },
    {
      "epoch": 0.13242040397124272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9068,
      "step": 3868
    },
    {
      "epoch": 0.1324546388223211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9116,
      "step": 3869
    },
    {
      "epoch": 0.1324888736733995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9224,
      "step": 3870
    },
    {
      "epoch": 0.1325231085244779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9038,
      "step": 3871
    },
    {
      "epoch": 0.13255734337555633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9055,
      "step": 3872
    },
    {
      "epoch": 0.13259157822663473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9297,
      "step": 3873
    },
    {
      "epoch": 0.13262581307771312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8973,
      "step": 3874
    },
    {
      "epoch": 0.13266004792879152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8912,
      "step": 3875
    },
    {
      "epoch": 0.13269428277986992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9129,
      "step": 3876
    },
    {
      "epoch": 0.1327285176309483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9311,
      "step": 3877
    },
    {
      "epoch": 0.1327627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9507,
      "step": 3878
    },
    {
      "epoch": 0.1327969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9152,
      "step": 3879
    },
    {
      "epoch": 0.1328312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9132,
      "step": 3880
    },
    {
      "epoch": 0.1328654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.903,
      "step": 3881
    },
    {
      "epoch": 0.1328996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9126,
      "step": 3882
    },
    {
      "epoch": 0.1329339267374187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8993,
      "step": 3883
    },
    {
      "epoch": 0.13296816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9199,
      "step": 3884
    },
    {
      "epoch": 0.13300239643957548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9205,
      "step": 3885
    },
    {
      "epoch": 0.13303663129065388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9129,
      "step": 3886
    },
    {
      "epoch": 0.13307086614173227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9246,
      "step": 3887
    },
    {
      "epoch": 0.13310510099281067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9215,
      "step": 3888
    },
    {
      "epoch": 0.13313933584388907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9169,
      "step": 3889
    },
    {
      "epoch": 0.13317357069496746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8993,
      "step": 3890
    },
    {
      "epoch": 0.1332078055460459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8993,
      "step": 3891
    },
    {
      "epoch": 0.13324204039712428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8932,
      "step": 3892
    },
    {
      "epoch": 0.13327627524820268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9129,
      "step": 3893
    },
    {
      "epoch": 0.13331051009928108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9011,
      "step": 3894
    },
    {
      "epoch": 0.13334474495035947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9235,
      "step": 3895
    },
    {
      "epoch": 0.13337897980143787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9099,
      "step": 3896
    },
    {
      "epoch": 0.13341321465251627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8953,
      "step": 3897
    },
    {
      "epoch": 0.13344744950359466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9142,
      "step": 3898
    },
    {
      "epoch": 0.13348168435467306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9298,
      "step": 3899
    },
    {
      "epoch": 0.13351591920575145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9019,
      "step": 3900
    },
    {
      "epoch": 0.13355015405682985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9094,
      "step": 3901
    },
    {
      "epoch": 0.13358438890790825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.94,
      "step": 3902
    },
    {
      "epoch": 0.13361862375898664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9018,
      "step": 3903
    },
    {
      "epoch": 0.13365285861006504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9202,
      "step": 3904
    },
    {
      "epoch": 0.13368709346114344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8864,
      "step": 3905
    },
    {
      "epoch": 0.13372132831222183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8984,
      "step": 3906
    },
    {
      "epoch": 0.13375556316330023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8875,
      "step": 3907
    },
    {
      "epoch": 0.13378979801437862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.915,
      "step": 3908
    },
    {
      "epoch": 0.13382403286545705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9139,
      "step": 3909
    },
    {
      "epoch": 0.13385826771653545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9143,
      "step": 3910
    },
    {
      "epoch": 0.13389250256761384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9036,
      "step": 3911
    },
    {
      "epoch": 0.13392673741869224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9233,
      "step": 3912
    },
    {
      "epoch": 0.13396097226977063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9314,
      "step": 3913
    },
    {
      "epoch": 0.13399520712084903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9218,
      "step": 3914
    },
    {
      "epoch": 0.13402944197192743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8916,
      "step": 3915
    },
    {
      "epoch": 0.13406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9288,
      "step": 3916
    },
    {
      "epoch": 0.13409791167408422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8853,
      "step": 3917
    },
    {
      "epoch": 0.13413214652516262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9054,
      "step": 3918
    },
    {
      "epoch": 0.134166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9221,
      "step": 3919
    },
    {
      "epoch": 0.1342006162273194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9,
      "step": 3920
    },
    {
      "epoch": 0.1342348510783978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9058,
      "step": 3921
    },
    {
      "epoch": 0.1342690859294762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8872,
      "step": 3922
    },
    {
      "epoch": 0.1343033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8925,
      "step": 3923
    },
    {
      "epoch": 0.134337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8991,
      "step": 3924
    },
    {
      "epoch": 0.1343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9032,
      "step": 3925
    },
    {
      "epoch": 0.13440602533378979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9076,
      "step": 3926
    },
    {
      "epoch": 0.13444026018486818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8969,
      "step": 3927
    },
    {
      "epoch": 0.1344744950359466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9177,
      "step": 3928
    },
    {
      "epoch": 0.134508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9005,
      "step": 3929
    },
    {
      "epoch": 0.1345429647381034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9255,
      "step": 3930
    },
    {
      "epoch": 0.1345771995891818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8993,
      "step": 3931
    },
    {
      "epoch": 0.1346114344402602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8964,
      "step": 3932
    },
    {
      "epoch": 0.1346456692913386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.891,
      "step": 3933
    },
    {
      "epoch": 0.13467990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9138,
      "step": 3934
    },
    {
      "epoch": 0.13471413899349538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9074,
      "step": 3935
    },
    {
      "epoch": 0.13474837384457378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9267,
      "step": 3936
    },
    {
      "epoch": 0.13478260869565217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9113,
      "step": 3937
    },
    {
      "epoch": 0.13481684354673057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8828,
      "step": 3938
    },
    {
      "epoch": 0.13485107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9275,
      "step": 3939
    },
    {
      "epoch": 0.13488531324888736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9064,
      "step": 3940
    },
    {
      "epoch": 0.13491954809996576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9156,
      "step": 3941
    },
    {
      "epoch": 0.13495378295104415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8859,
      "step": 3942
    },
    {
      "epoch": 0.13498801780212255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9111,
      "step": 3943
    },
    {
      "epoch": 0.13502225265320095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9123,
      "step": 3944
    },
    {
      "epoch": 0.13505648750427934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9224,
      "step": 3945
    },
    {
      "epoch": 0.13509072235535777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9129,
      "step": 3946
    },
    {
      "epoch": 0.13512495720643616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8955,
      "step": 3947
    },
    {
      "epoch": 0.13515919205751456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9543,
      "step": 3948
    },
    {
      "epoch": 0.13519342690859296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9238,
      "step": 3949
    },
    {
      "epoch": 0.13522766175967135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9143,
      "step": 3950
    },
    {
      "epoch": 0.13526189661074975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9268,
      "step": 3951
    },
    {
      "epoch": 0.13529613146182815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9454,
      "step": 3952
    },
    {
      "epoch": 0.13533036631290654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9074,
      "step": 3953
    },
    {
      "epoch": 0.13536460116398494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9154,
      "step": 3954
    },
    {
      "epoch": 0.13539883601506333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8836,
      "step": 3955
    },
    {
      "epoch": 0.13543307086614173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9041,
      "step": 3956
    },
    {
      "epoch": 0.13546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9142,
      "step": 3957
    },
    {
      "epoch": 0.13550154056829852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.912,
      "step": 3958
    },
    {
      "epoch": 0.13553577541937692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9042,
      "step": 3959
    },
    {
      "epoch": 0.13557001027045532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9069,
      "step": 3960
    },
    {
      "epoch": 0.1356042451215337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9082,
      "step": 3961
    },
    {
      "epoch": 0.1356384799726121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9012,
      "step": 3962
    },
    {
      "epoch": 0.1356727148236905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9134,
      "step": 3963
    },
    {
      "epoch": 0.13570694967476893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9077,
      "step": 3964
    },
    {
      "epoch": 0.13574118452584732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9108,
      "step": 3965
    },
    {
      "epoch": 0.13577541937692572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9034,
      "step": 3966
    },
    {
      "epoch": 0.13580965422800412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8967,
      "step": 3967
    },
    {
      "epoch": 0.1358438890790825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8977,
      "step": 3968
    },
    {
      "epoch": 0.1358781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9163,
      "step": 3969
    },
    {
      "epoch": 0.1359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8918,
      "step": 3970
    },
    {
      "epoch": 0.1359465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9157,
      "step": 3971
    },
    {
      "epoch": 0.1359808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9025,
      "step": 3972
    },
    {
      "epoch": 0.1360150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8776,
      "step": 3973
    },
    {
      "epoch": 0.1360492981855529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.913,
      "step": 3974
    },
    {
      "epoch": 0.1360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9349,
      "step": 3975
    },
    {
      "epoch": 0.13611776788770968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8983,
      "step": 3976
    },
    {
      "epoch": 0.13615200273878808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8963,
      "step": 3977
    },
    {
      "epoch": 0.13618623758986648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8737,
      "step": 3978
    },
    {
      "epoch": 0.13622047244094487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9112,
      "step": 3979
    },
    {
      "epoch": 0.13625470729202327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9011,
      "step": 3980
    },
    {
      "epoch": 0.13628894214310167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8731,
      "step": 3981
    },
    {
      "epoch": 0.13632317699418006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8812,
      "step": 3982
    },
    {
      "epoch": 0.13635741184525849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8877,
      "step": 3983
    },
    {
      "epoch": 0.13639164669633688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8881,
      "step": 3984
    },
    {
      "epoch": 0.13642588154741528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8776,
      "step": 3985
    },
    {
      "epoch": 0.13646011639849367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9006,
      "step": 3986
    },
    {
      "epoch": 0.13649435124957207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9252,
      "step": 3987
    },
    {
      "epoch": 0.13652858610065047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8941,
      "step": 3988
    },
    {
      "epoch": 0.13656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9013,
      "step": 3989
    },
    {
      "epoch": 0.13659705580280726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8798,
      "step": 3990
    },
    {
      "epoch": 0.13663129065388566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9027,
      "step": 3991
    },
    {
      "epoch": 0.13666552550496405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.886,
      "step": 3992
    },
    {
      "epoch": 0.13669976035604245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9132,
      "step": 3993
    },
    {
      "epoch": 0.13673399520712085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8875,
      "step": 3994
    },
    {
      "epoch": 0.13676823005819924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9033,
      "step": 3995
    },
    {
      "epoch": 0.13680246490927764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8733,
      "step": 3996
    },
    {
      "epoch": 0.13683669976035603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9017,
      "step": 3997
    },
    {
      "epoch": 0.13687093461143443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8887,
      "step": 3998
    },
    {
      "epoch": 0.13690516946251283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8806,
      "step": 3999
    },
    {
      "epoch": 0.13693940431359122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8607,
      "step": 4000
    },
    {
      "epoch": 0.13697363916466965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8928,
      "step": 4001
    },
    {
      "epoch": 0.13700787401574804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8969,
      "step": 4002
    },
    {
      "epoch": 0.13704210886682644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8881,
      "step": 4003
    },
    {
      "epoch": 0.13707634371790484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8814,
      "step": 4004
    },
    {
      "epoch": 0.13711057856898323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9126,
      "step": 4005
    },
    {
      "epoch": 0.13714481342006163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9202,
      "step": 4006
    },
    {
      "epoch": 0.13717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9085,
      "step": 4007
    },
    {
      "epoch": 0.13721328312221842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8851,
      "step": 4008
    },
    {
      "epoch": 0.13724751797329682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9057,
      "step": 4009
    },
    {
      "epoch": 0.1372817528243752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9099,
      "step": 4010
    },
    {
      "epoch": 0.1373159876754536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8739,
      "step": 4011
    },
    {
      "epoch": 0.137350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8717,
      "step": 4012
    },
    {
      "epoch": 0.1373844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8797,
      "step": 4013
    },
    {
      "epoch": 0.1374186922286888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8885,
      "step": 4014
    },
    {
      "epoch": 0.1374529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8704,
      "step": 4015
    },
    {
      "epoch": 0.1374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8824,
      "step": 4016
    },
    {
      "epoch": 0.137521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.913,
      "step": 4017
    },
    {
      "epoch": 0.13755563163300238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9088,
      "step": 4018
    },
    {
      "epoch": 0.13758986648408078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8857,
      "step": 4019
    },
    {
      "epoch": 0.1376241013351592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9059,
      "step": 4020
    },
    {
      "epoch": 0.1376583361862376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8684,
      "step": 4021
    },
    {
      "epoch": 0.137692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9048,
      "step": 4022
    },
    {
      "epoch": 0.1377268058883944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8841,
      "step": 4023
    },
    {
      "epoch": 0.1377610407394728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8956,
      "step": 4024
    },
    {
      "epoch": 0.1377952755905512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9033,
      "step": 4025
    },
    {
      "epoch": 0.13782951044162958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9147,
      "step": 4026
    },
    {
      "epoch": 0.13786374529270798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8897,
      "step": 4027
    },
    {
      "epoch": 0.13789798014378638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.89,
      "step": 4028
    },
    {
      "epoch": 0.13793221499486477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8903,
      "step": 4029
    },
    {
      "epoch": 0.13796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8792,
      "step": 4030
    },
    {
      "epoch": 0.13800068469702156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8625,
      "step": 4031
    },
    {
      "epoch": 0.13803491954809996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8833,
      "step": 4032
    },
    {
      "epoch": 0.13806915439917836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8826,
      "step": 4033
    },
    {
      "epoch": 0.13810338925025675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8759,
      "step": 4034
    },
    {
      "epoch": 0.13813762410133515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8702,
      "step": 4035
    },
    {
      "epoch": 0.13817185895241355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.908,
      "step": 4036
    },
    {
      "epoch": 0.13820609380349194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8899,
      "step": 4037
    },
    {
      "epoch": 0.13824032865457037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9119,
      "step": 4038
    },
    {
      "epoch": 0.13827456350564876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9121,
      "step": 4039
    },
    {
      "epoch": 0.13830879835672716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8802,
      "step": 4040
    },
    {
      "epoch": 0.13834303320780555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8875,
      "step": 4041
    },
    {
      "epoch": 0.13837726805888395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8923,
      "step": 4042
    },
    {
      "epoch": 0.13841150290996235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8718,
      "step": 4043
    },
    {
      "epoch": 0.13844573776104074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.888,
      "step": 4044
    },
    {
      "epoch": 0.13847997261211914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8792,
      "step": 4045
    },
    {
      "epoch": 0.13851420746319754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8681,
      "step": 4046
    },
    {
      "epoch": 0.13854844231427593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.893,
      "step": 4047
    },
    {
      "epoch": 0.13858267716535433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8716,
      "step": 4048
    },
    {
      "epoch": 0.13861691201643273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8928,
      "step": 4049
    },
    {
      "epoch": 0.13865114686751112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8835,
      "step": 4050
    },
    {
      "epoch": 0.13868538171858952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8836,
      "step": 4051
    },
    {
      "epoch": 0.13871961656966791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8809,
      "step": 4052
    },
    {
      "epoch": 0.1387538514207463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8537,
      "step": 4053
    },
    {
      "epoch": 0.1387880862718247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8705,
      "step": 4054
    },
    {
      "epoch": 0.1388223211229031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8847,
      "step": 4055
    },
    {
      "epoch": 0.13885655597398153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8928,
      "step": 4056
    },
    {
      "epoch": 0.13889079082505992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8764,
      "step": 4057
    },
    {
      "epoch": 0.13892502567613832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9139,
      "step": 4058
    },
    {
      "epoch": 0.13895926052721672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8896,
      "step": 4059
    },
    {
      "epoch": 0.1389934953782951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.885,
      "step": 4060
    },
    {
      "epoch": 0.1390277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8588,
      "step": 4061
    },
    {
      "epoch": 0.1390619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8785,
      "step": 4062
    },
    {
      "epoch": 0.1390961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8823,
      "step": 4063
    },
    {
      "epoch": 0.1391304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9069,
      "step": 4064
    },
    {
      "epoch": 0.1391646696336871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8688,
      "step": 4065
    },
    {
      "epoch": 0.1391989044847655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8776,
      "step": 4066
    },
    {
      "epoch": 0.1392331393358439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8663,
      "step": 4067
    },
    {
      "epoch": 0.13926737418692228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8647,
      "step": 4068
    },
    {
      "epoch": 0.13930160903800068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8919,
      "step": 4069
    },
    {
      "epoch": 0.13933584388907908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8942,
      "step": 4070
    },
    {
      "epoch": 0.13937007874015747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9174,
      "step": 4071
    },
    {
      "epoch": 0.13940431359123587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8906,
      "step": 4072
    },
    {
      "epoch": 0.13943854844231426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8581,
      "step": 4073
    },
    {
      "epoch": 0.13947278329339266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8847,
      "step": 4074
    },
    {
      "epoch": 0.13950701814447108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8938,
      "step": 4075
    },
    {
      "epoch": 0.13954125299554948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8941,
      "step": 4076
    },
    {
      "epoch": 0.13957548784662788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8993,
      "step": 4077
    },
    {
      "epoch": 0.13960972269770627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8927,
      "step": 4078
    },
    {
      "epoch": 0.13964395754878467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8894,
      "step": 4079
    },
    {
      "epoch": 0.13967819239986307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9007,
      "step": 4080
    },
    {
      "epoch": 0.13971242725094146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8872,
      "step": 4081
    },
    {
      "epoch": 0.13974666210201986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8692,
      "step": 4082
    },
    {
      "epoch": 0.13978089695309825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8669,
      "step": 4083
    },
    {
      "epoch": 0.13981513180417665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8772,
      "step": 4084
    },
    {
      "epoch": 0.13984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8723,
      "step": 4085
    },
    {
      "epoch": 0.13988360150633344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.888,
      "step": 4086
    },
    {
      "epoch": 0.13991783635741184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8954,
      "step": 4087
    },
    {
      "epoch": 0.13995207120849024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8703,
      "step": 4088
    },
    {
      "epoch": 0.13998630605956863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8459,
      "step": 4089
    },
    {
      "epoch": 0.14002054091064703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8639,
      "step": 4090
    },
    {
      "epoch": 0.14005477576172543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8838,
      "step": 4091
    },
    {
      "epoch": 0.14008901061280382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8707,
      "step": 4092
    },
    {
      "epoch": 0.14012324546388225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8919,
      "step": 4093
    },
    {
      "epoch": 0.14015748031496064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8859,
      "step": 4094
    },
    {
      "epoch": 0.14019171516603904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8637,
      "step": 4095
    },
    {
      "epoch": 0.14022595001711743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9048,
      "step": 4096
    },
    {
      "epoch": 0.14026018486819583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9017,
      "step": 4097
    },
    {
      "epoch": 0.14029441971927423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8646,
      "step": 4098
    },
    {
      "epoch": 0.14032865457035262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8651,
      "step": 4099
    },
    {
      "epoch": 0.14036288942143102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8731,
      "step": 4100
    },
    {
      "epoch": 0.14039712427250942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8913,
      "step": 4101
    },
    {
      "epoch": 0.1404313591235878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8694,
      "step": 4102
    },
    {
      "epoch": 0.1404655939746662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8771,
      "step": 4103
    },
    {
      "epoch": 0.1404998288257446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8967,
      "step": 4104
    },
    {
      "epoch": 0.140534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8338,
      "step": 4105
    },
    {
      "epoch": 0.1405682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8489,
      "step": 4106
    },
    {
      "epoch": 0.1406025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8748,
      "step": 4107
    },
    {
      "epoch": 0.1406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8261,
      "step": 4108
    },
    {
      "epoch": 0.1406710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8723,
      "step": 4109
    },
    {
      "epoch": 0.14070523793221498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.885,
      "step": 4110
    },
    {
      "epoch": 0.14073947278329338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8767,
      "step": 4111
    },
    {
      "epoch": 0.1407737076343718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8718,
      "step": 4112
    },
    {
      "epoch": 0.1408079424854502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8827,
      "step": 4113
    },
    {
      "epoch": 0.1408421773365286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8505,
      "step": 4114
    },
    {
      "epoch": 0.140876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8617,
      "step": 4115
    },
    {
      "epoch": 0.1409106470386854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8655,
      "step": 4116
    },
    {
      "epoch": 0.14094488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8592,
      "step": 4117
    },
    {
      "epoch": 0.14097911674084218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8743,
      "step": 4118
    },
    {
      "epoch": 0.14101335159192058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8815,
      "step": 4119
    },
    {
      "epoch": 0.14104758644299897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8781,
      "step": 4120
    },
    {
      "epoch": 0.14108182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8843,
      "step": 4121
    },
    {
      "epoch": 0.14111605614515577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8741,
      "step": 4122
    },
    {
      "epoch": 0.14115029099623416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8467,
      "step": 4123
    },
    {
      "epoch": 0.14118452584731256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8638,
      "step": 4124
    },
    {
      "epoch": 0.14121876069839095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8848,
      "step": 4125
    },
    {
      "epoch": 0.14125299554946935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8377,
      "step": 4126
    },
    {
      "epoch": 0.14128723040054775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8532,
      "step": 4127
    },
    {
      "epoch": 0.14132146525162614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8881,
      "step": 4128
    },
    {
      "epoch": 0.14135570010270454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8647,
      "step": 4129
    },
    {
      "epoch": 0.14138993495378296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.86,
      "step": 4130
    },
    {
      "epoch": 0.14142416980486136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.846,
      "step": 4131
    },
    {
      "epoch": 0.14145840465593976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8881,
      "step": 4132
    },
    {
      "epoch": 0.14149263950701815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8647,
      "step": 4133
    },
    {
      "epoch": 0.14152687435809655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8778,
      "step": 4134
    },
    {
      "epoch": 0.14156110920917495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8934,
      "step": 4135
    },
    {
      "epoch": 0.14159534406025334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8747,
      "step": 4136
    },
    {
      "epoch": 0.14162957891133174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8793,
      "step": 4137
    },
    {
      "epoch": 0.14166381376241013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8843,
      "step": 4138
    },
    {
      "epoch": 0.14169804861348853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8754,
      "step": 4139
    },
    {
      "epoch": 0.14173228346456693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8847,
      "step": 4140
    },
    {
      "epoch": 0.14176651831564532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8486,
      "step": 4141
    },
    {
      "epoch": 0.14180075316672372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.874,
      "step": 4142
    },
    {
      "epoch": 0.14183498801780212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9025,
      "step": 4143
    },
    {
      "epoch": 0.1418692228688805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8377,
      "step": 4144
    },
    {
      "epoch": 0.1419034577199589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8416,
      "step": 4145
    },
    {
      "epoch": 0.1419376925710373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8629,
      "step": 4146
    },
    {
      "epoch": 0.1419719274221157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8748,
      "step": 4147
    },
    {
      "epoch": 0.14200616227319413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.883,
      "step": 4148
    },
    {
      "epoch": 0.14204039712427252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8739,
      "step": 4149
    },
    {
      "epoch": 0.14207463197535092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.864,
      "step": 4150
    },
    {
      "epoch": 0.14210886682642931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.898,
      "step": 4151
    },
    {
      "epoch": 0.1421431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.85,
      "step": 4152
    },
    {
      "epoch": 0.1421773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.865,
      "step": 4153
    },
    {
      "epoch": 0.1422115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8446,
      "step": 4154
    },
    {
      "epoch": 0.1422458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8894,
      "step": 4155
    },
    {
      "epoch": 0.1422800410818213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8638,
      "step": 4156
    },
    {
      "epoch": 0.1423142759328997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8635,
      "step": 4157
    },
    {
      "epoch": 0.1423485107839781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8903,
      "step": 4158
    },
    {
      "epoch": 0.14238274563505648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9054,
      "step": 4159
    },
    {
      "epoch": 0.14241698048613488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8505,
      "step": 4160
    },
    {
      "epoch": 0.14245121533721328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8527,
      "step": 4161
    },
    {
      "epoch": 0.14248545018829167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.88,
      "step": 4162
    },
    {
      "epoch": 0.14251968503937007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8751,
      "step": 4163
    },
    {
      "epoch": 0.14255391989044847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8566,
      "step": 4164
    },
    {
      "epoch": 0.14258815474152686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8763,
      "step": 4165
    },
    {
      "epoch": 0.14262238959260526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.861,
      "step": 4166
    },
    {
      "epoch": 0.14265662444368368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8561,
      "step": 4167
    },
    {
      "epoch": 0.14269085929476208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8674,
      "step": 4168
    },
    {
      "epoch": 0.14272509414584048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8854,
      "step": 4169
    },
    {
      "epoch": 0.14275932899691887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8774,
      "step": 4170
    },
    {
      "epoch": 0.14279356384799727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8995,
      "step": 4171
    },
    {
      "epoch": 0.14282779869907566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8939,
      "step": 4172
    },
    {
      "epoch": 0.14286203355015406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9071,
      "step": 4173
    },
    {
      "epoch": 0.14289626840123246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8505,
      "step": 4174
    },
    {
      "epoch": 0.14293050325231085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8676,
      "step": 4175
    },
    {
      "epoch": 0.14296473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8572,
      "step": 4176
    },
    {
      "epoch": 0.14299897295446765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8678,
      "step": 4177
    },
    {
      "epoch": 0.14303320780554604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8919,
      "step": 4178
    },
    {
      "epoch": 0.14306744265662444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8813,
      "step": 4179
    },
    {
      "epoch": 0.14310167750770283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8588,
      "step": 4180
    },
    {
      "epoch": 0.14313591235878123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8685,
      "step": 4181
    },
    {
      "epoch": 0.14317014720985963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8618,
      "step": 4182
    },
    {
      "epoch": 0.14320438206093802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8874,
      "step": 4183
    },
    {
      "epoch": 0.14323861691201642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.84,
      "step": 4184
    },
    {
      "epoch": 0.14327285176309484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8699,
      "step": 4185
    },
    {
      "epoch": 0.14330708661417324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8694,
      "step": 4186
    },
    {
      "epoch": 0.14334132146525164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9088,
      "step": 4187
    },
    {
      "epoch": 0.14337555631633003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8325,
      "step": 4188
    },
    {
      "epoch": 0.14340979116740843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8666,
      "step": 4189
    },
    {
      "epoch": 0.14344402601848683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8605,
      "step": 4190
    },
    {
      "epoch": 0.14347826086956522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8659,
      "step": 4191
    },
    {
      "epoch": 0.14351249572064362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8665,
      "step": 4192
    },
    {
      "epoch": 0.14354673057172201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8674,
      "step": 4193
    },
    {
      "epoch": 0.1435809654228004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8809,
      "step": 4194
    },
    {
      "epoch": 0.1436152002738788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8602,
      "step": 4195
    },
    {
      "epoch": 0.1436494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8696,
      "step": 4196
    },
    {
      "epoch": 0.1436836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8746,
      "step": 4197
    },
    {
      "epoch": 0.143717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8651,
      "step": 4198
    },
    {
      "epoch": 0.1437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8887,
      "step": 4199
    },
    {
      "epoch": 0.1437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8709,
      "step": 4200
    },
    {
      "epoch": 0.14382060938034918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8793,
      "step": 4201
    },
    {
      "epoch": 0.14385484423142758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8472,
      "step": 4202
    },
    {
      "epoch": 0.14388907908250598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8845,
      "step": 4203
    },
    {
      "epoch": 0.1439233139335844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8764,
      "step": 4204
    },
    {
      "epoch": 0.1439575487846628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8525,
      "step": 4205
    },
    {
      "epoch": 0.1439917836357412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8754,
      "step": 4206
    },
    {
      "epoch": 0.1440260184868196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8857,
      "step": 4207
    },
    {
      "epoch": 0.144060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8927,
      "step": 4208
    },
    {
      "epoch": 0.14409448818897638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8704,
      "step": 4209
    },
    {
      "epoch": 0.14412872304005478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8463,
      "step": 4210
    },
    {
      "epoch": 0.14416295789113318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8673,
      "step": 4211
    },
    {
      "epoch": 0.14419719274221157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8402,
      "step": 4212
    },
    {
      "epoch": 0.14423142759328997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8757,
      "step": 4213
    },
    {
      "epoch": 0.14426566244436836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8848,
      "step": 4214
    },
    {
      "epoch": 0.14429989729544676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8595,
      "step": 4215
    },
    {
      "epoch": 0.14433413214652516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8563,
      "step": 4216
    },
    {
      "epoch": 0.14436836699760355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8696,
      "step": 4217
    },
    {
      "epoch": 0.14440260184868195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8632,
      "step": 4218
    },
    {
      "epoch": 0.14443683669976035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8677,
      "step": 4219
    },
    {
      "epoch": 0.14447107155083874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8625,
      "step": 4220
    },
    {
      "epoch": 0.14450530640191714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8796,
      "step": 4221
    },
    {
      "epoch": 0.14453954125299556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8544,
      "step": 4222
    },
    {
      "epoch": 0.14457377610407396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8507,
      "step": 4223
    },
    {
      "epoch": 0.14460801095515236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8298,
      "step": 4224
    },
    {
      "epoch": 0.14464224580623075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8508,
      "step": 4225
    },
    {
      "epoch": 0.14467648065730915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8696,
      "step": 4226
    },
    {
      "epoch": 0.14471071550838754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8602,
      "step": 4227
    },
    {
      "epoch": 0.14474495035946594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8521,
      "step": 4228
    },
    {
      "epoch": 0.14477918521054434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8554,
      "step": 4229
    },
    {
      "epoch": 0.14481342006162273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8432,
      "step": 4230
    },
    {
      "epoch": 0.14484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8723,
      "step": 4231
    },
    {
      "epoch": 0.14488188976377953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.849,
      "step": 4232
    },
    {
      "epoch": 0.14491612461485792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8492,
      "step": 4233
    },
    {
      "epoch": 0.14495035946593632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8557,
      "step": 4234
    },
    {
      "epoch": 0.14498459431701471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8443,
      "step": 4235
    },
    {
      "epoch": 0.1450188291680931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.848,
      "step": 4236
    },
    {
      "epoch": 0.1450530640191715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8312,
      "step": 4237
    },
    {
      "epoch": 0.1450872988702499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8399,
      "step": 4238
    },
    {
      "epoch": 0.1451215337213283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8598,
      "step": 4239
    },
    {
      "epoch": 0.14515576857240672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.84,
      "step": 4240
    },
    {
      "epoch": 0.14519000342348512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8415,
      "step": 4241
    },
    {
      "epoch": 0.14522423827456352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.828,
      "step": 4242
    },
    {
      "epoch": 0.1452584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8886,
      "step": 4243
    },
    {
      "epoch": 0.1452927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8654,
      "step": 4244
    },
    {
      "epoch": 0.1453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8529,
      "step": 4245
    },
    {
      "epoch": 0.1453611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8544,
      "step": 4246
    },
    {
      "epoch": 0.1453954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8673,
      "step": 4247
    },
    {
      "epoch": 0.1454296473810339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8703,
      "step": 4248
    },
    {
      "epoch": 0.1454638822321123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8802,
      "step": 4249
    },
    {
      "epoch": 0.1454981170831907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8689,
      "step": 4250
    },
    {
      "epoch": 0.14553235193426908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8272,
      "step": 4251
    },
    {
      "epoch": 0.14556658678534748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.856,
      "step": 4252
    },
    {
      "epoch": 0.14560082163642588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8451,
      "step": 4253
    },
    {
      "epoch": 0.14563505648750427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8421,
      "step": 4254
    },
    {
      "epoch": 0.14566929133858267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8528,
      "step": 4255
    },
    {
      "epoch": 0.14570352618966106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8795,
      "step": 4256
    },
    {
      "epoch": 0.14573776104073946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8623,
      "step": 4257
    },
    {
      "epoch": 0.14577199589181786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8486,
      "step": 4258
    },
    {
      "epoch": 0.14580623074289628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8802,
      "step": 4259
    },
    {
      "epoch": 0.14584046559397468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8346,
      "step": 4260
    },
    {
      "epoch": 0.14587470044505307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8624,
      "step": 4261
    },
    {
      "epoch": 0.14590893529613147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8569,
      "step": 4262
    },
    {
      "epoch": 0.14594317014720987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8353,
      "step": 4263
    },
    {
      "epoch": 0.14597740499828826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8652,
      "step": 4264
    },
    {
      "epoch": 0.14601163984936666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8376,
      "step": 4265
    },
    {
      "epoch": 0.14604587470044506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8745,
      "step": 4266
    },
    {
      "epoch": 0.14608010955152345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8615,
      "step": 4267
    },
    {
      "epoch": 0.14611434440260185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8438,
      "step": 4268
    },
    {
      "epoch": 0.14614857925368024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8713,
      "step": 4269
    },
    {
      "epoch": 0.14618281410475864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8544,
      "step": 4270
    },
    {
      "epoch": 0.14621704895583704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8796,
      "step": 4271
    },
    {
      "epoch": 0.14625128380691543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8685,
      "step": 4272
    },
    {
      "epoch": 0.14628551865799383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8532,
      "step": 4273
    },
    {
      "epoch": 0.14631975350907223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8603,
      "step": 4274
    },
    {
      "epoch": 0.14635398836015062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8648,
      "step": 4275
    },
    {
      "epoch": 0.14638822321122902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.853,
      "step": 4276
    },
    {
      "epoch": 0.14642245806230744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.862,
      "step": 4277
    },
    {
      "epoch": 0.14645669291338584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8436,
      "step": 4278
    },
    {
      "epoch": 0.14649092776446423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8642,
      "step": 4279
    },
    {
      "epoch": 0.14652516261554263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8531,
      "step": 4280
    },
    {
      "epoch": 0.14655939746662103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8423,
      "step": 4281
    },
    {
      "epoch": 0.14659363231769942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8584,
      "step": 4282
    },
    {
      "epoch": 0.14662786716877782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8411,
      "step": 4283
    },
    {
      "epoch": 0.14666210201985622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8229,
      "step": 4284
    },
    {
      "epoch": 0.1466963368709346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8759,
      "step": 4285
    },
    {
      "epoch": 0.146730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8565,
      "step": 4286
    },
    {
      "epoch": 0.1467648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8502,
      "step": 4287
    },
    {
      "epoch": 0.1467990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8452,
      "step": 4288
    },
    {
      "epoch": 0.1468332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8644,
      "step": 4289
    },
    {
      "epoch": 0.1468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8297,
      "step": 4290
    },
    {
      "epoch": 0.146901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8751,
      "step": 4291
    },
    {
      "epoch": 0.1469359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8744,
      "step": 4292
    },
    {
      "epoch": 0.14697021567956178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8535,
      "step": 4293
    },
    {
      "epoch": 0.14700445053064018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8504,
      "step": 4294
    },
    {
      "epoch": 0.14703868538171858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8384,
      "step": 4295
    },
    {
      "epoch": 0.147072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8475,
      "step": 4296
    },
    {
      "epoch": 0.1471071550838754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8311,
      "step": 4297
    },
    {
      "epoch": 0.1471413899349538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8665,
      "step": 4298
    },
    {
      "epoch": 0.1471756247860322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8553,
      "step": 4299
    },
    {
      "epoch": 0.14720985963711058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8669,
      "step": 4300
    },
    {
      "epoch": 0.14724409448818898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.869,
      "step": 4301
    },
    {
      "epoch": 0.14727832933926738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.837,
      "step": 4302
    },
    {
      "epoch": 0.14731256419034577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8391,
      "step": 4303
    },
    {
      "epoch": 0.14734679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8552,
      "step": 4304
    },
    {
      "epoch": 0.14738103389250257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8585,
      "step": 4305
    },
    {
      "epoch": 0.14741526874358096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8531,
      "step": 4306
    },
    {
      "epoch": 0.14744950359465936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8713,
      "step": 4307
    },
    {
      "epoch": 0.14748373844573776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8452,
      "step": 4308
    },
    {
      "epoch": 0.14751797329681615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8397,
      "step": 4309
    },
    {
      "epoch": 0.14755220814789455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8421,
      "step": 4310
    },
    {
      "epoch": 0.14758644299897294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8425,
      "step": 4311
    },
    {
      "epoch": 0.14762067785005134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8527,
      "step": 4312
    },
    {
      "epoch": 0.14765491270112974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8536,
      "step": 4313
    },
    {
      "epoch": 0.14768914755220816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8569,
      "step": 4314
    },
    {
      "epoch": 0.14772338240328656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8563,
      "step": 4315
    },
    {
      "epoch": 0.14775761725436495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8559,
      "step": 4316
    },
    {
      "epoch": 0.14779185210544335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8555,
      "step": 4317
    },
    {
      "epoch": 0.14782608695652175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8461,
      "step": 4318
    },
    {
      "epoch": 0.14786032180760014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8536,
      "step": 4319
    },
    {
      "epoch": 0.14789455665867854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8604,
      "step": 4320
    },
    {
      "epoch": 0.14792879150975694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8114,
      "step": 4321
    },
    {
      "epoch": 0.14796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8667,
      "step": 4322
    },
    {
      "epoch": 0.14799726121191373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8567,
      "step": 4323
    },
    {
      "epoch": 0.14803149606299212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8434,
      "step": 4324
    },
    {
      "epoch": 0.14806573091407052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8349,
      "step": 4325
    },
    {
      "epoch": 0.14809996576514892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.858,
      "step": 4326
    },
    {
      "epoch": 0.1481342006162273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8326,
      "step": 4327
    },
    {
      "epoch": 0.1481684354673057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8367,
      "step": 4328
    },
    {
      "epoch": 0.1482026703183841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8486,
      "step": 4329
    },
    {
      "epoch": 0.1482369051694625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8395,
      "step": 4330
    },
    {
      "epoch": 0.1482711400205409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8516,
      "step": 4331
    },
    {
      "epoch": 0.14830537487161932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8449,
      "step": 4332
    },
    {
      "epoch": 0.14833960972269772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8427,
      "step": 4333
    },
    {
      "epoch": 0.14837384457377611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.848,
      "step": 4334
    },
    {
      "epoch": 0.1484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8466,
      "step": 4335
    },
    {
      "epoch": 0.1484423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8477,
      "step": 4336
    },
    {
      "epoch": 0.1484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8433,
      "step": 4337
    },
    {
      "epoch": 0.1485107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8369,
      "step": 4338
    },
    {
      "epoch": 0.1485450188291681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8499,
      "step": 4339
    },
    {
      "epoch": 0.1485792536802465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8207,
      "step": 4340
    },
    {
      "epoch": 0.1486134885313249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8297,
      "step": 4341
    },
    {
      "epoch": 0.14864772338240329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8613,
      "step": 4342
    },
    {
      "epoch": 0.14868195823348168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8298,
      "step": 4343
    },
    {
      "epoch": 0.14871619308456008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8414,
      "step": 4344
    },
    {
      "epoch": 0.14875042793563847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8319,
      "step": 4345
    },
    {
      "epoch": 0.14878466278671687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8554,
      "step": 4346
    },
    {
      "epoch": 0.14881889763779527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8403,
      "step": 4347
    },
    {
      "epoch": 0.14885313248887366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8645,
      "step": 4348
    },
    {
      "epoch": 0.14888736733995206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8469,
      "step": 4349
    },
    {
      "epoch": 0.14892160219103046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.856,
      "step": 4350
    },
    {
      "epoch": 0.14895583704210888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8477,
      "step": 4351
    },
    {
      "epoch": 0.14899007189318728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8459,
      "step": 4352
    },
    {
      "epoch": 0.14902430674426567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8345,
      "step": 4353
    },
    {
      "epoch": 0.14905854159534407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8333,
      "step": 4354
    },
    {
      "epoch": 0.14909277644642246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8251,
      "step": 4355
    },
    {
      "epoch": 0.14912701129750086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8282,
      "step": 4356
    },
    {
      "epoch": 0.14916124614857926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8318,
      "step": 4357
    },
    {
      "epoch": 0.14919548099965765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8464,
      "step": 4358
    },
    {
      "epoch": 0.14922971585073605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8345,
      "step": 4359
    },
    {
      "epoch": 0.14926395070181445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8386,
      "step": 4360
    },
    {
      "epoch": 0.14929818555289284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.807,
      "step": 4361
    },
    {
      "epoch": 0.14933242040397124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8168,
      "step": 4362
    },
    {
      "epoch": 0.14936665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8441,
      "step": 4363
    },
    {
      "epoch": 0.14940089010612803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8149,
      "step": 4364
    },
    {
      "epoch": 0.14943512495720643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8302,
      "step": 4365
    },
    {
      "epoch": 0.14946935980828482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8408,
      "step": 4366
    },
    {
      "epoch": 0.14950359465936322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8251,
      "step": 4367
    },
    {
      "epoch": 0.14953782951044162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8441,
      "step": 4368
    },
    {
      "epoch": 0.14957206436152004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8411,
      "step": 4369
    },
    {
      "epoch": 0.14960629921259844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8533,
      "step": 4370
    },
    {
      "epoch": 0.14964053406367683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8434,
      "step": 4371
    },
    {
      "epoch": 0.14967476891475523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8259,
      "step": 4372
    },
    {
      "epoch": 0.14970900376583363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8213,
      "step": 4373
    },
    {
      "epoch": 0.14974323861691202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8342,
      "step": 4374
    },
    {
      "epoch": 0.14977747346799042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8186,
      "step": 4375
    },
    {
      "epoch": 0.14981170831906881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.825,
      "step": 4376
    },
    {
      "epoch": 0.1498459431701472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8373,
      "step": 4377
    },
    {
      "epoch": 0.1498801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8092,
      "step": 4378
    },
    {
      "epoch": 0.149914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8523,
      "step": 4379
    },
    {
      "epoch": 0.1499486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8435,
      "step": 4380
    },
    {
      "epoch": 0.1499828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8525,
      "step": 4381
    },
    {
      "epoch": 0.1500171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.816,
      "step": 4382
    },
    {
      "epoch": 0.1500513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8429,
      "step": 4383
    },
    {
      "epoch": 0.15008558712769599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8508,
      "step": 4384
    },
    {
      "epoch": 0.15011982197877438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8238,
      "step": 4385
    },
    {
      "epoch": 0.15015405682985278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8163,
      "step": 4386
    },
    {
      "epoch": 0.15018829168093117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8423,
      "step": 4387
    },
    {
      "epoch": 0.1502225265320096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8373,
      "step": 4388
    },
    {
      "epoch": 0.150256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8198,
      "step": 4389
    },
    {
      "epoch": 0.1502909962341664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8414,
      "step": 4390
    },
    {
      "epoch": 0.1503252310852448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.84,
      "step": 4391
    },
    {
      "epoch": 0.15035946593632318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8372,
      "step": 4392
    },
    {
      "epoch": 0.15039370078740158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8153,
      "step": 4393
    },
    {
      "epoch": 0.15042793563847998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8206,
      "step": 4394
    },
    {
      "epoch": 0.15046217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8204,
      "step": 4395
    },
    {
      "epoch": 0.15049640534063677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.832,
      "step": 4396
    },
    {
      "epoch": 0.15053064019171516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8107,
      "step": 4397
    },
    {
      "epoch": 0.15056487504279356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8486,
      "step": 4398
    },
    {
      "epoch": 0.15059910989387196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8275,
      "step": 4399
    },
    {
      "epoch": 0.15063334474495035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8423,
      "step": 4400
    },
    {
      "epoch": 0.15066757959602875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8234,
      "step": 4401
    },
    {
      "epoch": 0.15070181444710715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8068,
      "step": 4402
    },
    {
      "epoch": 0.15073604929818554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8486,
      "step": 4403
    },
    {
      "epoch": 0.15077028414926394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.861,
      "step": 4404
    },
    {
      "epoch": 0.15080451900034234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8193,
      "step": 4405
    },
    {
      "epoch": 0.15083875385142076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8489,
      "step": 4406
    },
    {
      "epoch": 0.15087298870249916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8449,
      "step": 4407
    },
    {
      "epoch": 0.15090722355357755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8452,
      "step": 4408
    },
    {
      "epoch": 0.15094145840465595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8416,
      "step": 4409
    },
    {
      "epoch": 0.15097569325573434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8256,
      "step": 4410
    },
    {
      "epoch": 0.15100992810681274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8187,
      "step": 4411
    },
    {
      "epoch": 0.15104416295789114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8428,
      "step": 4412
    },
    {
      "epoch": 0.15107839780896953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8467,
      "step": 4413
    },
    {
      "epoch": 0.15111263266004793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8563,
      "step": 4414
    },
    {
      "epoch": 0.15114686751112633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.831,
      "step": 4415
    },
    {
      "epoch": 0.15118110236220472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8011,
      "step": 4416
    },
    {
      "epoch": 0.15121533721328312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8401,
      "step": 4417
    },
    {
      "epoch": 0.15124957206436151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8217,
      "step": 4418
    },
    {
      "epoch": 0.1512838069154399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8248,
      "step": 4419
    },
    {
      "epoch": 0.1513180417665183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8339,
      "step": 4420
    },
    {
      "epoch": 0.1513522766175967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8508,
      "step": 4421
    },
    {
      "epoch": 0.1513865114686751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8184,
      "step": 4422
    },
    {
      "epoch": 0.1514207463197535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8403,
      "step": 4423
    },
    {
      "epoch": 0.15145498117083192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8259,
      "step": 4424
    },
    {
      "epoch": 0.15148921602191032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.837,
      "step": 4425
    },
    {
      "epoch": 0.1515234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8315,
      "step": 4426
    },
    {
      "epoch": 0.1515576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8431,
      "step": 4427
    },
    {
      "epoch": 0.1515919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8358,
      "step": 4428
    },
    {
      "epoch": 0.1516261554262239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8214,
      "step": 4429
    },
    {
      "epoch": 0.1516603902773023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8255,
      "step": 4430
    },
    {
      "epoch": 0.1516946251283807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.834,
      "step": 4431
    },
    {
      "epoch": 0.1517288599794591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8344,
      "step": 4432
    },
    {
      "epoch": 0.1517630948305375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8104,
      "step": 4433
    },
    {
      "epoch": 0.15179732968161588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7941,
      "step": 4434
    },
    {
      "epoch": 0.15183156453269428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.819,
      "step": 4435
    },
    {
      "epoch": 0.15186579938377268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8222,
      "step": 4436
    },
    {
      "epoch": 0.15190003423485107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8425,
      "step": 4437
    },
    {
      "epoch": 0.15193426908592947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8129,
      "step": 4438
    },
    {
      "epoch": 0.15196850393700786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8086,
      "step": 4439
    },
    {
      "epoch": 0.15200273878808626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8225,
      "step": 4440
    },
    {
      "epoch": 0.15203697363916466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8299,
      "step": 4441
    },
    {
      "epoch": 0.15207120849024305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8381,
      "step": 4442
    },
    {
      "epoch": 0.15210544334132148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8253,
      "step": 4443
    },
    {
      "epoch": 0.15213967819239987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8562,
      "step": 4444
    },
    {
      "epoch": 0.15217391304347827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7999,
      "step": 4445
    },
    {
      "epoch": 0.15220814789455667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8339,
      "step": 4446
    },
    {
      "epoch": 0.15224238274563506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8203,
      "step": 4447
    },
    {
      "epoch": 0.15227661759671346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8307,
      "step": 4448
    },
    {
      "epoch": 0.15231085244779186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8191,
      "step": 4449
    },
    {
      "epoch": 0.15234508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8497,
      "step": 4450
    },
    {
      "epoch": 0.15237932214994865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.822,
      "step": 4451
    },
    {
      "epoch": 0.15241355700102704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8229,
      "step": 4452
    },
    {
      "epoch": 0.15244779185210544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8299,
      "step": 4453
    },
    {
      "epoch": 0.15248202670318384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8206,
      "step": 4454
    },
    {
      "epoch": 0.15251626155426223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8323,
      "step": 4455
    },
    {
      "epoch": 0.15255049640534063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8344,
      "step": 4456
    },
    {
      "epoch": 0.15258473125641903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8159,
      "step": 4457
    },
    {
      "epoch": 0.15261896610749742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8243,
      "step": 4458
    },
    {
      "epoch": 0.15265320095857582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8199,
      "step": 4459
    },
    {
      "epoch": 0.15268743580965422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8357,
      "step": 4460
    },
    {
      "epoch": 0.15272167066073264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.819,
      "step": 4461
    },
    {
      "epoch": 0.15275590551181104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8367,
      "step": 4462
    },
    {
      "epoch": 0.15279014036288943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8373,
      "step": 4463
    },
    {
      "epoch": 0.15282437521396783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8507,
      "step": 4464
    },
    {
      "epoch": 0.15285861006504622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8458,
      "step": 4465
    },
    {
      "epoch": 0.15289284491612462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8366,
      "step": 4466
    },
    {
      "epoch": 0.15292707976720302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8145,
      "step": 4467
    },
    {
      "epoch": 0.1529613146182814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8166,
      "step": 4468
    },
    {
      "epoch": 0.1529955494693598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.837,
      "step": 4469
    },
    {
      "epoch": 0.1530297843204382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8177,
      "step": 4470
    },
    {
      "epoch": 0.1530640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8328,
      "step": 4471
    },
    {
      "epoch": 0.153098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.832,
      "step": 4472
    },
    {
      "epoch": 0.1531324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8019,
      "step": 4473
    },
    {
      "epoch": 0.1531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8345,
      "step": 4474
    },
    {
      "epoch": 0.1532009585758302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8245,
      "step": 4475
    },
    {
      "epoch": 0.15323519342690858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8217,
      "step": 4476
    },
    {
      "epoch": 0.15326942827798698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8309,
      "step": 4477
    },
    {
      "epoch": 0.15330366312906538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.828,
      "step": 4478
    },
    {
      "epoch": 0.1533378979801438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8275,
      "step": 4479
    },
    {
      "epoch": 0.1533721328312222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.829,
      "step": 4480
    },
    {
      "epoch": 0.1534063676823006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8033,
      "step": 4481
    },
    {
      "epoch": 0.153440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.814,
      "step": 4482
    },
    {
      "epoch": 0.15347483738445739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8069,
      "step": 4483
    },
    {
      "epoch": 0.15350907223553578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8092,
      "step": 4484
    },
    {
      "epoch": 0.15354330708661418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8603,
      "step": 4485
    },
    {
      "epoch": 0.15357754193769257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.814,
      "step": 4486
    },
    {
      "epoch": 0.15361177678877097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.798,
      "step": 4487
    },
    {
      "epoch": 0.15364601163984937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8165,
      "step": 4488
    },
    {
      "epoch": 0.15368024649092776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8038,
      "step": 4489
    },
    {
      "epoch": 0.15371448134200616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8434,
      "step": 4490
    },
    {
      "epoch": 0.15374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8337,
      "step": 4491
    },
    {
      "epoch": 0.15378295104416295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8414,
      "step": 4492
    },
    {
      "epoch": 0.15381718589524135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8163,
      "step": 4493
    },
    {
      "epoch": 0.15385142074631974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8411,
      "step": 4494
    },
    {
      "epoch": 0.15388565559739814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8082,
      "step": 4495
    },
    {
      "epoch": 0.15391989044847654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8108,
      "step": 4496
    },
    {
      "epoch": 0.15395412529955493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8134,
      "step": 4497
    },
    {
      "epoch": 0.15398836015063336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8518,
      "step": 4498
    },
    {
      "epoch": 0.15402259500171175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7969,
      "step": 4499
    },
    {
      "epoch": 0.15405682985279015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8107,
      "step": 4500
    },
    {
      "epoch": 0.15409106470386855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8118,
      "step": 4501
    },
    {
      "epoch": 0.15412529955494694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8056,
      "step": 4502
    },
    {
      "epoch": 0.15415953440602534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8234,
      "step": 4503
    },
    {
      "epoch": 0.15419376925710374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8124,
      "step": 4504
    },
    {
      "epoch": 0.15422800410818213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8115,
      "step": 4505
    },
    {
      "epoch": 0.15426223895926053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7989,
      "step": 4506
    },
    {
      "epoch": 0.15429647381033892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8257,
      "step": 4507
    },
    {
      "epoch": 0.15433070866141732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8254,
      "step": 4508
    },
    {
      "epoch": 0.15436494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8401,
      "step": 4509
    },
    {
      "epoch": 0.1543991783635741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8325,
      "step": 4510
    },
    {
      "epoch": 0.1544334132146525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.822,
      "step": 4511
    },
    {
      "epoch": 0.1544676480657309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8071,
      "step": 4512
    },
    {
      "epoch": 0.1545018829168093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8377,
      "step": 4513
    },
    {
      "epoch": 0.1545361177678877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8213,
      "step": 4514
    },
    {
      "epoch": 0.1545703526189661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8309,
      "step": 4515
    },
    {
      "epoch": 0.15460458747004452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7896,
      "step": 4516
    },
    {
      "epoch": 0.15463882232112292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8249,
      "step": 4517
    },
    {
      "epoch": 0.1546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8308,
      "step": 4518
    },
    {
      "epoch": 0.1547072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8162,
      "step": 4519
    },
    {
      "epoch": 0.1547415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7928,
      "step": 4520
    },
    {
      "epoch": 0.1547757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8157,
      "step": 4521
    },
    {
      "epoch": 0.1548099965765149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7931,
      "step": 4522
    },
    {
      "epoch": 0.1548442314275933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8213,
      "step": 4523
    },
    {
      "epoch": 0.1548784662786717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8326,
      "step": 4524
    },
    {
      "epoch": 0.15491270112975009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.802,
      "step": 4525
    },
    {
      "epoch": 0.15494693598082848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8261,
      "step": 4526
    },
    {
      "epoch": 0.15498117083190688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7903,
      "step": 4527
    },
    {
      "epoch": 0.15501540568298527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8133,
      "step": 4528
    },
    {
      "epoch": 0.15504964053406367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8003,
      "step": 4529
    },
    {
      "epoch": 0.15508387538514207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8245,
      "step": 4530
    },
    {
      "epoch": 0.15511811023622046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8225,
      "step": 4531
    },
    {
      "epoch": 0.15515234508729886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8282,
      "step": 4532
    },
    {
      "epoch": 0.15518657993837726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8164,
      "step": 4533
    },
    {
      "epoch": 0.15522081478945565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8388,
      "step": 4534
    },
    {
      "epoch": 0.15525504964053408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.841,
      "step": 4535
    },
    {
      "epoch": 0.15528928449161247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8138,
      "step": 4536
    },
    {
      "epoch": 0.15532351934269087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8443,
      "step": 4537
    },
    {
      "epoch": 0.15535775419376927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8019,
      "step": 4538
    },
    {
      "epoch": 0.15539198904484766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8249,
      "step": 4539
    },
    {
      "epoch": 0.15542622389592606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8024,
      "step": 4540
    },
    {
      "epoch": 0.15546045874700445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8391,
      "step": 4541
    },
    {
      "epoch": 0.15549469359808285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7978,
      "step": 4542
    },
    {
      "epoch": 0.15552892844916125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8085,
      "step": 4543
    },
    {
      "epoch": 0.15556316330023964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7983,
      "step": 4544
    },
    {
      "epoch": 0.15559739815131804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8196,
      "step": 4545
    },
    {
      "epoch": 0.15563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8104,
      "step": 4546
    },
    {
      "epoch": 0.15566586785347483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.804,
      "step": 4547
    },
    {
      "epoch": 0.15570010270455323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8184,
      "step": 4548
    },
    {
      "epoch": 0.15573433755563162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8086,
      "step": 4549
    },
    {
      "epoch": 0.15576857240671002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8132,
      "step": 4550
    },
    {
      "epoch": 0.15580280725778842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8301,
      "step": 4551
    },
    {
      "epoch": 0.1558370421088668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8086,
      "step": 4552
    },
    {
      "epoch": 0.15587127695994524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8204,
      "step": 4553
    },
    {
      "epoch": 0.15590551181102363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7974,
      "step": 4554
    },
    {
      "epoch": 0.15593974666210203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.817,
      "step": 4555
    },
    {
      "epoch": 0.15597398151318043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8234,
      "step": 4556
    },
    {
      "epoch": 0.15600821636425882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8263,
      "step": 4557
    },
    {
      "epoch": 0.15604245121533722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8069,
      "step": 4558
    },
    {
      "epoch": 0.15607668606641562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8039,
      "step": 4559
    },
    {
      "epoch": 0.156110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7889,
      "step": 4560
    },
    {
      "epoch": 0.1561451557685724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8022,
      "step": 4561
    },
    {
      "epoch": 0.1561793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8044,
      "step": 4562
    },
    {
      "epoch": 0.1562136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8112,
      "step": 4563
    },
    {
      "epoch": 0.1562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8295,
      "step": 4564
    },
    {
      "epoch": 0.156282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8102,
      "step": 4565
    },
    {
      "epoch": 0.1563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8361,
      "step": 4566
    },
    {
      "epoch": 0.15635056487504279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.812,
      "step": 4567
    },
    {
      "epoch": 0.15638479972612118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.817,
      "step": 4568
    },
    {
      "epoch": 0.15641903457719958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.806,
      "step": 4569
    },
    {
      "epoch": 0.15645326942827797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7917,
      "step": 4570
    },
    {
      "epoch": 0.1564875042793564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.827,
      "step": 4571
    },
    {
      "epoch": 0.1565217391304348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7994,
      "step": 4572
    },
    {
      "epoch": 0.1565559739815132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7878,
      "step": 4573
    },
    {
      "epoch": 0.1565902088325916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8131,
      "step": 4574
    },
    {
      "epoch": 0.15662444368366998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7949,
      "step": 4575
    },
    {
      "epoch": 0.15665867853474838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.839,
      "step": 4576
    },
    {
      "epoch": 0.15669291338582678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8014,
      "step": 4577
    },
    {
      "epoch": 0.15672714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8229,
      "step": 4578
    },
    {
      "epoch": 0.15676138308798357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8234,
      "step": 4579
    },
    {
      "epoch": 0.15679561793906197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8256,
      "step": 4580
    },
    {
      "epoch": 0.15682985279014036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8267,
      "step": 4581
    },
    {
      "epoch": 0.15686408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8384,
      "step": 4582
    },
    {
      "epoch": 0.15689832249229715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8141,
      "step": 4583
    },
    {
      "epoch": 0.15693255734337555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8027,
      "step": 4584
    },
    {
      "epoch": 0.15696679219445395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8016,
      "step": 4585
    },
    {
      "epoch": 0.15700102704553234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8011,
      "step": 4586
    },
    {
      "epoch": 0.15703526189661074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.819,
      "step": 4587
    },
    {
      "epoch": 0.15706949674768914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.809,
      "step": 4588
    },
    {
      "epoch": 0.15710373159876753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.819,
      "step": 4589
    },
    {
      "epoch": 0.15713796644984596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.791,
      "step": 4590
    },
    {
      "epoch": 0.15717220130092435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7805,
      "step": 4591
    },
    {
      "epoch": 0.15720643615200275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7945,
      "step": 4592
    },
    {
      "epoch": 0.15724067100308114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8084,
      "step": 4593
    },
    {
      "epoch": 0.15727490585415954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7917,
      "step": 4594
    },
    {
      "epoch": 0.15730914070523794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8224,
      "step": 4595
    },
    {
      "epoch": 0.15734337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8102,
      "step": 4596
    },
    {
      "epoch": 0.15737761040739473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7961,
      "step": 4597
    },
    {
      "epoch": 0.15741184525847313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8035,
      "step": 4598
    },
    {
      "epoch": 0.15744608010955152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8182,
      "step": 4599
    },
    {
      "epoch": 0.15748031496062992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7811,
      "step": 4600
    },
    {
      "epoch": 0.15751454981170832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8024,
      "step": 4601
    },
    {
      "epoch": 0.1575487846627867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7869,
      "step": 4602
    },
    {
      "epoch": 0.1575830195138651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7977,
      "step": 4603
    },
    {
      "epoch": 0.1576172543649435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8279,
      "step": 4604
    },
    {
      "epoch": 0.1576514892160219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7881,
      "step": 4605
    },
    {
      "epoch": 0.1576857240671003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7746,
      "step": 4606
    },
    {
      "epoch": 0.1577199589181787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8,
      "step": 4607
    },
    {
      "epoch": 0.15775419376925712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8096,
      "step": 4608
    },
    {
      "epoch": 0.1577884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8067,
      "step": 4609
    },
    {
      "epoch": 0.1578226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8159,
      "step": 4610
    },
    {
      "epoch": 0.1578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8121,
      "step": 4611
    },
    {
      "epoch": 0.1578911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.801,
      "step": 4612
    },
    {
      "epoch": 0.1579253680246491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7919,
      "step": 4613
    },
    {
      "epoch": 0.1579596028757275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7997,
      "step": 4614
    },
    {
      "epoch": 0.1579938377268059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7983,
      "step": 4615
    },
    {
      "epoch": 0.1580280725778843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8,
      "step": 4616
    },
    {
      "epoch": 0.15806230742896268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8041,
      "step": 4617
    },
    {
      "epoch": 0.15809654228004108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7952,
      "step": 4618
    },
    {
      "epoch": 0.15813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8057,
      "step": 4619
    },
    {
      "epoch": 0.15816501198219787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8299,
      "step": 4620
    },
    {
      "epoch": 0.15819924683327627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7787,
      "step": 4621
    },
    {
      "epoch": 0.15823348168435467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7934,
      "step": 4622
    },
    {
      "epoch": 0.15826771653543306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.776,
      "step": 4623
    },
    {
      "epoch": 0.15830195138651146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7988,
      "step": 4624
    },
    {
      "epoch": 0.15833618623758985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8087,
      "step": 4625
    },
    {
      "epoch": 0.15837042108866825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7981,
      "step": 4626
    },
    {
      "epoch": 0.15840465593974667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8144,
      "step": 4627
    },
    {
      "epoch": 0.15843889079082507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8219,
      "step": 4628
    },
    {
      "epoch": 0.15847312564190347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7941,
      "step": 4629
    },
    {
      "epoch": 0.15850736049298186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7872,
      "step": 4630
    },
    {
      "epoch": 0.15854159534406026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7957,
      "step": 4631
    },
    {
      "epoch": 0.15857583019513866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8073,
      "step": 4632
    },
    {
      "epoch": 0.15861006504621705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8163,
      "step": 4633
    },
    {
      "epoch": 0.15864429989729545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7835,
      "step": 4634
    },
    {
      "epoch": 0.15867853474837385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8066,
      "step": 4635
    },
    {
      "epoch": 0.15871276959945224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8163,
      "step": 4636
    },
    {
      "epoch": 0.15874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8106,
      "step": 4637
    },
    {
      "epoch": 0.15878123930160903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8112,
      "step": 4638
    },
    {
      "epoch": 0.15881547415268743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8148,
      "step": 4639
    },
    {
      "epoch": 0.15884970900376583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8127,
      "step": 4640
    },
    {
      "epoch": 0.15888394385484422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7988,
      "step": 4641
    },
    {
      "epoch": 0.15891817870592262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8074,
      "step": 4642
    },
    {
      "epoch": 0.15895241355700102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8191,
      "step": 4643
    },
    {
      "epoch": 0.1589866484080794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7967,
      "step": 4644
    },
    {
      "epoch": 0.15902088325915784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8156,
      "step": 4645
    },
    {
      "epoch": 0.15905511811023623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8077,
      "step": 4646
    },
    {
      "epoch": 0.15908935296131463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8059,
      "step": 4647
    },
    {
      "epoch": 0.15912358781239302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7982,
      "step": 4648
    },
    {
      "epoch": 0.15915782266347142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.796,
      "step": 4649
    },
    {
      "epoch": 0.15919205751454982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7683,
      "step": 4650
    },
    {
      "epoch": 0.1592262923656282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8051,
      "step": 4651
    },
    {
      "epoch": 0.1592605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7783,
      "step": 4652
    },
    {
      "epoch": 0.159294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.785,
      "step": 4653
    },
    {
      "epoch": 0.1593289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8197,
      "step": 4654
    },
    {
      "epoch": 0.1593632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8135,
      "step": 4655
    },
    {
      "epoch": 0.1593974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8097,
      "step": 4656
    },
    {
      "epoch": 0.1594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8215,
      "step": 4657
    },
    {
      "epoch": 0.159465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.81,
      "step": 4658
    },
    {
      "epoch": 0.15950017117425538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7951,
      "step": 4659
    },
    {
      "epoch": 0.15953440602533378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8176,
      "step": 4660
    },
    {
      "epoch": 0.15956864087641218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.777,
      "step": 4661
    },
    {
      "epoch": 0.15960287572749057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7913,
      "step": 4662
    },
    {
      "epoch": 0.159637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7891,
      "step": 4663
    },
    {
      "epoch": 0.1596713454296474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8264,
      "step": 4664
    },
    {
      "epoch": 0.1597055802807258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7838,
      "step": 4665
    },
    {
      "epoch": 0.15973981513180419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7917,
      "step": 4666
    },
    {
      "epoch": 0.15977404998288258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8194,
      "step": 4667
    },
    {
      "epoch": 0.15980828483396098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.794,
      "step": 4668
    },
    {
      "epoch": 0.15984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8289,
      "step": 4669
    },
    {
      "epoch": 0.15987675453611777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8023,
      "step": 4670
    },
    {
      "epoch": 0.15991098938719617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8027,
      "step": 4671
    },
    {
      "epoch": 0.15994522423827456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7827,
      "step": 4672
    },
    {
      "epoch": 0.15997945908935296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8044,
      "step": 4673
    },
    {
      "epoch": 0.16001369394043136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7942,
      "step": 4674
    },
    {
      "epoch": 0.16004792879150975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.789,
      "step": 4675
    },
    {
      "epoch": 0.16008216364258815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8055,
      "step": 4676
    },
    {
      "epoch": 0.16011639849366655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8073,
      "step": 4677
    },
    {
      "epoch": 0.16015063334474494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.811,
      "step": 4678
    },
    {
      "epoch": 0.16018486819582334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7993,
      "step": 4679
    },
    {
      "epoch": 0.16021910304690173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7723,
      "step": 4680
    },
    {
      "epoch": 0.16025333789798013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8029,
      "step": 4681
    },
    {
      "epoch": 0.16028757274905855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.795,
      "step": 4682
    },
    {
      "epoch": 0.16032180760013695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7877,
      "step": 4683
    },
    {
      "epoch": 0.16035604245121535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7929,
      "step": 4684
    },
    {
      "epoch": 0.16039027730229374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7908,
      "step": 4685
    },
    {
      "epoch": 0.16042451215337214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8027,
      "step": 4686
    },
    {
      "epoch": 0.16045874700445054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.792,
      "step": 4687
    },
    {
      "epoch": 0.16049298185552893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7613,
      "step": 4688
    },
    {
      "epoch": 0.16052721670660733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.821,
      "step": 4689
    },
    {
      "epoch": 0.16056145155768572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7905,
      "step": 4690
    },
    {
      "epoch": 0.16059568640876412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7658,
      "step": 4691
    },
    {
      "epoch": 0.16062992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7971,
      "step": 4692
    },
    {
      "epoch": 0.1606641561109209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7968,
      "step": 4693
    },
    {
      "epoch": 0.1606983909619993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8011,
      "step": 4694
    },
    {
      "epoch": 0.1607326258130777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8082,
      "step": 4695
    },
    {
      "epoch": 0.1607668606641561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7999,
      "step": 4696
    },
    {
      "epoch": 0.1608010955152345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.774,
      "step": 4697
    },
    {
      "epoch": 0.1608353303663129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7793,
      "step": 4698
    },
    {
      "epoch": 0.1608695652173913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8074,
      "step": 4699
    },
    {
      "epoch": 0.16090380006846972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.803,
      "step": 4700
    },
    {
      "epoch": 0.1609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8061,
      "step": 4701
    },
    {
      "epoch": 0.1609722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.787,
      "step": 4702
    },
    {
      "epoch": 0.1610065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7788,
      "step": 4703
    },
    {
      "epoch": 0.1610407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7881,
      "step": 4704
    },
    {
      "epoch": 0.1610749743238617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.797,
      "step": 4705
    },
    {
      "epoch": 0.1611092091749401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7947,
      "step": 4706
    },
    {
      "epoch": 0.1611434440260185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8123,
      "step": 4707
    },
    {
      "epoch": 0.16117767887709689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8126,
      "step": 4708
    },
    {
      "epoch": 0.16121191372817528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8139,
      "step": 4709
    },
    {
      "epoch": 0.16124614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7933,
      "step": 4710
    },
    {
      "epoch": 0.16128038343033207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7924,
      "step": 4711
    },
    {
      "epoch": 0.16131461828141047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.793,
      "step": 4712
    },
    {
      "epoch": 0.16134885313248887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8062,
      "step": 4713
    },
    {
      "epoch": 0.16138308798356726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.766,
      "step": 4714
    },
    {
      "epoch": 0.16141732283464566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7933,
      "step": 4715
    },
    {
      "epoch": 0.16145155768572406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7984,
      "step": 4716
    },
    {
      "epoch": 0.16148579253680245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8035,
      "step": 4717
    },
    {
      "epoch": 0.16152002738788085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7993,
      "step": 4718
    },
    {
      "epoch": 0.16155426223895927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8069,
      "step": 4719
    },
    {
      "epoch": 0.16158849709003767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.796,
      "step": 4720
    },
    {
      "epoch": 0.16162273194111607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8105,
      "step": 4721
    },
    {
      "epoch": 0.16165696679219446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7889,
      "step": 4722
    },
    {
      "epoch": 0.16169120164327286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8224,
      "step": 4723
    },
    {
      "epoch": 0.16172543649435125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7948,
      "step": 4724
    },
    {
      "epoch": 0.16175967134542965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7798,
      "step": 4725
    },
    {
      "epoch": 0.16179390619650805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7767,
      "step": 4726
    },
    {
      "epoch": 0.16182814104758644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7781,
      "step": 4727
    },
    {
      "epoch": 0.16186237589866484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7711,
      "step": 4728
    },
    {
      "epoch": 0.16189661074974324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7829,
      "step": 4729
    },
    {
      "epoch": 0.16193084560082163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7607,
      "step": 4730
    },
    {
      "epoch": 0.16196508045190003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8005,
      "step": 4731
    },
    {
      "epoch": 0.16199931530297842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.778,
      "step": 4732
    },
    {
      "epoch": 0.16203355015405682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7659,
      "step": 4733
    },
    {
      "epoch": 0.16206778500513522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7905,
      "step": 4734
    },
    {
      "epoch": 0.1621020198562136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8005,
      "step": 4735
    },
    {
      "epoch": 0.162136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7798,
      "step": 4736
    },
    {
      "epoch": 0.16217048955837043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.798,
      "step": 4737
    },
    {
      "epoch": 0.16220472440944883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7852,
      "step": 4738
    },
    {
      "epoch": 0.16223895926052723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7878,
      "step": 4739
    },
    {
      "epoch": 0.16227319411160562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7832,
      "step": 4740
    },
    {
      "epoch": 0.16230742896268402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7867,
      "step": 4741
    },
    {
      "epoch": 0.16234166381376242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.751,
      "step": 4742
    },
    {
      "epoch": 0.1623758986648408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8022,
      "step": 4743
    },
    {
      "epoch": 0.1624101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.797,
      "step": 4744
    },
    {
      "epoch": 0.1624443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7805,
      "step": 4745
    },
    {
      "epoch": 0.162478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7746,
      "step": 4746
    },
    {
      "epoch": 0.1625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.773,
      "step": 4747
    },
    {
      "epoch": 0.1625470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7895,
      "step": 4748
    },
    {
      "epoch": 0.1625813077713112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7895,
      "step": 4749
    },
    {
      "epoch": 0.1626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7853,
      "step": 4750
    },
    {
      "epoch": 0.16264977747346798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7789,
      "step": 4751
    },
    {
      "epoch": 0.16268401232454638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7909,
      "step": 4752
    },
    {
      "epoch": 0.16271824717562478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7907,
      "step": 4753
    },
    {
      "epoch": 0.16275248202670317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7663,
      "step": 4754
    },
    {
      "epoch": 0.1627867168777816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8041,
      "step": 4755
    },
    {
      "epoch": 0.16282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7953,
      "step": 4756
    },
    {
      "epoch": 0.1628551865799384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7914,
      "step": 4757
    },
    {
      "epoch": 0.16288942143101678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8161,
      "step": 4758
    },
    {
      "epoch": 0.16292365628209518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8064,
      "step": 4759
    },
    {
      "epoch": 0.16295789113317358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7966,
      "step": 4760
    },
    {
      "epoch": 0.16299212598425197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7917,
      "step": 4761
    },
    {
      "epoch": 0.16302636083533037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7964,
      "step": 4762
    },
    {
      "epoch": 0.16306059568640877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7847,
      "step": 4763
    },
    {
      "epoch": 0.16309483053748716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8116,
      "step": 4764
    },
    {
      "epoch": 0.16312906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7906,
      "step": 4765
    },
    {
      "epoch": 0.16316330023964395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7889,
      "step": 4766
    },
    {
      "epoch": 0.16319753509072235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8069,
      "step": 4767
    },
    {
      "epoch": 0.16323176994180075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8022,
      "step": 4768
    },
    {
      "epoch": 0.16326600479287914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7587,
      "step": 4769
    },
    {
      "epoch": 0.16330023964395754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7809,
      "step": 4770
    },
    {
      "epoch": 0.16333447449503594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7803,
      "step": 4771
    },
    {
      "epoch": 0.16336870934611433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7984,
      "step": 4772
    },
    {
      "epoch": 0.16340294419719273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7901,
      "step": 4773
    },
    {
      "epoch": 0.16343717904827115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7669,
      "step": 4774
    },
    {
      "epoch": 0.16347141389934955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7687,
      "step": 4775
    },
    {
      "epoch": 0.16350564875042795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7968,
      "step": 4776
    },
    {
      "epoch": 0.16353988360150634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.788,
      "step": 4777
    },
    {
      "epoch": 0.16357411845258474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7881,
      "step": 4778
    },
    {
      "epoch": 0.16360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7869,
      "step": 4779
    },
    {
      "epoch": 0.16364258815474153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7739,
      "step": 4780
    },
    {
      "epoch": 0.16367682300581993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7987,
      "step": 4781
    },
    {
      "epoch": 0.16371105785689832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7895,
      "step": 4782
    },
    {
      "epoch": 0.16374529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7576,
      "step": 4783
    },
    {
      "epoch": 0.16377952755905512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7922,
      "step": 4784
    },
    {
      "epoch": 0.1638137624101335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7933,
      "step": 4785
    },
    {
      "epoch": 0.1638479972612119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7604,
      "step": 4786
    },
    {
      "epoch": 0.1638822321122903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7789,
      "step": 4787
    },
    {
      "epoch": 0.1639164669633687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7965,
      "step": 4788
    },
    {
      "epoch": 0.1639507018144471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7589,
      "step": 4789
    },
    {
      "epoch": 0.1639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7814,
      "step": 4790
    },
    {
      "epoch": 0.1640191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7714,
      "step": 4791
    },
    {
      "epoch": 0.16405340636768231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7807,
      "step": 4792
    },
    {
      "epoch": 0.1640876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.802,
      "step": 4793
    },
    {
      "epoch": 0.1641218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7706,
      "step": 4794
    },
    {
      "epoch": 0.1641561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7784,
      "step": 4795
    },
    {
      "epoch": 0.1641903457719959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7967,
      "step": 4796
    },
    {
      "epoch": 0.1642245806230743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7433,
      "step": 4797
    },
    {
      "epoch": 0.1642588154741527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7637,
      "step": 4798
    },
    {
      "epoch": 0.1642930503252311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7715,
      "step": 4799
    },
    {
      "epoch": 0.16432728517630948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.793,
      "step": 4800
    },
    {
      "epoch": 0.16436152002738788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7607,
      "step": 4801
    },
    {
      "epoch": 0.16439575487846628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7779,
      "step": 4802
    },
    {
      "epoch": 0.16442998972954467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7815,
      "step": 4803
    },
    {
      "epoch": 0.16446422458062307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7797,
      "step": 4804
    },
    {
      "epoch": 0.16449845943170147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8053,
      "step": 4805
    },
    {
      "epoch": 0.16453269428277986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7828,
      "step": 4806
    },
    {
      "epoch": 0.16456692913385826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7902,
      "step": 4807
    },
    {
      "epoch": 0.16460116398493665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7997,
      "step": 4808
    },
    {
      "epoch": 0.16463539883601505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7976,
      "step": 4809
    },
    {
      "epoch": 0.16466963368709345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7694,
      "step": 4810
    },
    {
      "epoch": 0.16470386853817187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7783,
      "step": 4811
    },
    {
      "epoch": 0.16473810338925027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7766,
      "step": 4812
    },
    {
      "epoch": 0.16477233824032866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7915,
      "step": 4813
    },
    {
      "epoch": 0.16480657309140706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7914,
      "step": 4814
    },
    {
      "epoch": 0.16484080794248546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7952,
      "step": 4815
    },
    {
      "epoch": 0.16487504279356385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7822,
      "step": 4816
    },
    {
      "epoch": 0.16490927764464225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7754,
      "step": 4817
    },
    {
      "epoch": 0.16494351249572065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7657,
      "step": 4818
    },
    {
      "epoch": 0.16497774734679904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8058,
      "step": 4819
    },
    {
      "epoch": 0.16501198219787744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7659,
      "step": 4820
    },
    {
      "epoch": 0.16504621704895583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7555,
      "step": 4821
    },
    {
      "epoch": 0.16508045190003423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.769,
      "step": 4822
    },
    {
      "epoch": 0.16511468675111263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.775,
      "step": 4823
    },
    {
      "epoch": 0.16514892160219102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7735,
      "step": 4824
    },
    {
      "epoch": 0.16518315645326942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7914,
      "step": 4825
    },
    {
      "epoch": 0.16521739130434782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7806,
      "step": 4826
    },
    {
      "epoch": 0.1652516261554262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7814,
      "step": 4827
    },
    {
      "epoch": 0.1652858610065046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7781,
      "step": 4828
    },
    {
      "epoch": 0.16532009585758303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7652,
      "step": 4829
    },
    {
      "epoch": 0.16535433070866143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7884,
      "step": 4830
    },
    {
      "epoch": 0.16538856555973983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7917,
      "step": 4831
    },
    {
      "epoch": 0.16542280041081822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.788,
      "step": 4832
    },
    {
      "epoch": 0.16545703526189662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7666,
      "step": 4833
    },
    {
      "epoch": 0.16549127011297501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7678,
      "step": 4834
    },
    {
      "epoch": 0.1655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7759,
      "step": 4835
    },
    {
      "epoch": 0.1655597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.808,
      "step": 4836
    },
    {
      "epoch": 0.1655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7592,
      "step": 4837
    },
    {
      "epoch": 0.1656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7696,
      "step": 4838
    },
    {
      "epoch": 0.165662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7506,
      "step": 4839
    },
    {
      "epoch": 0.1656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7794,
      "step": 4840
    },
    {
      "epoch": 0.1657309140705238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7792,
      "step": 4841
    },
    {
      "epoch": 0.16576514892160218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7707,
      "step": 4842
    },
    {
      "epoch": 0.16579938377268058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7637,
      "step": 4843
    },
    {
      "epoch": 0.16583361862375898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7943,
      "step": 4844
    },
    {
      "epoch": 0.16586785347483737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7783,
      "step": 4845
    },
    {
      "epoch": 0.16590208832591577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7671,
      "step": 4846
    },
    {
      "epoch": 0.1659363231769942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7837,
      "step": 4847
    },
    {
      "epoch": 0.1659705580280726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8004,
      "step": 4848
    },
    {
      "epoch": 0.166004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7518,
      "step": 4849
    },
    {
      "epoch": 0.16603902773022938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7755,
      "step": 4850
    },
    {
      "epoch": 0.16607326258130778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7502,
      "step": 4851
    },
    {
      "epoch": 0.16610749743238618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7865,
      "step": 4852
    },
    {
      "epoch": 0.16614173228346457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7796,
      "step": 4853
    },
    {
      "epoch": 0.16617596713454297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7709,
      "step": 4854
    },
    {
      "epoch": 0.16621020198562136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7574,
      "step": 4855
    },
    {
      "epoch": 0.16624443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7582,
      "step": 4856
    },
    {
      "epoch": 0.16627867168777816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7926,
      "step": 4857
    },
    {
      "epoch": 0.16631290653885655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7908,
      "step": 4858
    },
    {
      "epoch": 0.16634714138993495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7867,
      "step": 4859
    },
    {
      "epoch": 0.16638137624101335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7923,
      "step": 4860
    },
    {
      "epoch": 0.16641561109209174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7607,
      "step": 4861
    },
    {
      "epoch": 0.16644984594317014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7798,
      "step": 4862
    },
    {
      "epoch": 0.16648408079424853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7779,
      "step": 4863
    },
    {
      "epoch": 0.16651831564532693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7786,
      "step": 4864
    },
    {
      "epoch": 0.16655255049640533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7569,
      "step": 4865
    },
    {
      "epoch": 0.16658678534748375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7706,
      "step": 4866
    },
    {
      "epoch": 0.16662102019856215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7997,
      "step": 4867
    },
    {
      "epoch": 0.16665525504964054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7623,
      "step": 4868
    },
    {
      "epoch": 0.16668948990071894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7599,
      "step": 4869
    },
    {
      "epoch": 0.16672372475179734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7536,
      "step": 4870
    },
    {
      "epoch": 0.16675795960287573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7715,
      "step": 4871
    },
    {
      "epoch": 0.16679219445395413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7738,
      "step": 4872
    },
    {
      "epoch": 0.16682642930503253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7701,
      "step": 4873
    },
    {
      "epoch": 0.16686066415611092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7653,
      "step": 4874
    },
    {
      "epoch": 0.16689489900718932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7762,
      "step": 4875
    },
    {
      "epoch": 0.16692913385826771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.782,
      "step": 4876
    },
    {
      "epoch": 0.1669633687093461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7458,
      "step": 4877
    },
    {
      "epoch": 0.1669976035604245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7725,
      "step": 4878
    },
    {
      "epoch": 0.1670318384115029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7576,
      "step": 4879
    },
    {
      "epoch": 0.1670660732625813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7677,
      "step": 4880
    },
    {
      "epoch": 0.1671003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7692,
      "step": 4881
    },
    {
      "epoch": 0.1671345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7552,
      "step": 4882
    },
    {
      "epoch": 0.1671687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7709,
      "step": 4883
    },
    {
      "epoch": 0.1672030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7473,
      "step": 4884
    },
    {
      "epoch": 0.1672372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7595,
      "step": 4885
    },
    {
      "epoch": 0.1672714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7555,
      "step": 4886
    },
    {
      "epoch": 0.1673057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7572,
      "step": 4887
    },
    {
      "epoch": 0.1673399520712085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7496,
      "step": 4888
    },
    {
      "epoch": 0.1673741869222869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7953,
      "step": 4889
    },
    {
      "epoch": 0.1674084217733653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7508,
      "step": 4890
    },
    {
      "epoch": 0.1674426566244437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7697,
      "step": 4891
    },
    {
      "epoch": 0.16747689147552208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7615,
      "step": 4892
    },
    {
      "epoch": 0.16751112632660048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7668,
      "step": 4893
    },
    {
      "epoch": 0.16754536117767888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7659,
      "step": 4894
    },
    {
      "epoch": 0.16757959602875727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7782,
      "step": 4895
    },
    {
      "epoch": 0.16761383087983567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7693,
      "step": 4896
    },
    {
      "epoch": 0.16764806573091406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7848,
      "step": 4897
    },
    {
      "epoch": 0.16768230058199246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7781,
      "step": 4898
    },
    {
      "epoch": 0.16771653543307086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7591,
      "step": 4899
    },
    {
      "epoch": 0.16775077028414925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7582,
      "step": 4900
    },
    {
      "epoch": 0.16778500513522765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7657,
      "step": 4901
    },
    {
      "epoch": 0.16781923998630605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7583,
      "step": 4902
    },
    {
      "epoch": 0.16785347483738447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7558,
      "step": 4903
    },
    {
      "epoch": 0.16788770968846287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7684,
      "step": 4904
    },
    {
      "epoch": 0.16792194453954126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7493,
      "step": 4905
    },
    {
      "epoch": 0.16795617939061966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7808,
      "step": 4906
    },
    {
      "epoch": 0.16799041424169806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7839,
      "step": 4907
    },
    {
      "epoch": 0.16802464909277645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7706,
      "step": 4908
    },
    {
      "epoch": 0.16805888394385485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7871,
      "step": 4909
    },
    {
      "epoch": 0.16809311879493324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7754,
      "step": 4910
    },
    {
      "epoch": 0.16812735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7811,
      "step": 4911
    },
    {
      "epoch": 0.16816158849709004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7563,
      "step": 4912
    },
    {
      "epoch": 0.16819582334816843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7592,
      "step": 4913
    },
    {
      "epoch": 0.16823005819924683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7636,
      "step": 4914
    },
    {
      "epoch": 0.16826429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7693,
      "step": 4915
    },
    {
      "epoch": 0.16829852790140362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7672,
      "step": 4916
    },
    {
      "epoch": 0.16833276275248202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7811,
      "step": 4917
    },
    {
      "epoch": 0.16836699760356041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7649,
      "step": 4918
    },
    {
      "epoch": 0.1684012324546388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7887,
      "step": 4919
    },
    {
      "epoch": 0.1684354673057172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7514,
      "step": 4920
    },
    {
      "epoch": 0.16846970215679563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7777,
      "step": 4921
    },
    {
      "epoch": 0.16850393700787403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7633,
      "step": 4922
    },
    {
      "epoch": 0.16853817185895242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7544,
      "step": 4923
    },
    {
      "epoch": 0.16857240671003082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7638,
      "step": 4924
    },
    {
      "epoch": 0.16860664156110922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7912,
      "step": 4925
    },
    {
      "epoch": 0.1686408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7756,
      "step": 4926
    },
    {
      "epoch": 0.168675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7722,
      "step": 4927
    },
    {
      "epoch": 0.1687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7666,
      "step": 4928
    },
    {
      "epoch": 0.1687435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7597,
      "step": 4929
    },
    {
      "epoch": 0.1687778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7707,
      "step": 4930
    },
    {
      "epoch": 0.1688120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7679,
      "step": 4931
    },
    {
      "epoch": 0.168846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7859,
      "step": 4932
    },
    {
      "epoch": 0.1688805203697364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.785,
      "step": 4933
    },
    {
      "epoch": 0.16891475522081478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7566,
      "step": 4934
    },
    {
      "epoch": 0.16894899007189318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7808,
      "step": 4935
    },
    {
      "epoch": 0.16898322492297158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7803,
      "step": 4936
    },
    {
      "epoch": 0.16901745977404997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7652,
      "step": 4937
    },
    {
      "epoch": 0.16905169462512837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7694,
      "step": 4938
    },
    {
      "epoch": 0.1690859294762068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7646,
      "step": 4939
    },
    {
      "epoch": 0.1691201643272852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7765,
      "step": 4940
    },
    {
      "epoch": 0.16915439917836358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7647,
      "step": 4941
    },
    {
      "epoch": 0.16918863402944198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7533,
      "step": 4942
    },
    {
      "epoch": 0.16922286888052038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7502,
      "step": 4943
    },
    {
      "epoch": 0.16925710373159877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7756,
      "step": 4944
    },
    {
      "epoch": 0.16929133858267717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7399,
      "step": 4945
    },
    {
      "epoch": 0.16932557343375557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7703,
      "step": 4946
    },
    {
      "epoch": 0.16935980828483396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7657,
      "step": 4947
    },
    {
      "epoch": 0.16939404313591236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7737,
      "step": 4948
    },
    {
      "epoch": 0.16942827798699076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7754,
      "step": 4949
    },
    {
      "epoch": 0.16946251283806915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7674,
      "step": 4950
    },
    {
      "epoch": 0.16949674768914755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7531,
      "step": 4951
    },
    {
      "epoch": 0.16953098254022594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7678,
      "step": 4952
    },
    {
      "epoch": 0.16956521739130434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7478,
      "step": 4953
    },
    {
      "epoch": 0.16959945224238274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7664,
      "step": 4954
    },
    {
      "epoch": 0.16963368709346113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.745,
      "step": 4955
    },
    {
      "epoch": 0.16966792194453953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7637,
      "step": 4956
    },
    {
      "epoch": 0.16970215679561793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7496,
      "step": 4957
    },
    {
      "epoch": 0.16973639164669635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7876,
      "step": 4958
    },
    {
      "epoch": 0.16977062649777475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7655,
      "step": 4959
    },
    {
      "epoch": 0.16980486134885314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7649,
      "step": 4960
    },
    {
      "epoch": 0.16983909619993154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7758,
      "step": 4961
    },
    {
      "epoch": 0.16987333105100993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7594,
      "step": 4962
    },
    {
      "epoch": 0.16990756590208833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.753,
      "step": 4963
    },
    {
      "epoch": 0.16994180075316673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7414,
      "step": 4964
    },
    {
      "epoch": 0.16997603560424512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7775,
      "step": 4965
    },
    {
      "epoch": 0.17001027045532352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7537,
      "step": 4966
    },
    {
      "epoch": 0.17004450530640192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7534,
      "step": 4967
    },
    {
      "epoch": 0.1700787401574803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7385,
      "step": 4968
    },
    {
      "epoch": 0.1701129750085587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7825,
      "step": 4969
    },
    {
      "epoch": 0.1701472098596371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7601,
      "step": 4970
    },
    {
      "epoch": 0.1701814447107155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7696,
      "step": 4971
    },
    {
      "epoch": 0.1702156795617939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7764,
      "step": 4972
    },
    {
      "epoch": 0.1702499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.776,
      "step": 4973
    },
    {
      "epoch": 0.1702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7852,
      "step": 4974
    },
    {
      "epoch": 0.1703183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7428,
      "step": 4975
    },
    {
      "epoch": 0.1703526189661075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7474,
      "step": 4976
    },
    {
      "epoch": 0.1703868538171859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7707,
      "step": 4977
    },
    {
      "epoch": 0.1704210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7471,
      "step": 4978
    },
    {
      "epoch": 0.1704553235193427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7637,
      "step": 4979
    },
    {
      "epoch": 0.1704895583704211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7544,
      "step": 4980
    },
    {
      "epoch": 0.1705237932214995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7778,
      "step": 4981
    },
    {
      "epoch": 0.1705580280725779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7545,
      "step": 4982
    },
    {
      "epoch": 0.17059226292365628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7514,
      "step": 4983
    },
    {
      "epoch": 0.17062649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7482,
      "step": 4984
    },
    {
      "epoch": 0.17066073262581308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7406,
      "step": 4985
    },
    {
      "epoch": 0.17069496747689147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7479,
      "step": 4986
    },
    {
      "epoch": 0.17072920232796987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7908,
      "step": 4987
    },
    {
      "epoch": 0.17076343717904827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7633,
      "step": 4988
    },
    {
      "epoch": 0.17079767203012666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7699,
      "step": 4989
    },
    {
      "epoch": 0.17083190688120506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7469,
      "step": 4990
    },
    {
      "epoch": 0.17086614173228346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7643,
      "step": 4991
    },
    {
      "epoch": 0.17090037658336185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7554,
      "step": 4992
    },
    {
      "epoch": 0.17093461143444025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7651,
      "step": 4993
    },
    {
      "epoch": 0.17096884628551864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7584,
      "step": 4994
    },
    {
      "epoch": 0.17100308113659707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7535,
      "step": 4995
    },
    {
      "epoch": 0.17103731598767546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.755,
      "step": 4996
    },
    {
      "epoch": 0.17107155083875386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7763,
      "step": 4997
    },
    {
      "epoch": 0.17110578568983226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7383,
      "step": 4998
    },
    {
      "epoch": 0.17114002054091065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7695,
      "step": 4999
    },
    {
      "epoch": 0.17117425539198905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7269,
      "step": 5000
    },
    {
      "epoch": 0.17120849024306745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7624,
      "step": 5001
    },
    {
      "epoch": 0.17124272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7709,
      "step": 5002
    },
    {
      "epoch": 0.17127695994522424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7672,
      "step": 5003
    },
    {
      "epoch": 0.17131119479630263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7505,
      "step": 5004
    },
    {
      "epoch": 0.17134542964738103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.752,
      "step": 5005
    },
    {
      "epoch": 0.17137966449845943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7818,
      "step": 5006
    },
    {
      "epoch": 0.17141389934953782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7654,
      "step": 5007
    },
    {
      "epoch": 0.17144813420061622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7444,
      "step": 5008
    },
    {
      "epoch": 0.17148236905169462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7545,
      "step": 5009
    },
    {
      "epoch": 0.171516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7765,
      "step": 5010
    },
    {
      "epoch": 0.1715508387538514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.767,
      "step": 5011
    },
    {
      "epoch": 0.1715850736049298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7701,
      "step": 5012
    },
    {
      "epoch": 0.17161930845600823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7575,
      "step": 5013
    },
    {
      "epoch": 0.17165354330708663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7528,
      "step": 5014
    },
    {
      "epoch": 0.17168777815816502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7595,
      "step": 5015
    },
    {
      "epoch": 0.17172201300924342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7275,
      "step": 5016
    },
    {
      "epoch": 0.17175624786032181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7615,
      "step": 5017
    },
    {
      "epoch": 0.1717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.768,
      "step": 5018
    },
    {
      "epoch": 0.1718247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7642,
      "step": 5019
    },
    {
      "epoch": 0.171858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7372,
      "step": 5020
    },
    {
      "epoch": 0.1718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.751,
      "step": 5021
    },
    {
      "epoch": 0.1719274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7617,
      "step": 5022
    },
    {
      "epoch": 0.1719616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7534,
      "step": 5023
    },
    {
      "epoch": 0.1719958918178706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.748,
      "step": 5024
    },
    {
      "epoch": 0.17203012666894898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7482,
      "step": 5025
    },
    {
      "epoch": 0.17206436152002738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7532,
      "step": 5026
    },
    {
      "epoch": 0.17209859637110578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7573,
      "step": 5027
    },
    {
      "epoch": 0.17213283122218417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7538,
      "step": 5028
    },
    {
      "epoch": 0.17216706607326257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7453,
      "step": 5029
    },
    {
      "epoch": 0.17220130092434097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.772,
      "step": 5030
    },
    {
      "epoch": 0.1722355357754194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7424,
      "step": 5031
    },
    {
      "epoch": 0.1722697706264978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.747,
      "step": 5032
    },
    {
      "epoch": 0.17230400547757618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7413,
      "step": 5033
    },
    {
      "epoch": 0.17233824032865458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7695,
      "step": 5034
    },
    {
      "epoch": 0.17237247517973298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7427,
      "step": 5035
    },
    {
      "epoch": 0.17240671003081137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7445,
      "step": 5036
    },
    {
      "epoch": 0.17244094488188977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7511,
      "step": 5037
    },
    {
      "epoch": 0.17247517973296816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7515,
      "step": 5038
    },
    {
      "epoch": 0.17250941458404656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7453,
      "step": 5039
    },
    {
      "epoch": 0.17254364943512496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7486,
      "step": 5040
    },
    {
      "epoch": 0.17257788428620335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.766,
      "step": 5041
    },
    {
      "epoch": 0.17261211913728175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7696,
      "step": 5042
    },
    {
      "epoch": 0.17264635398836015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7671,
      "step": 5043
    },
    {
      "epoch": 0.17268058883943854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7503,
      "step": 5044
    },
    {
      "epoch": 0.17271482369051694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7754,
      "step": 5045
    },
    {
      "epoch": 0.17274905854159534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7496,
      "step": 5046
    },
    {
      "epoch": 0.17278329339267373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.747,
      "step": 5047
    },
    {
      "epoch": 0.17281752824375213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7819,
      "step": 5048
    },
    {
      "epoch": 0.17285176309483052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7478,
      "step": 5049
    },
    {
      "epoch": 0.17288599794590895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7524,
      "step": 5050
    },
    {
      "epoch": 0.17292023279698734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7571,
      "step": 5051
    },
    {
      "epoch": 0.17295446764806574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7549,
      "step": 5052
    },
    {
      "epoch": 0.17298870249914414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7334,
      "step": 5053
    },
    {
      "epoch": 0.17302293735022253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7552,
      "step": 5054
    },
    {
      "epoch": 0.17305717220130093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7474,
      "step": 5055
    },
    {
      "epoch": 0.17309140705237933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.749,
      "step": 5056
    },
    {
      "epoch": 0.17312564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7452,
      "step": 5057
    },
    {
      "epoch": 0.17315987675453612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7395,
      "step": 5058
    },
    {
      "epoch": 0.17319411160561451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7513,
      "step": 5059
    },
    {
      "epoch": 0.1732283464566929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7427,
      "step": 5060
    },
    {
      "epoch": 0.1732625813077713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7141,
      "step": 5061
    },
    {
      "epoch": 0.1732968161588497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7471,
      "step": 5062
    },
    {
      "epoch": 0.1733310510099281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7657,
      "step": 5063
    },
    {
      "epoch": 0.1733652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7433,
      "step": 5064
    },
    {
      "epoch": 0.1733995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7435,
      "step": 5065
    },
    {
      "epoch": 0.1734337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7512,
      "step": 5066
    },
    {
      "epoch": 0.17346799041424169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7164,
      "step": 5067
    },
    {
      "epoch": 0.1735022252653201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7375,
      "step": 5068
    },
    {
      "epoch": 0.1735364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7351,
      "step": 5069
    },
    {
      "epoch": 0.1735706949674769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7677,
      "step": 5070
    },
    {
      "epoch": 0.1736049298185553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7427,
      "step": 5071
    },
    {
      "epoch": 0.1736391646696337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7436,
      "step": 5072
    },
    {
      "epoch": 0.1736733995207121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7159,
      "step": 5073
    },
    {
      "epoch": 0.1737076343717905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7361,
      "step": 5074
    },
    {
      "epoch": 0.17374186922286888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7509,
      "step": 5075
    },
    {
      "epoch": 0.17377610407394728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7456,
      "step": 5076
    },
    {
      "epoch": 0.17381033892502568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7369,
      "step": 5077
    },
    {
      "epoch": 0.17384457377610407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7685,
      "step": 5078
    },
    {
      "epoch": 0.17387880862718247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7597,
      "step": 5079
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7461,
      "step": 5080
    },
    {
      "epoch": 0.17394727832933926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7262,
      "step": 5081
    },
    {
      "epoch": 0.17398151318041766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7416,
      "step": 5082
    },
    {
      "epoch": 0.17401574803149605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7469,
      "step": 5083
    },
    {
      "epoch": 0.17404998288257445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7549,
      "step": 5084
    },
    {
      "epoch": 0.17408421773365285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7605,
      "step": 5085
    },
    {
      "epoch": 0.17411845258473127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7375,
      "step": 5086
    },
    {
      "epoch": 0.17415268743580967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7528,
      "step": 5087
    },
    {
      "epoch": 0.17418692228688806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7409,
      "step": 5088
    },
    {
      "epoch": 0.17422115713796646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7346,
      "step": 5089
    },
    {
      "epoch": 0.17425539198904486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7316,
      "step": 5090
    },
    {
      "epoch": 0.17428962684012325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7429,
      "step": 5091
    },
    {
      "epoch": 0.17432386169120165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7521,
      "step": 5092
    },
    {
      "epoch": 0.17435809654228004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7366,
      "step": 5093
    },
    {
      "epoch": 0.17439233139335844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7585,
      "step": 5094
    },
    {
      "epoch": 0.17442656624443684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7366,
      "step": 5095
    },
    {
      "epoch": 0.17446080109551523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7684,
      "step": 5096
    },
    {
      "epoch": 0.17449503594659363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7708,
      "step": 5097
    },
    {
      "epoch": 0.17452927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7643,
      "step": 5098
    },
    {
      "epoch": 0.17456350564875042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7405,
      "step": 5099
    },
    {
      "epoch": 0.17459774049982882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7384,
      "step": 5100
    },
    {
      "epoch": 0.17463197535090721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7419,
      "step": 5101
    },
    {
      "epoch": 0.1746662102019856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7429,
      "step": 5102
    },
    {
      "epoch": 0.174700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7332,
      "step": 5103
    },
    {
      "epoch": 0.1747346799041424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7314,
      "step": 5104
    },
    {
      "epoch": 0.17476891475522083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.735,
      "step": 5105
    },
    {
      "epoch": 0.17480314960629922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7527,
      "step": 5106
    },
    {
      "epoch": 0.17483738445737762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7611,
      "step": 5107
    },
    {
      "epoch": 0.17487161930845602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.734,
      "step": 5108
    },
    {
      "epoch": 0.1749058541595344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7394,
      "step": 5109
    },
    {
      "epoch": 0.1749400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7533,
      "step": 5110
    },
    {
      "epoch": 0.1749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7405,
      "step": 5111
    },
    {
      "epoch": 0.1750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7494,
      "step": 5112
    },
    {
      "epoch": 0.175042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7521,
      "step": 5113
    },
    {
      "epoch": 0.1750770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7715,
      "step": 5114
    },
    {
      "epoch": 0.1751112632660048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7212,
      "step": 5115
    },
    {
      "epoch": 0.1751454981170832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.741,
      "step": 5116
    },
    {
      "epoch": 0.17517973296816158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7336,
      "step": 5117
    },
    {
      "epoch": 0.17521396781923998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7501,
      "step": 5118
    },
    {
      "epoch": 0.17524820267031838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7312,
      "step": 5119
    },
    {
      "epoch": 0.17528243752139677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7711,
      "step": 5120
    },
    {
      "epoch": 0.17531667237247517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7236,
      "step": 5121
    },
    {
      "epoch": 0.17535090722355356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7355,
      "step": 5122
    },
    {
      "epoch": 0.175385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7531,
      "step": 5123
    },
    {
      "epoch": 0.17541937692571039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.734,
      "step": 5124
    },
    {
      "epoch": 0.17545361177678878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7315,
      "step": 5125
    },
    {
      "epoch": 0.17548784662786718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7132,
      "step": 5126
    },
    {
      "epoch": 0.17552208147894557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7386,
      "step": 5127
    },
    {
      "epoch": 0.17555631633002397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7355,
      "step": 5128
    },
    {
      "epoch": 0.17559055118110237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7464,
      "step": 5129
    },
    {
      "epoch": 0.17562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7663,
      "step": 5130
    },
    {
      "epoch": 0.17565902088325916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7417,
      "step": 5131
    },
    {
      "epoch": 0.17569325573433756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7402,
      "step": 5132
    },
    {
      "epoch": 0.17572749058541595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7329,
      "step": 5133
    },
    {
      "epoch": 0.17576172543649435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7416,
      "step": 5134
    },
    {
      "epoch": 0.17579596028757274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7485,
      "step": 5135
    },
    {
      "epoch": 0.17583019513865114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.734,
      "step": 5136
    },
    {
      "epoch": 0.17586442998972954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7381,
      "step": 5137
    },
    {
      "epoch": 0.17589866484080793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7424,
      "step": 5138
    },
    {
      "epoch": 0.17593289969188633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7204,
      "step": 5139
    },
    {
      "epoch": 0.17596713454296473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.764,
      "step": 5140
    },
    {
      "epoch": 0.17600136939404312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7384,
      "step": 5141
    },
    {
      "epoch": 0.17603560424512155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7357,
      "step": 5142
    },
    {
      "epoch": 0.17606983909619994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7588,
      "step": 5143
    },
    {
      "epoch": 0.17610407394727834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7414,
      "step": 5144
    },
    {
      "epoch": 0.17613830879835674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7351,
      "step": 5145
    },
    {
      "epoch": 0.17617254364943513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7406,
      "step": 5146
    },
    {
      "epoch": 0.17620677850051353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7274,
      "step": 5147
    },
    {
      "epoch": 0.17624101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7366,
      "step": 5148
    },
    {
      "epoch": 0.17627524820267032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7245,
      "step": 5149
    },
    {
      "epoch": 0.17630948305374872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6909,
      "step": 5150
    },
    {
      "epoch": 0.1763437179048271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7643,
      "step": 5151
    },
    {
      "epoch": 0.1763779527559055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.764,
      "step": 5152
    },
    {
      "epoch": 0.1764121876069839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7387,
      "step": 5153
    },
    {
      "epoch": 0.1764464224580623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7281,
      "step": 5154
    },
    {
      "epoch": 0.1764806573091407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7366,
      "step": 5155
    },
    {
      "epoch": 0.1765148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7394,
      "step": 5156
    },
    {
      "epoch": 0.1765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7519,
      "step": 5157
    },
    {
      "epoch": 0.1765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7449,
      "step": 5158
    },
    {
      "epoch": 0.17661759671345428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7373,
      "step": 5159
    },
    {
      "epoch": 0.1766518315645327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7471,
      "step": 5160
    },
    {
      "epoch": 0.1766860664156111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7218,
      "step": 5161
    },
    {
      "epoch": 0.1767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.728,
      "step": 5162
    },
    {
      "epoch": 0.1767545361177679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7463,
      "step": 5163
    },
    {
      "epoch": 0.1767887709688463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7384,
      "step": 5164
    },
    {
      "epoch": 0.1768230058199247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.744,
      "step": 5165
    },
    {
      "epoch": 0.17685724067100309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7347,
      "step": 5166
    },
    {
      "epoch": 0.17689147552208148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7385,
      "step": 5167
    },
    {
      "epoch": 0.17692571037315988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7365,
      "step": 5168
    },
    {
      "epoch": 0.17695994522423827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7372,
      "step": 5169
    },
    {
      "epoch": 0.17699418007531667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7166,
      "step": 5170
    },
    {
      "epoch": 0.17702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7404,
      "step": 5171
    },
    {
      "epoch": 0.17706264977747346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7207,
      "step": 5172
    },
    {
      "epoch": 0.17709688462855186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7396,
      "step": 5173
    },
    {
      "epoch": 0.17713111947963026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7419,
      "step": 5174
    },
    {
      "epoch": 0.17716535433070865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7445,
      "step": 5175
    },
    {
      "epoch": 0.17719958918178705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7275,
      "step": 5176
    },
    {
      "epoch": 0.17723382403286544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7131,
      "step": 5177
    },
    {
      "epoch": 0.17726805888394387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7288,
      "step": 5178
    },
    {
      "epoch": 0.17730229373502226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7533,
      "step": 5179
    },
    {
      "epoch": 0.17733652858610066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7674,
      "step": 5180
    },
    {
      "epoch": 0.17737076343717906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.75,
      "step": 5181
    },
    {
      "epoch": 0.17740499828825745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7374,
      "step": 5182
    },
    {
      "epoch": 0.17743923313933585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.737,
      "step": 5183
    },
    {
      "epoch": 0.17747346799041425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7469,
      "step": 5184
    },
    {
      "epoch": 0.17750770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7051,
      "step": 5185
    },
    {
      "epoch": 0.17754193769257104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7259,
      "step": 5186
    },
    {
      "epoch": 0.17757617254364944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7397,
      "step": 5187
    },
    {
      "epoch": 0.17761040739472783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7393,
      "step": 5188
    },
    {
      "epoch": 0.17764464224580623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7195,
      "step": 5189
    },
    {
      "epoch": 0.17767887709688462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7335,
      "step": 5190
    },
    {
      "epoch": 0.17771311194796302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7266,
      "step": 5191
    },
    {
      "epoch": 0.17774734679904142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7389,
      "step": 5192
    },
    {
      "epoch": 0.1777815816501198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7407,
      "step": 5193
    },
    {
      "epoch": 0.1778158165011982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7044,
      "step": 5194
    },
    {
      "epoch": 0.1778500513522766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7534,
      "step": 5195
    },
    {
      "epoch": 0.177884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7304,
      "step": 5196
    },
    {
      "epoch": 0.17791852105443343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7064,
      "step": 5197
    },
    {
      "epoch": 0.17795275590551182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.726,
      "step": 5198
    },
    {
      "epoch": 0.17798699075659022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7254,
      "step": 5199
    },
    {
      "epoch": 0.17802122560766862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.73,
      "step": 5200
    },
    {
      "epoch": 0.178055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7231,
      "step": 5201
    },
    {
      "epoch": 0.1780896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7325,
      "step": 5202
    },
    {
      "epoch": 0.1781239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7445,
      "step": 5203
    },
    {
      "epoch": 0.1781581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7425,
      "step": 5204
    },
    {
      "epoch": 0.1781923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7145,
      "step": 5205
    },
    {
      "epoch": 0.178226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.738,
      "step": 5206
    },
    {
      "epoch": 0.1782608695652174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7305,
      "step": 5207
    },
    {
      "epoch": 0.17829510441629579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7151,
      "step": 5208
    },
    {
      "epoch": 0.17832933926737418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7217,
      "step": 5209
    },
    {
      "epoch": 0.17836357411845258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7157,
      "step": 5210
    },
    {
      "epoch": 0.17839780896953097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7367,
      "step": 5211
    },
    {
      "epoch": 0.17843204382060937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7358,
      "step": 5212
    },
    {
      "epoch": 0.17846627867168777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7163,
      "step": 5213
    },
    {
      "epoch": 0.17850051352276616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7282,
      "step": 5214
    },
    {
      "epoch": 0.1785347483738446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7053,
      "step": 5215
    },
    {
      "epoch": 0.17856898322492298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7267,
      "step": 5216
    },
    {
      "epoch": 0.17860321807600138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7179,
      "step": 5217
    },
    {
      "epoch": 0.17863745292707978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7493,
      "step": 5218
    },
    {
      "epoch": 0.17867168777815817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7384,
      "step": 5219
    },
    {
      "epoch": 0.17870592262923657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.733,
      "step": 5220
    },
    {
      "epoch": 0.17874015748031497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7286,
      "step": 5221
    },
    {
      "epoch": 0.17877439233139336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7326,
      "step": 5222
    },
    {
      "epoch": 0.17880862718247176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7589,
      "step": 5223
    },
    {
      "epoch": 0.17884286203355015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7547,
      "step": 5224
    },
    {
      "epoch": 0.17887709688462855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7278,
      "step": 5225
    },
    {
      "epoch": 0.17891133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7488,
      "step": 5226
    },
    {
      "epoch": 0.17894556658678534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7152,
      "step": 5227
    },
    {
      "epoch": 0.17897980143786374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7178,
      "step": 5228
    },
    {
      "epoch": 0.17901403628894214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7323,
      "step": 5229
    },
    {
      "epoch": 0.17904827114002053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7552,
      "step": 5230
    },
    {
      "epoch": 0.17908250599109893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.742,
      "step": 5231
    },
    {
      "epoch": 0.17911674084217732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7184,
      "step": 5232
    },
    {
      "epoch": 0.17915097569325572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7176,
      "step": 5233
    },
    {
      "epoch": 0.17918521054433414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7232,
      "step": 5234
    },
    {
      "epoch": 0.17921944539541254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7294,
      "step": 5235
    },
    {
      "epoch": 0.17925368024649094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7046,
      "step": 5236
    },
    {
      "epoch": 0.17928791509756933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.714,
      "step": 5237
    },
    {
      "epoch": 0.17932214994864773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7267,
      "step": 5238
    },
    {
      "epoch": 0.17935638479972613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7293,
      "step": 5239
    },
    {
      "epoch": 0.17939061965080452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.721,
      "step": 5240
    },
    {
      "epoch": 0.17942485450188292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7403,
      "step": 5241
    },
    {
      "epoch": 0.17945908935296132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.74,
      "step": 5242
    },
    {
      "epoch": 0.1794933242040397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7195,
      "step": 5243
    },
    {
      "epoch": 0.1795275590551181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7434,
      "step": 5244
    },
    {
      "epoch": 0.1795617939061965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7366,
      "step": 5245
    },
    {
      "epoch": 0.1795960287572749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7273,
      "step": 5246
    },
    {
      "epoch": 0.1796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7182,
      "step": 5247
    },
    {
      "epoch": 0.1796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7282,
      "step": 5248
    },
    {
      "epoch": 0.1796987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7234,
      "step": 5249
    },
    {
      "epoch": 0.17973296816158849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7423,
      "step": 5250
    },
    {
      "epoch": 0.17976720301266688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7207,
      "step": 5251
    },
    {
      "epoch": 0.1798014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7406,
      "step": 5252
    },
    {
      "epoch": 0.1798356727148237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7392,
      "step": 5253
    },
    {
      "epoch": 0.1798699075659021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7446,
      "step": 5254
    },
    {
      "epoch": 0.1799041424169805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7342,
      "step": 5255
    },
    {
      "epoch": 0.1799383772680589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7473,
      "step": 5256
    },
    {
      "epoch": 0.1799726121191373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7146,
      "step": 5257
    },
    {
      "epoch": 0.18000684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7338,
      "step": 5258
    },
    {
      "epoch": 0.18004108182129408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7322,
      "step": 5259
    },
    {
      "epoch": 0.18007531667237248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7322,
      "step": 5260
    },
    {
      "epoch": 0.18010955152345087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7282,
      "step": 5261
    },
    {
      "epoch": 0.18014378637452927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7382,
      "step": 5262
    },
    {
      "epoch": 0.18017802122560767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7326,
      "step": 5263
    },
    {
      "epoch": 0.18021225607668606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7517,
      "step": 5264
    },
    {
      "epoch": 0.18024649092776446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7082,
      "step": 5265
    },
    {
      "epoch": 0.18028072577884285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7245,
      "step": 5266
    },
    {
      "epoch": 0.18031496062992125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.733,
      "step": 5267
    },
    {
      "epoch": 0.18034919548099965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7353,
      "step": 5268
    },
    {
      "epoch": 0.18038343033207804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7241,
      "step": 5269
    },
    {
      "epoch": 0.18041766518315647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7558,
      "step": 5270
    },
    {
      "epoch": 0.18045190003423486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7298,
      "step": 5271
    },
    {
      "epoch": 0.18048613488531326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.714,
      "step": 5272
    },
    {
      "epoch": 0.18052036973639166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7329,
      "step": 5273
    },
    {
      "epoch": 0.18055460458747005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7093,
      "step": 5274
    },
    {
      "epoch": 0.18058883943854845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7242,
      "step": 5275
    },
    {
      "epoch": 0.18062307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7337,
      "step": 5276
    },
    {
      "epoch": 0.18065730914070524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7132,
      "step": 5277
    },
    {
      "epoch": 0.18069154399178364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.701,
      "step": 5278
    },
    {
      "epoch": 0.18072577884286203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7325,
      "step": 5279
    },
    {
      "epoch": 0.18076001369394043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7342,
      "step": 5280
    },
    {
      "epoch": 0.18079424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7176,
      "step": 5281
    },
    {
      "epoch": 0.18082848339609722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7223,
      "step": 5282
    },
    {
      "epoch": 0.18086271824717562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7246,
      "step": 5283
    },
    {
      "epoch": 0.18089695309825402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6949,
      "step": 5284
    },
    {
      "epoch": 0.1809311879493324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7098,
      "step": 5285
    },
    {
      "epoch": 0.1809654228004108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6932,
      "step": 5286
    },
    {
      "epoch": 0.1809996576514892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7157,
      "step": 5287
    },
    {
      "epoch": 0.1810338925025676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7027,
      "step": 5288
    },
    {
      "epoch": 0.18106812735364602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7304,
      "step": 5289
    },
    {
      "epoch": 0.18110236220472442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7338,
      "step": 5290
    },
    {
      "epoch": 0.18113659705580282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7053,
      "step": 5291
    },
    {
      "epoch": 0.1811708319068812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7134,
      "step": 5292
    },
    {
      "epoch": 0.1812050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7314,
      "step": 5293
    },
    {
      "epoch": 0.181239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7022,
      "step": 5294
    },
    {
      "epoch": 0.1812735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7262,
      "step": 5295
    },
    {
      "epoch": 0.1813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7259,
      "step": 5296
    },
    {
      "epoch": 0.1813420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7564,
      "step": 5297
    },
    {
      "epoch": 0.1813762410133516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7362,
      "step": 5298
    },
    {
      "epoch": 0.18141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.73,
      "step": 5299
    },
    {
      "epoch": 0.18144471071550838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7194,
      "step": 5300
    },
    {
      "epoch": 0.18147894556658678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7344,
      "step": 5301
    },
    {
      "epoch": 0.18151318041766518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7279,
      "step": 5302
    },
    {
      "epoch": 0.18154741526874357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6946,
      "step": 5303
    },
    {
      "epoch": 0.18158165011982197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7141,
      "step": 5304
    },
    {
      "epoch": 0.18161588497090037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7331,
      "step": 5305
    },
    {
      "epoch": 0.18165011982197876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.726,
      "step": 5306
    },
    {
      "epoch": 0.18168435467305719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7077,
      "step": 5307
    },
    {
      "epoch": 0.18171858952413558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7521,
      "step": 5308
    },
    {
      "epoch": 0.18175282437521398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7204,
      "step": 5309
    },
    {
      "epoch": 0.18178705922629237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7298,
      "step": 5310
    },
    {
      "epoch": 0.18182129407737077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7079,
      "step": 5311
    },
    {
      "epoch": 0.18185552892844917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7087,
      "step": 5312
    },
    {
      "epoch": 0.18188976377952756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7129,
      "step": 5313
    },
    {
      "epoch": 0.18192399863060596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7304,
      "step": 5314
    },
    {
      "epoch": 0.18195823348168436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6842,
      "step": 5315
    },
    {
      "epoch": 0.18199246833276275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7177,
      "step": 5316
    },
    {
      "epoch": 0.18202670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7213,
      "step": 5317
    },
    {
      "epoch": 0.18206093803491954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6996,
      "step": 5318
    },
    {
      "epoch": 0.18209517288599794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7123,
      "step": 5319
    },
    {
      "epoch": 0.18212940773707634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7085,
      "step": 5320
    },
    {
      "epoch": 0.18216364258815473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7198,
      "step": 5321
    },
    {
      "epoch": 0.18219787743923313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7076,
      "step": 5322
    },
    {
      "epoch": 0.18223211229031153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7371,
      "step": 5323
    },
    {
      "epoch": 0.18226634714138992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.73,
      "step": 5324
    },
    {
      "epoch": 0.18230058199246832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7024,
      "step": 5325
    },
    {
      "epoch": 0.18233481684354674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.712,
      "step": 5326
    },
    {
      "epoch": 0.18236905169462514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7303,
      "step": 5327
    },
    {
      "epoch": 0.18240328654570354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7072,
      "step": 5328
    },
    {
      "epoch": 0.18243752139678193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7149,
      "step": 5329
    },
    {
      "epoch": 0.18247175624786033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7052,
      "step": 5330
    },
    {
      "epoch": 0.18250599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.724,
      "step": 5331
    },
    {
      "epoch": 0.18254022595001712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7157,
      "step": 5332
    },
    {
      "epoch": 0.18257446080109552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7289,
      "step": 5333
    },
    {
      "epoch": 0.1826086956521739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6841,
      "step": 5334
    },
    {
      "epoch": 0.1826429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7478,
      "step": 5335
    },
    {
      "epoch": 0.1826771653543307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7087,
      "step": 5336
    },
    {
      "epoch": 0.1827114002054091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7268,
      "step": 5337
    },
    {
      "epoch": 0.1827456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.73,
      "step": 5338
    },
    {
      "epoch": 0.1827798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6975,
      "step": 5339
    },
    {
      "epoch": 0.1828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7184,
      "step": 5340
    },
    {
      "epoch": 0.1828483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7305,
      "step": 5341
    },
    {
      "epoch": 0.18288257446080108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7371,
      "step": 5342
    },
    {
      "epoch": 0.18291680931187948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.714,
      "step": 5343
    },
    {
      "epoch": 0.1829510441629579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7352,
      "step": 5344
    },
    {
      "epoch": 0.1829852790140363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7178,
      "step": 5345
    },
    {
      "epoch": 0.1830195138651147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7137,
      "step": 5346
    },
    {
      "epoch": 0.1830537487161931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7126,
      "step": 5347
    },
    {
      "epoch": 0.1830879835672715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7017,
      "step": 5348
    },
    {
      "epoch": 0.18312221841834989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7198,
      "step": 5349
    },
    {
      "epoch": 0.18315645326942828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6969,
      "step": 5350
    },
    {
      "epoch": 0.18319068812050668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7042,
      "step": 5351
    },
    {
      "epoch": 0.18322492297158507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.732,
      "step": 5352
    },
    {
      "epoch": 0.18325915782266347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7134,
      "step": 5353
    },
    {
      "epoch": 0.18329339267374187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7057,
      "step": 5354
    },
    {
      "epoch": 0.18332762752482026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7005,
      "step": 5355
    },
    {
      "epoch": 0.18336186237589866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7251,
      "step": 5356
    },
    {
      "epoch": 0.18339609722697706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.713,
      "step": 5357
    },
    {
      "epoch": 0.18343033207805545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6886,
      "step": 5358
    },
    {
      "epoch": 0.18346456692913385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7192,
      "step": 5359
    },
    {
      "epoch": 0.18349880178021225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7201,
      "step": 5360
    },
    {
      "epoch": 0.18353303663129064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7272,
      "step": 5361
    },
    {
      "epoch": 0.18356727148236907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7071,
      "step": 5362
    },
    {
      "epoch": 0.18360150633344746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7109,
      "step": 5363
    },
    {
      "epoch": 0.18363574118452586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7225,
      "step": 5364
    },
    {
      "epoch": 0.18366997603560425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7058,
      "step": 5365
    },
    {
      "epoch": 0.18370421088668265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7208,
      "step": 5366
    },
    {
      "epoch": 0.18373844573776105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7063,
      "step": 5367
    },
    {
      "epoch": 0.18377268058883944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7088,
      "step": 5368
    },
    {
      "epoch": 0.18380691543991784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7015,
      "step": 5369
    },
    {
      "epoch": 0.18384115029099624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7306,
      "step": 5370
    },
    {
      "epoch": 0.18387538514207463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7163,
      "step": 5371
    },
    {
      "epoch": 0.18390961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6972,
      "step": 5372
    },
    {
      "epoch": 0.18394385484423142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7388,
      "step": 5373
    },
    {
      "epoch": 0.18397808969530982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6975,
      "step": 5374
    },
    {
      "epoch": 0.18401232454638822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7142,
      "step": 5375
    },
    {
      "epoch": 0.1840465593974666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7065,
      "step": 5376
    },
    {
      "epoch": 0.184080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7289,
      "step": 5377
    },
    {
      "epoch": 0.1841150290996234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.699,
      "step": 5378
    },
    {
      "epoch": 0.1841492639507018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7287,
      "step": 5379
    },
    {
      "epoch": 0.1841834988017802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6981,
      "step": 5380
    },
    {
      "epoch": 0.18421773365285862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6825,
      "step": 5381
    },
    {
      "epoch": 0.18425196850393702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7214,
      "step": 5382
    },
    {
      "epoch": 0.18428620335501542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7238,
      "step": 5383
    },
    {
      "epoch": 0.1843204382060938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7292,
      "step": 5384
    },
    {
      "epoch": 0.1843546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7129,
      "step": 5385
    },
    {
      "epoch": 0.1843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7137,
      "step": 5386
    },
    {
      "epoch": 0.184423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6883,
      "step": 5387
    },
    {
      "epoch": 0.1844573776104074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7046,
      "step": 5388
    },
    {
      "epoch": 0.1844916124614858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6985,
      "step": 5389
    },
    {
      "epoch": 0.1845258473125642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6985,
      "step": 5390
    },
    {
      "epoch": 0.18456008216364259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.748,
      "step": 5391
    },
    {
      "epoch": 0.18459431701472098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7187,
      "step": 5392
    },
    {
      "epoch": 0.18462855186579938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7195,
      "step": 5393
    },
    {
      "epoch": 0.18466278671687777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7083,
      "step": 5394
    },
    {
      "epoch": 0.18469702156795617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7157,
      "step": 5395
    },
    {
      "epoch": 0.18473125641903457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7022,
      "step": 5396
    },
    {
      "epoch": 0.18476549127011296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7053,
      "step": 5397
    },
    {
      "epoch": 0.18479972612119136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6772,
      "step": 5398
    },
    {
      "epoch": 0.18483396097226978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7278,
      "step": 5399
    },
    {
      "epoch": 0.18486819582334818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6888,
      "step": 5400
    },
    {
      "epoch": 0.18490243067442658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7085,
      "step": 5401
    },
    {
      "epoch": 0.18493666552550497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7043,
      "step": 5402
    },
    {
      "epoch": 0.18497090037658337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7505,
      "step": 5403
    },
    {
      "epoch": 0.18500513522766177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7006,
      "step": 5404
    },
    {
      "epoch": 0.18503937007874016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7286,
      "step": 5405
    },
    {
      "epoch": 0.18507360492981856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7142,
      "step": 5406
    },
    {
      "epoch": 0.18510783978089695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7065,
      "step": 5407
    },
    {
      "epoch": 0.18514207463197535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7136,
      "step": 5408
    },
    {
      "epoch": 0.18517630948305375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7237,
      "step": 5409
    },
    {
      "epoch": 0.18521054433413214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6994,
      "step": 5410
    },
    {
      "epoch": 0.18524477918521054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6958,
      "step": 5411
    },
    {
      "epoch": 0.18527901403628894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7154,
      "step": 5412
    },
    {
      "epoch": 0.18531324888736733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7058,
      "step": 5413
    },
    {
      "epoch": 0.18534748373844573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7111,
      "step": 5414
    },
    {
      "epoch": 0.18538171858952412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7254,
      "step": 5415
    },
    {
      "epoch": 0.18541595344060252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7043,
      "step": 5416
    },
    {
      "epoch": 0.18545018829168092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6963,
      "step": 5417
    },
    {
      "epoch": 0.18548442314275934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7215,
      "step": 5418
    },
    {
      "epoch": 0.18551865799383774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7112,
      "step": 5419
    },
    {
      "epoch": 0.18555289284491613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7081,
      "step": 5420
    },
    {
      "epoch": 0.18558712769599453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7419,
      "step": 5421
    },
    {
      "epoch": 0.18562136254707293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7162,
      "step": 5422
    },
    {
      "epoch": 0.18565559739815132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6898,
      "step": 5423
    },
    {
      "epoch": 0.18568983224922972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6916,
      "step": 5424
    },
    {
      "epoch": 0.18572406710030812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6986,
      "step": 5425
    },
    {
      "epoch": 0.1857583019513865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7297,
      "step": 5426
    },
    {
      "epoch": 0.1857925368024649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6944,
      "step": 5427
    },
    {
      "epoch": 0.1858267716535433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7138,
      "step": 5428
    },
    {
      "epoch": 0.1858610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7032,
      "step": 5429
    },
    {
      "epoch": 0.1858952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7163,
      "step": 5430
    },
    {
      "epoch": 0.1859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6917,
      "step": 5431
    },
    {
      "epoch": 0.1859637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.723,
      "step": 5432
    },
    {
      "epoch": 0.18599794590893529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7077,
      "step": 5433
    },
    {
      "epoch": 0.18603218076001368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6827,
      "step": 5434
    },
    {
      "epoch": 0.18606641561109208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.672,
      "step": 5435
    },
    {
      "epoch": 0.1861006504621705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6891,
      "step": 5436
    },
    {
      "epoch": 0.1861348853132489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7121,
      "step": 5437
    },
    {
      "epoch": 0.1861691201643273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6899,
      "step": 5438
    },
    {
      "epoch": 0.1862033550154057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7015,
      "step": 5439
    },
    {
      "epoch": 0.1862375898664841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7088,
      "step": 5440
    },
    {
      "epoch": 0.18627182471756248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7109,
      "step": 5441
    },
    {
      "epoch": 0.18630605956864088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.698,
      "step": 5442
    },
    {
      "epoch": 0.18634029441971928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7122,
      "step": 5443
    },
    {
      "epoch": 0.18637452927079767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6955,
      "step": 5444
    },
    {
      "epoch": 0.18640876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6917,
      "step": 5445
    },
    {
      "epoch": 0.18644299897295447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6923,
      "step": 5446
    },
    {
      "epoch": 0.18647723382403286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6888,
      "step": 5447
    },
    {
      "epoch": 0.18651146867511126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6853,
      "step": 5448
    },
    {
      "epoch": 0.18654570352618965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7317,
      "step": 5449
    },
    {
      "epoch": 0.18657993837726805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7094,
      "step": 5450
    },
    {
      "epoch": 0.18661417322834645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7004,
      "step": 5451
    },
    {
      "epoch": 0.18664840807942484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6899,
      "step": 5452
    },
    {
      "epoch": 0.18668264293050324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6925,
      "step": 5453
    },
    {
      "epoch": 0.18671687778158166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7158,
      "step": 5454
    },
    {
      "epoch": 0.18675111263266006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7024,
      "step": 5455
    },
    {
      "epoch": 0.18678534748373846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6918,
      "step": 5456
    },
    {
      "epoch": 0.18681958233481685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.698,
      "step": 5457
    },
    {
      "epoch": 0.18685381718589525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7269,
      "step": 5458
    },
    {
      "epoch": 0.18688805203697365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7083,
      "step": 5459
    },
    {
      "epoch": 0.18692228688805204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7226,
      "step": 5460
    },
    {
      "epoch": 0.18695652173913044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6903,
      "step": 5461
    },
    {
      "epoch": 0.18699075659020883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7338,
      "step": 5462
    },
    {
      "epoch": 0.18702499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.708,
      "step": 5463
    },
    {
      "epoch": 0.18705922629236563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6935,
      "step": 5464
    },
    {
      "epoch": 0.18709346114344402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.675,
      "step": 5465
    },
    {
      "epoch": 0.18712769599452242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.695,
      "step": 5466
    },
    {
      "epoch": 0.18716193084560082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6883,
      "step": 5467
    },
    {
      "epoch": 0.1871961656966792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7032,
      "step": 5468
    },
    {
      "epoch": 0.1872304005477576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6955,
      "step": 5469
    },
    {
      "epoch": 0.187264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6902,
      "step": 5470
    },
    {
      "epoch": 0.1872988702499144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6819,
      "step": 5471
    },
    {
      "epoch": 0.1873331051009928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7178,
      "step": 5472
    },
    {
      "epoch": 0.18736733995207122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6938,
      "step": 5473
    },
    {
      "epoch": 0.18740157480314962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.693,
      "step": 5474
    },
    {
      "epoch": 0.187435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7143,
      "step": 5475
    },
    {
      "epoch": 0.1874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7041,
      "step": 5476
    },
    {
      "epoch": 0.1875042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6902,
      "step": 5477
    },
    {
      "epoch": 0.1875385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7247,
      "step": 5478
    },
    {
      "epoch": 0.1875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6851,
      "step": 5479
    },
    {
      "epoch": 0.18760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7222,
      "step": 5480
    },
    {
      "epoch": 0.1876412187606984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7142,
      "step": 5481
    },
    {
      "epoch": 0.1876754536117768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6913,
      "step": 5482
    },
    {
      "epoch": 0.18770968846285518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7126,
      "step": 5483
    },
    {
      "epoch": 0.18774392331393358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7137,
      "step": 5484
    },
    {
      "epoch": 0.18777815816501198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.701,
      "step": 5485
    },
    {
      "epoch": 0.18781239301609037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6924,
      "step": 5486
    },
    {
      "epoch": 0.18784662786716877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6888,
      "step": 5487
    },
    {
      "epoch": 0.18788086271824717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6901,
      "step": 5488
    },
    {
      "epoch": 0.18791509756932556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6914,
      "step": 5489
    },
    {
      "epoch": 0.18794933242040396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7051,
      "step": 5490
    },
    {
      "epoch": 0.18798356727148238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7194,
      "step": 5491
    },
    {
      "epoch": 0.18801780212256078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6861,
      "step": 5492
    },
    {
      "epoch": 0.18805203697363918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6886,
      "step": 5493
    },
    {
      "epoch": 0.18808627182471757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7191,
      "step": 5494
    },
    {
      "epoch": 0.18812050667579597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6819,
      "step": 5495
    },
    {
      "epoch": 0.18815474152687436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6748,
      "step": 5496
    },
    {
      "epoch": 0.18818897637795276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7052,
      "step": 5497
    },
    {
      "epoch": 0.18822321122903116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7182,
      "step": 5498
    },
    {
      "epoch": 0.18825744608010955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6996,
      "step": 5499
    },
    {
      "epoch": 0.18829168093118795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7097,
      "step": 5500
    },
    {
      "epoch": 0.18832591578226635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7381,
      "step": 5501
    },
    {
      "epoch": 0.18836015063334474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7127,
      "step": 5502
    },
    {
      "epoch": 0.18839438548442314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6983,
      "step": 5503
    },
    {
      "epoch": 0.18842862033550153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7012,
      "step": 5504
    },
    {
      "epoch": 0.18846285518657993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6983,
      "step": 5505
    },
    {
      "epoch": 0.18849709003765833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7108,
      "step": 5506
    },
    {
      "epoch": 0.18853132488873672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6853,
      "step": 5507
    },
    {
      "epoch": 0.18856555973981512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6947,
      "step": 5508
    },
    {
      "epoch": 0.18859979459089352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.695,
      "step": 5509
    },
    {
      "epoch": 0.18863402944197194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7256,
      "step": 5510
    },
    {
      "epoch": 0.18866826429305034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6975,
      "step": 5511
    },
    {
      "epoch": 0.18870249914412873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7121,
      "step": 5512
    },
    {
      "epoch": 0.18873673399520713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6992,
      "step": 5513
    },
    {
      "epoch": 0.18877096884628553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6965,
      "step": 5514
    },
    {
      "epoch": 0.18880520369736392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6912,
      "step": 5515
    },
    {
      "epoch": 0.18883943854844232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6726,
      "step": 5516
    },
    {
      "epoch": 0.18887367339952071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.685,
      "step": 5517
    },
    {
      "epoch": 0.1889079082505991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6894,
      "step": 5518
    },
    {
      "epoch": 0.1889421431016775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7021,
      "step": 5519
    },
    {
      "epoch": 0.1889763779527559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.685,
      "step": 5520
    },
    {
      "epoch": 0.1890106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6858,
      "step": 5521
    },
    {
      "epoch": 0.1890448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6545,
      "step": 5522
    },
    {
      "epoch": 0.1890790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7045,
      "step": 5523
    },
    {
      "epoch": 0.1891133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7121,
      "step": 5524
    },
    {
      "epoch": 0.18914755220814788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6957,
      "step": 5525
    },
    {
      "epoch": 0.18918178705922628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7304,
      "step": 5526
    },
    {
      "epoch": 0.18921602191030468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.703,
      "step": 5527
    },
    {
      "epoch": 0.1892502567613831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7114,
      "step": 5528
    },
    {
      "epoch": 0.1892844916124615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7001,
      "step": 5529
    },
    {
      "epoch": 0.1893187264635399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6999,
      "step": 5530
    },
    {
      "epoch": 0.1893529613146183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6839,
      "step": 5531
    },
    {
      "epoch": 0.1893871961656967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7035,
      "step": 5532
    },
    {
      "epoch": 0.18942143101677508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6964,
      "step": 5533
    },
    {
      "epoch": 0.18945566586785348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6932,
      "step": 5534
    },
    {
      "epoch": 0.18948990071893188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6998,
      "step": 5535
    },
    {
      "epoch": 0.18952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6856,
      "step": 5536
    },
    {
      "epoch": 0.18955837042108867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6703,
      "step": 5537
    },
    {
      "epoch": 0.18959260527216706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.709,
      "step": 5538
    },
    {
      "epoch": 0.18962684012324546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.692,
      "step": 5539
    },
    {
      "epoch": 0.18966107497432386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6897,
      "step": 5540
    },
    {
      "epoch": 0.18969530982540225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.693,
      "step": 5541
    },
    {
      "epoch": 0.18972954467648065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.678,
      "step": 5542
    },
    {
      "epoch": 0.18976377952755905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6848,
      "step": 5543
    },
    {
      "epoch": 0.18979801437863744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6834,
      "step": 5544
    },
    {
      "epoch": 0.18983224922971584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7151,
      "step": 5545
    },
    {
      "epoch": 0.18986648408079426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7014,
      "step": 5546
    },
    {
      "epoch": 0.18990071893187266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6841,
      "step": 5547
    },
    {
      "epoch": 0.18993495378295105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.708,
      "step": 5548
    },
    {
      "epoch": 0.18996918863402945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7024,
      "step": 5549
    },
    {
      "epoch": 0.19000342348510785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7016,
      "step": 5550
    },
    {
      "epoch": 0.19003765833618624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7037,
      "step": 5551
    },
    {
      "epoch": 0.19007189318726464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7025,
      "step": 5552
    },
    {
      "epoch": 0.19010612803834304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6984,
      "step": 5553
    },
    {
      "epoch": 0.19014036288942143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6981,
      "step": 5554
    },
    {
      "epoch": 0.19017459774049983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6903,
      "step": 5555
    },
    {
      "epoch": 0.19020883259157823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6736,
      "step": 5556
    },
    {
      "epoch": 0.19024306744265662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7044,
      "step": 5557
    },
    {
      "epoch": 0.19027730229373502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6843,
      "step": 5558
    },
    {
      "epoch": 0.19031153714481341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6956,
      "step": 5559
    },
    {
      "epoch": 0.1903457719958918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6906,
      "step": 5560
    },
    {
      "epoch": 0.1903800068469702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6854,
      "step": 5561
    },
    {
      "epoch": 0.1904142416980486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6864,
      "step": 5562
    },
    {
      "epoch": 0.190448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7083,
      "step": 5563
    },
    {
      "epoch": 0.1904827114002054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6847,
      "step": 5564
    },
    {
      "epoch": 0.19051694625128382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7105,
      "step": 5565
    },
    {
      "epoch": 0.19055118110236222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6688,
      "step": 5566
    },
    {
      "epoch": 0.1905854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6845,
      "step": 5567
    },
    {
      "epoch": 0.190619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6747,
      "step": 5568
    },
    {
      "epoch": 0.1906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6894,
      "step": 5569
    },
    {
      "epoch": 0.1906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6757,
      "step": 5570
    },
    {
      "epoch": 0.1907223553577542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6853,
      "step": 5571
    },
    {
      "epoch": 0.1907565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7052,
      "step": 5572
    },
    {
      "epoch": 0.190790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7119,
      "step": 5573
    },
    {
      "epoch": 0.1908250599109894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6473,
      "step": 5574
    },
    {
      "epoch": 0.19085929476206778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6794,
      "step": 5575
    },
    {
      "epoch": 0.19089352961314618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6955,
      "step": 5576
    },
    {
      "epoch": 0.19092776446422458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6933,
      "step": 5577
    },
    {
      "epoch": 0.19096199931530297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6724,
      "step": 5578
    },
    {
      "epoch": 0.19099623416638137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6805,
      "step": 5579
    },
    {
      "epoch": 0.19103046901745976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6767,
      "step": 5580
    },
    {
      "epoch": 0.19106470386853816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6822,
      "step": 5581
    },
    {
      "epoch": 0.19109893871961656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7007,
      "step": 5582
    },
    {
      "epoch": 0.19113317357069498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6768,
      "step": 5583
    },
    {
      "epoch": 0.19116740842177338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6756,
      "step": 5584
    },
    {
      "epoch": 0.19120164327285177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6612,
      "step": 5585
    },
    {
      "epoch": 0.19123587812393017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7009,
      "step": 5586
    },
    {
      "epoch": 0.19127011297500857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6978,
      "step": 5587
    },
    {
      "epoch": 0.19130434782608696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7043,
      "step": 5588
    },
    {
      "epoch": 0.19133858267716536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6936,
      "step": 5589
    },
    {
      "epoch": 0.19137281752824375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6761,
      "step": 5590
    },
    {
      "epoch": 0.19140705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6866,
      "step": 5591
    },
    {
      "epoch": 0.19144128723040055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6838,
      "step": 5592
    },
    {
      "epoch": 0.19147552208147894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6771,
      "step": 5593
    },
    {
      "epoch": 0.19150975693255734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6987,
      "step": 5594
    },
    {
      "epoch": 0.19154399178363574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.698,
      "step": 5595
    },
    {
      "epoch": 0.19157822663471413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6637,
      "step": 5596
    },
    {
      "epoch": 0.19161246148579253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6755,
      "step": 5597
    },
    {
      "epoch": 0.19164669633687093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7154,
      "step": 5598
    },
    {
      "epoch": 0.19168093118794932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6977,
      "step": 5599
    },
    {
      "epoch": 0.19171516603902772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7118,
      "step": 5600
    },
    {
      "epoch": 0.19174940089010611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6914,
      "step": 5601
    },
    {
      "epoch": 0.19178363574118454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6971,
      "step": 5602
    },
    {
      "epoch": 0.19181787059226293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6917,
      "step": 5603
    },
    {
      "epoch": 0.19185210544334133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6911,
      "step": 5604
    },
    {
      "epoch": 0.19188634029441973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.685,
      "step": 5605
    },
    {
      "epoch": 0.19192057514549812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6975,
      "step": 5606
    },
    {
      "epoch": 0.19195480999657652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6737,
      "step": 5607
    },
    {
      "epoch": 0.19198904484765492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7209,
      "step": 5608
    },
    {
      "epoch": 0.1920232796987333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6756,
      "step": 5609
    },
    {
      "epoch": 0.1920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6771,
      "step": 5610
    },
    {
      "epoch": 0.1920917494008901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.689,
      "step": 5611
    },
    {
      "epoch": 0.1921259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.679,
      "step": 5612
    },
    {
      "epoch": 0.1921602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7128,
      "step": 5613
    },
    {
      "epoch": 0.1921944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6982,
      "step": 5614
    },
    {
      "epoch": 0.1922286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6939,
      "step": 5615
    },
    {
      "epoch": 0.1922629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6933,
      "step": 5616
    },
    {
      "epoch": 0.19229715850736048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7022,
      "step": 5617
    },
    {
      "epoch": 0.19233139335843888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7289,
      "step": 5618
    },
    {
      "epoch": 0.19236562820951728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6863,
      "step": 5619
    },
    {
      "epoch": 0.1923998630605957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6782,
      "step": 5620
    },
    {
      "epoch": 0.1924340979116741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6808,
      "step": 5621
    },
    {
      "epoch": 0.1924683327627525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6643,
      "step": 5622
    },
    {
      "epoch": 0.1925025676138309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6669,
      "step": 5623
    },
    {
      "epoch": 0.19253680246490928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6714,
      "step": 5624
    },
    {
      "epoch": 0.19257103731598768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.69,
      "step": 5625
    },
    {
      "epoch": 0.19260527216706608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7094,
      "step": 5626
    },
    {
      "epoch": 0.19263950701814447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6623,
      "step": 5627
    },
    {
      "epoch": 0.19267374186922287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7123,
      "step": 5628
    },
    {
      "epoch": 0.19270797672030127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6551,
      "step": 5629
    },
    {
      "epoch": 0.19274221157137966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6891,
      "step": 5630
    },
    {
      "epoch": 0.19277644642245806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6648,
      "step": 5631
    },
    {
      "epoch": 0.19281068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6811,
      "step": 5632
    },
    {
      "epoch": 0.19284491612461485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.659,
      "step": 5633
    },
    {
      "epoch": 0.19287915097569325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6844,
      "step": 5634
    },
    {
      "epoch": 0.19291338582677164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.677,
      "step": 5635
    },
    {
      "epoch": 0.19294762067785004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6734,
      "step": 5636
    },
    {
      "epoch": 0.19298185552892844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6992,
      "step": 5637
    },
    {
      "epoch": 0.19301609038000686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6725,
      "step": 5638
    },
    {
      "epoch": 0.19305032523108526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.679,
      "step": 5639
    },
    {
      "epoch": 0.19308456008216365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6844,
      "step": 5640
    },
    {
      "epoch": 0.19311879493324205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6922,
      "step": 5641
    },
    {
      "epoch": 0.19315302978432045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7006,
      "step": 5642
    },
    {
      "epoch": 0.19318726463539884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6786,
      "step": 5643
    },
    {
      "epoch": 0.19322149948647724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6991,
      "step": 5644
    },
    {
      "epoch": 0.19325573433755563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6887,
      "step": 5645
    },
    {
      "epoch": 0.19328996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7002,
      "step": 5646
    },
    {
      "epoch": 0.19332420403971243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6941,
      "step": 5647
    },
    {
      "epoch": 0.19335843889079082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7011,
      "step": 5648
    },
    {
      "epoch": 0.19339267374186922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6823,
      "step": 5649
    },
    {
      "epoch": 0.19342690859294762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7014,
      "step": 5650
    },
    {
      "epoch": 0.193461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6764,
      "step": 5651
    },
    {
      "epoch": 0.1934953782951044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6731,
      "step": 5652
    },
    {
      "epoch": 0.1935296131461828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6881,
      "step": 5653
    },
    {
      "epoch": 0.1935638479972612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6875,
      "step": 5654
    },
    {
      "epoch": 0.1935980828483396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.698,
      "step": 5655
    },
    {
      "epoch": 0.193632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6727,
      "step": 5656
    },
    {
      "epoch": 0.19366655255049642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6864,
      "step": 5657
    },
    {
      "epoch": 0.19370078740157481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6908,
      "step": 5658
    },
    {
      "epoch": 0.1937350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6773,
      "step": 5659
    },
    {
      "epoch": 0.1937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6999,
      "step": 5660
    },
    {
      "epoch": 0.19380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7056,
      "step": 5661
    },
    {
      "epoch": 0.1938377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6777,
      "step": 5662
    },
    {
      "epoch": 0.1938719616569668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6855,
      "step": 5663
    },
    {
      "epoch": 0.1939061965080452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6505,
      "step": 5664
    },
    {
      "epoch": 0.1939404313591236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6736,
      "step": 5665
    },
    {
      "epoch": 0.19397466621020198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6685,
      "step": 5666
    },
    {
      "epoch": 0.19400890106128038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6689,
      "step": 5667
    },
    {
      "epoch": 0.19404313591235878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6658,
      "step": 5668
    },
    {
      "epoch": 0.19407737076343717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6845,
      "step": 5669
    },
    {
      "epoch": 0.19411160561451557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7026,
      "step": 5670
    },
    {
      "epoch": 0.19414584046559397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6889,
      "step": 5671
    },
    {
      "epoch": 0.19418007531667236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.664,
      "step": 5672
    },
    {
      "epoch": 0.19421431016775076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6753,
      "step": 5673
    },
    {
      "epoch": 0.19424854501882916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6748,
      "step": 5674
    },
    {
      "epoch": 0.19428277986990758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6847,
      "step": 5675
    },
    {
      "epoch": 0.19431701472098598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6844,
      "step": 5676
    },
    {
      "epoch": 0.19435124957206437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6995,
      "step": 5677
    },
    {
      "epoch": 0.19438548442314277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6872,
      "step": 5678
    },
    {
      "epoch": 0.19441971927422116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6966,
      "step": 5679
    },
    {
      "epoch": 0.19445395412529956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6814,
      "step": 5680
    },
    {
      "epoch": 0.19448818897637796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6555,
      "step": 5681
    },
    {
      "epoch": 0.19452242382745635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6703,
      "step": 5682
    },
    {
      "epoch": 0.19455665867853475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6776,
      "step": 5683
    },
    {
      "epoch": 0.19459089352961315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.679,
      "step": 5684
    },
    {
      "epoch": 0.19462512838069154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6912,
      "step": 5685
    },
    {
      "epoch": 0.19465936323176994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6737,
      "step": 5686
    },
    {
      "epoch": 0.19469359808284833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6542,
      "step": 5687
    },
    {
      "epoch": 0.19472783293392673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6846,
      "step": 5688
    },
    {
      "epoch": 0.19476206778500513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6878,
      "step": 5689
    },
    {
      "epoch": 0.19479630263608352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6518,
      "step": 5690
    },
    {
      "epoch": 0.19483053748716192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.703,
      "step": 5691
    },
    {
      "epoch": 0.19486477233824032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6809,
      "step": 5692
    },
    {
      "epoch": 0.1948990071893187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6734,
      "step": 5693
    },
    {
      "epoch": 0.19493324204039714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.671,
      "step": 5694
    },
    {
      "epoch": 0.19496747689147553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6723,
      "step": 5695
    },
    {
      "epoch": 0.19500171174255393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6853,
      "step": 5696
    },
    {
      "epoch": 0.19503594659363233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7016,
      "step": 5697
    },
    {
      "epoch": 0.19507018144471072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6572,
      "step": 5698
    },
    {
      "epoch": 0.19510441629578912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6571,
      "step": 5699
    },
    {
      "epoch": 0.19513865114686751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6637,
      "step": 5700
    },
    {
      "epoch": 0.1951728859979459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6896,
      "step": 5701
    },
    {
      "epoch": 0.1952071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6842,
      "step": 5702
    },
    {
      "epoch": 0.1952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6589,
      "step": 5703
    },
    {
      "epoch": 0.1952755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.693,
      "step": 5704
    },
    {
      "epoch": 0.1953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6947,
      "step": 5705
    },
    {
      "epoch": 0.1953440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.675,
      "step": 5706
    },
    {
      "epoch": 0.1953782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6791,
      "step": 5707
    },
    {
      "epoch": 0.19541252995549468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6675,
      "step": 5708
    },
    {
      "epoch": 0.19544676480657308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6834,
      "step": 5709
    },
    {
      "epoch": 0.19548099965765148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6978,
      "step": 5710
    },
    {
      "epoch": 0.19551523450872987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6935,
      "step": 5711
    },
    {
      "epoch": 0.1955494693598083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.672,
      "step": 5712
    },
    {
      "epoch": 0.1955837042108867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6729,
      "step": 5713
    },
    {
      "epoch": 0.1956179390619651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6824,
      "step": 5714
    },
    {
      "epoch": 0.1956521739130435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6767,
      "step": 5715
    },
    {
      "epoch": 0.19568640876412188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6791,
      "step": 5716
    },
    {
      "epoch": 0.19572064361520028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7003,
      "step": 5717
    },
    {
      "epoch": 0.19575487846627868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6744,
      "step": 5718
    },
    {
      "epoch": 0.19578911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6748,
      "step": 5719
    },
    {
      "epoch": 0.19582334816843547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6841,
      "step": 5720
    },
    {
      "epoch": 0.19585758301951386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.66,
      "step": 5721
    },
    {
      "epoch": 0.19589181787059226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6402,
      "step": 5722
    },
    {
      "epoch": 0.19592605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6706,
      "step": 5723
    },
    {
      "epoch": 0.19596028757274905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6837,
      "step": 5724
    },
    {
      "epoch": 0.19599452242382745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6867,
      "step": 5725
    },
    {
      "epoch": 0.19602875727490585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6783,
      "step": 5726
    },
    {
      "epoch": 0.19606299212598424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7011,
      "step": 5727
    },
    {
      "epoch": 0.19609722697706264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6651,
      "step": 5728
    },
    {
      "epoch": 0.19613146182814103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6861,
      "step": 5729
    },
    {
      "epoch": 0.19616569667921946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6664,
      "step": 5730
    },
    {
      "epoch": 0.19619993153029786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6608,
      "step": 5731
    },
    {
      "epoch": 0.19623416638137625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6718,
      "step": 5732
    },
    {
      "epoch": 0.19626840123245465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6693,
      "step": 5733
    },
    {
      "epoch": 0.19630263608353304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6944,
      "step": 5734
    },
    {
      "epoch": 0.19633687093461144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6966,
      "step": 5735
    },
    {
      "epoch": 0.19637110578568984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6911,
      "step": 5736
    },
    {
      "epoch": 0.19640534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6671,
      "step": 5737
    },
    {
      "epoch": 0.19643957548784663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6912,
      "step": 5738
    },
    {
      "epoch": 0.19647381033892503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.681,
      "step": 5739
    },
    {
      "epoch": 0.19650804519000342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6758,
      "step": 5740
    },
    {
      "epoch": 0.19654228004108182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6769,
      "step": 5741
    },
    {
      "epoch": 0.19657651489216021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6738,
      "step": 5742
    },
    {
      "epoch": 0.1966107497432386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6461,
      "step": 5743
    },
    {
      "epoch": 0.196644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6521,
      "step": 5744
    },
    {
      "epoch": 0.1966792194453954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6627,
      "step": 5745
    },
    {
      "epoch": 0.1967134542964738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6609,
      "step": 5746
    },
    {
      "epoch": 0.1967476891475522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6684,
      "step": 5747
    },
    {
      "epoch": 0.1967819239986306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6754,
      "step": 5748
    },
    {
      "epoch": 0.19681615884970902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6687,
      "step": 5749
    },
    {
      "epoch": 0.1968503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6952,
      "step": 5750
    },
    {
      "epoch": 0.1968846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6748,
      "step": 5751
    },
    {
      "epoch": 0.1969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6851,
      "step": 5752
    },
    {
      "epoch": 0.1969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6797,
      "step": 5753
    },
    {
      "epoch": 0.196987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6748,
      "step": 5754
    },
    {
      "epoch": 0.1970215679561794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6691,
      "step": 5755
    },
    {
      "epoch": 0.1970558028072578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6834,
      "step": 5756
    },
    {
      "epoch": 0.1970900376583362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6793,
      "step": 5757
    },
    {
      "epoch": 0.19712427250941458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6596,
      "step": 5758
    },
    {
      "epoch": 0.19715850736049298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6509,
      "step": 5759
    },
    {
      "epoch": 0.19719274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6811,
      "step": 5760
    },
    {
      "epoch": 0.19722697706264977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6701,
      "step": 5761
    },
    {
      "epoch": 0.19726121191372817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.674,
      "step": 5762
    },
    {
      "epoch": 0.19729544676480656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6696,
      "step": 5763
    },
    {
      "epoch": 0.19732968161588496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6662,
      "step": 5764
    },
    {
      "epoch": 0.19736391646696336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6715,
      "step": 5765
    },
    {
      "epoch": 0.19739815131804175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6659,
      "step": 5766
    },
    {
      "epoch": 0.19743238616912018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6724,
      "step": 5767
    },
    {
      "epoch": 0.19746662102019857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6957,
      "step": 5768
    },
    {
      "epoch": 0.19750085587127697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6804,
      "step": 5769
    },
    {
      "epoch": 0.19753509072235537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6698,
      "step": 5770
    },
    {
      "epoch": 0.19756932557343376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6563,
      "step": 5771
    },
    {
      "epoch": 0.19760356042451216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6838,
      "step": 5772
    },
    {
      "epoch": 0.19763779527559056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6739,
      "step": 5773
    },
    {
      "epoch": 0.19767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6772,
      "step": 5774
    },
    {
      "epoch": 0.19770626497774735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6815,
      "step": 5775
    },
    {
      "epoch": 0.19774049982882574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6575,
      "step": 5776
    },
    {
      "epoch": 0.19777473467990414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6624,
      "step": 5777
    },
    {
      "epoch": 0.19780896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6792,
      "step": 5778
    },
    {
      "epoch": 0.19784320438206093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6839,
      "step": 5779
    },
    {
      "epoch": 0.19787743923313933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6667,
      "step": 5780
    },
    {
      "epoch": 0.19791167408421773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.673,
      "step": 5781
    },
    {
      "epoch": 0.19794590893529612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6656,
      "step": 5782
    },
    {
      "epoch": 0.19798014378637452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6573,
      "step": 5783
    },
    {
      "epoch": 0.19801437863745291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6928,
      "step": 5784
    },
    {
      "epoch": 0.19804861348853134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6709,
      "step": 5785
    },
    {
      "epoch": 0.19808284833960974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.668,
      "step": 5786
    },
    {
      "epoch": 0.19811708319068813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6837,
      "step": 5787
    },
    {
      "epoch": 0.19815131804176653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6647,
      "step": 5788
    },
    {
      "epoch": 0.19818555289284492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6876,
      "step": 5789
    },
    {
      "epoch": 0.19821978774392332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6579,
      "step": 5790
    },
    {
      "epoch": 0.19825402259500172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6669,
      "step": 5791
    },
    {
      "epoch": 0.1982882574460801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6664,
      "step": 5792
    },
    {
      "epoch": 0.1983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6482,
      "step": 5793
    },
    {
      "epoch": 0.1983567271482369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6792,
      "step": 5794
    },
    {
      "epoch": 0.1983909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6817,
      "step": 5795
    },
    {
      "epoch": 0.1984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6973,
      "step": 5796
    },
    {
      "epoch": 0.1984594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6777,
      "step": 5797
    },
    {
      "epoch": 0.1984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.685,
      "step": 5798
    },
    {
      "epoch": 0.1985279014036289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6581,
      "step": 5799
    },
    {
      "epoch": 0.19856213625470728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6708,
      "step": 5800
    },
    {
      "epoch": 0.19859637110578568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6754,
      "step": 5801
    },
    {
      "epoch": 0.19863060595686408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6618,
      "step": 5802
    },
    {
      "epoch": 0.19866484080794247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6836,
      "step": 5803
    },
    {
      "epoch": 0.1986990756590209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6743,
      "step": 5804
    },
    {
      "epoch": 0.1987333105100993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6538,
      "step": 5805
    },
    {
      "epoch": 0.1987675453611777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6853,
      "step": 5806
    },
    {
      "epoch": 0.19880178021225609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6685,
      "step": 5807
    },
    {
      "epoch": 0.19883601506333448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7042,
      "step": 5808
    },
    {
      "epoch": 0.19887024991441288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6663,
      "step": 5809
    },
    {
      "epoch": 0.19890448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6674,
      "step": 5810
    },
    {
      "epoch": 0.19893871961656967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6965,
      "step": 5811
    },
    {
      "epoch": 0.19897295446764807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6737,
      "step": 5812
    },
    {
      "epoch": 0.19900718931872646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6709,
      "step": 5813
    },
    {
      "epoch": 0.19904142416980486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6951,
      "step": 5814
    },
    {
      "epoch": 0.19907565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6598,
      "step": 5815
    },
    {
      "epoch": 0.19910989387196165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6624,
      "step": 5816
    },
    {
      "epoch": 0.19914412872304005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6724,
      "step": 5817
    },
    {
      "epoch": 0.19917836357411844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6673,
      "step": 5818
    },
    {
      "epoch": 0.19921259842519684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6784,
      "step": 5819
    },
    {
      "epoch": 0.19924683327627524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6541,
      "step": 5820
    },
    {
      "epoch": 0.19928106812735363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6853,
      "step": 5821
    },
    {
      "epoch": 0.19931530297843206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.68,
      "step": 5822
    },
    {
      "epoch": 0.19934953782951045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.654,
      "step": 5823
    },
    {
      "epoch": 0.19938377268058885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.659,
      "step": 5824
    },
    {
      "epoch": 0.19941800753166725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6696,
      "step": 5825
    },
    {
      "epoch": 0.19945224238274564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6692,
      "step": 5826
    },
    {
      "epoch": 0.19948647723382404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6466,
      "step": 5827
    },
    {
      "epoch": 0.19952071208490244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6661,
      "step": 5828
    },
    {
      "epoch": 0.19955494693598083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6645,
      "step": 5829
    },
    {
      "epoch": 0.19958918178705923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6625,
      "step": 5830
    },
    {
      "epoch": 0.19962341663813762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6585,
      "step": 5831
    },
    {
      "epoch": 0.19965765148921602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6717,
      "step": 5832
    },
    {
      "epoch": 0.19969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6436,
      "step": 5833
    },
    {
      "epoch": 0.1997261211913728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6701,
      "step": 5834
    },
    {
      "epoch": 0.1997603560424512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6537,
      "step": 5835
    },
    {
      "epoch": 0.1997945908935296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6775,
      "step": 5836
    },
    {
      "epoch": 0.199828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6477,
      "step": 5837
    },
    {
      "epoch": 0.1998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6534,
      "step": 5838
    },
    {
      "epoch": 0.1998972954467648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6606,
      "step": 5839
    },
    {
      "epoch": 0.1999315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.658,
      "step": 5840
    },
    {
      "epoch": 0.19996576514892161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6775,
      "step": 5841
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6469,
      "step": 5842
    },
    {
      "epoch": 0.2000342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6991,
      "step": 5843
    },
    {
      "epoch": 0.2000684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6791,
      "step": 5844
    },
    {
      "epoch": 0.2001027045532352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6669,
      "step": 5845
    },
    {
      "epoch": 0.2001369394043136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6703,
      "step": 5846
    },
    {
      "epoch": 0.200171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6856,
      "step": 5847
    },
    {
      "epoch": 0.2002054091064704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6559,
      "step": 5848
    },
    {
      "epoch": 0.20023964395754879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.674,
      "step": 5849
    },
    {
      "epoch": 0.20027387880862718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.676,
      "step": 5850
    },
    {
      "epoch": 0.20030811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6636,
      "step": 5851
    },
    {
      "epoch": 0.20034234851078397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6656,
      "step": 5852
    },
    {
      "epoch": 0.20037658336186237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6839,
      "step": 5853
    },
    {
      "epoch": 0.20041081821294077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6477,
      "step": 5854
    },
    {
      "epoch": 0.20044505306401916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.67,
      "step": 5855
    },
    {
      "epoch": 0.20047928791509756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6429,
      "step": 5856
    },
    {
      "epoch": 0.20051352276617596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6512,
      "step": 5857
    },
    {
      "epoch": 0.20054775761725435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6295,
      "step": 5858
    },
    {
      "epoch": 0.20058199246833278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6563,
      "step": 5859
    },
    {
      "epoch": 0.20061622731941117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6726,
      "step": 5860
    },
    {
      "epoch": 0.20065046217048957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6713,
      "step": 5861
    },
    {
      "epoch": 0.20068469702156796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6618,
      "step": 5862
    },
    {
      "epoch": 0.20071893187264636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6821,
      "step": 5863
    },
    {
      "epoch": 0.20075316672372476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6554,
      "step": 5864
    },
    {
      "epoch": 0.20078740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6629,
      "step": 5865
    },
    {
      "epoch": 0.20082163642588155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6441,
      "step": 5866
    },
    {
      "epoch": 0.20085587127695995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6722,
      "step": 5867
    },
    {
      "epoch": 0.20089010612803834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6604,
      "step": 5868
    },
    {
      "epoch": 0.20092434097911674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.672,
      "step": 5869
    },
    {
      "epoch": 0.20095857583019514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6524,
      "step": 5870
    },
    {
      "epoch": 0.20099281068127353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6539,
      "step": 5871
    },
    {
      "epoch": 0.20102704553235193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6648,
      "step": 5872
    },
    {
      "epoch": 0.20106128038343032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6424,
      "step": 5873
    },
    {
      "epoch": 0.20109551523450872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6688,
      "step": 5874
    },
    {
      "epoch": 0.20112975008558712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6532,
      "step": 5875
    },
    {
      "epoch": 0.2011639849366655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6817,
      "step": 5876
    },
    {
      "epoch": 0.20119821978774394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6468,
      "step": 5877
    },
    {
      "epoch": 0.20123245463882233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6637,
      "step": 5878
    },
    {
      "epoch": 0.20126668948990073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6654,
      "step": 5879
    },
    {
      "epoch": 0.20130092434097913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6335,
      "step": 5880
    },
    {
      "epoch": 0.20133515919205752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6805,
      "step": 5881
    },
    {
      "epoch": 0.20136939404313592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6821,
      "step": 5882
    },
    {
      "epoch": 0.20140362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6534,
      "step": 5883
    },
    {
      "epoch": 0.2014378637452927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6501,
      "step": 5884
    },
    {
      "epoch": 0.2014720985963711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6881,
      "step": 5885
    },
    {
      "epoch": 0.2015063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6577,
      "step": 5886
    },
    {
      "epoch": 0.2015405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6507,
      "step": 5887
    },
    {
      "epoch": 0.2015748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6436,
      "step": 5888
    },
    {
      "epoch": 0.2016090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.674,
      "step": 5889
    },
    {
      "epoch": 0.2016432728517631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6453,
      "step": 5890
    },
    {
      "epoch": 0.20167750770284149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6525,
      "step": 5891
    },
    {
      "epoch": 0.20171174255391988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6349,
      "step": 5892
    },
    {
      "epoch": 0.20174597740499828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6692,
      "step": 5893
    },
    {
      "epoch": 0.20178021225607667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6646,
      "step": 5894
    },
    {
      "epoch": 0.20181444710715507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6657,
      "step": 5895
    },
    {
      "epoch": 0.2018486819582335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6693,
      "step": 5896
    },
    {
      "epoch": 0.2018829168093119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6654,
      "step": 5897
    },
    {
      "epoch": 0.2019171516603903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6526,
      "step": 5898
    },
    {
      "epoch": 0.20195138651146868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6807,
      "step": 5899
    },
    {
      "epoch": 0.20198562136254708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6626,
      "step": 5900
    },
    {
      "epoch": 0.20201985621362548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6527,
      "step": 5901
    },
    {
      "epoch": 0.20205409106470387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6588,
      "step": 5902
    },
    {
      "epoch": 0.20208832591578227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6575,
      "step": 5903
    },
    {
      "epoch": 0.20212256076686066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6621,
      "step": 5904
    },
    {
      "epoch": 0.20215679561793906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6515,
      "step": 5905
    },
    {
      "epoch": 0.20219103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6597,
      "step": 5906
    },
    {
      "epoch": 0.20222526532009585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6716,
      "step": 5907
    },
    {
      "epoch": 0.20225950017117425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6588,
      "step": 5908
    },
    {
      "epoch": 0.20229373502225265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6609,
      "step": 5909
    },
    {
      "epoch": 0.20232796987333104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6648,
      "step": 5910
    },
    {
      "epoch": 0.20236220472440944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6657,
      "step": 5911
    },
    {
      "epoch": 0.20239643957548784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.642,
      "step": 5912
    },
    {
      "epoch": 0.20243067442656623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6604,
      "step": 5913
    },
    {
      "epoch": 0.20246490927764466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6567,
      "step": 5914
    },
    {
      "epoch": 0.20249914412872305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6573,
      "step": 5915
    },
    {
      "epoch": 0.20253337897980145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6764,
      "step": 5916
    },
    {
      "epoch": 0.20256761383087984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6684,
      "step": 5917
    },
    {
      "epoch": 0.20260184868195824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6563,
      "step": 5918
    },
    {
      "epoch": 0.20263608353303664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6729,
      "step": 5919
    },
    {
      "epoch": 0.20267031838411503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6638,
      "step": 5920
    },
    {
      "epoch": 0.20270455323519343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6453,
      "step": 5921
    },
    {
      "epoch": 0.20273878808627183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6574,
      "step": 5922
    },
    {
      "epoch": 0.20277302293735022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6633,
      "step": 5923
    },
    {
      "epoch": 0.20280725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6803,
      "step": 5924
    },
    {
      "epoch": 0.20284149263950702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.653,
      "step": 5925
    },
    {
      "epoch": 0.2028757274905854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6424,
      "step": 5926
    },
    {
      "epoch": 0.2029099623416638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6322,
      "step": 5927
    },
    {
      "epoch": 0.2029441971927422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6306,
      "step": 5928
    },
    {
      "epoch": 0.2029784320438206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.637,
      "step": 5929
    },
    {
      "epoch": 0.203012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6596,
      "step": 5930
    },
    {
      "epoch": 0.2030469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6516,
      "step": 5931
    },
    {
      "epoch": 0.2030811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6706,
      "step": 5932
    },
    {
      "epoch": 0.2031153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6744,
      "step": 5933
    },
    {
      "epoch": 0.2031496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6706,
      "step": 5934
    },
    {
      "epoch": 0.203183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6709,
      "step": 5935
    },
    {
      "epoch": 0.2032180760013694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6449,
      "step": 5936
    },
    {
      "epoch": 0.2032523108524478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6664,
      "step": 5937
    },
    {
      "epoch": 0.2032865457035262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6385,
      "step": 5938
    },
    {
      "epoch": 0.2033207805546046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6759,
      "step": 5939
    },
    {
      "epoch": 0.203355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6499,
      "step": 5940
    },
    {
      "epoch": 0.20338925025676138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6552,
      "step": 5941
    },
    {
      "epoch": 0.20342348510783978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6649,
      "step": 5942
    },
    {
      "epoch": 0.20345771995891818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.674,
      "step": 5943
    },
    {
      "epoch": 0.20349195480999657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6614,
      "step": 5944
    },
    {
      "epoch": 0.20352618966107497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6484,
      "step": 5945
    },
    {
      "epoch": 0.20356042451215337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6353,
      "step": 5946
    },
    {
      "epoch": 0.20359465936323176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6544,
      "step": 5947
    },
    {
      "epoch": 0.20362889421431016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6513,
      "step": 5948
    },
    {
      "epoch": 0.20366312906538855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6555,
      "step": 5949
    },
    {
      "epoch": 0.20369736391646695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6245,
      "step": 5950
    },
    {
      "epoch": 0.20373159876754537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.671,
      "step": 5951
    },
    {
      "epoch": 0.20376583361862377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.649,
      "step": 5952
    },
    {
      "epoch": 0.20380006846970217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6514,
      "step": 5953
    },
    {
      "epoch": 0.20383430332078056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6648,
      "step": 5954
    },
    {
      "epoch": 0.20386853817185896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6577,
      "step": 5955
    },
    {
      "epoch": 0.20390277302293736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6616,
      "step": 5956
    },
    {
      "epoch": 0.20393700787401575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.62,
      "step": 5957
    },
    {
      "epoch": 0.20397124272509415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6823,
      "step": 5958
    },
    {
      "epoch": 0.20400547757617254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6605,
      "step": 5959
    },
    {
      "epoch": 0.20403971242725094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.662,
      "step": 5960
    },
    {
      "epoch": 0.20407394727832934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6529,
      "step": 5961
    },
    {
      "epoch": 0.20410818212940773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6667,
      "step": 5962
    },
    {
      "epoch": 0.20414241698048613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6569,
      "step": 5963
    },
    {
      "epoch": 0.20417665183156453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.675,
      "step": 5964
    },
    {
      "epoch": 0.20421088668264292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.661,
      "step": 5965
    },
    {
      "epoch": 0.20424512153372132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6635,
      "step": 5966
    },
    {
      "epoch": 0.20427935638479972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6776,
      "step": 5967
    },
    {
      "epoch": 0.2043135912358781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6395,
      "step": 5968
    },
    {
      "epoch": 0.20434782608695654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6458,
      "step": 5969
    },
    {
      "epoch": 0.20438206093803493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6513,
      "step": 5970
    },
    {
      "epoch": 0.20441629578911333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6623,
      "step": 5971
    },
    {
      "epoch": 0.20445053064019172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.649,
      "step": 5972
    },
    {
      "epoch": 0.20448476549127012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6483,
      "step": 5973
    },
    {
      "epoch": 0.20451900034234852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6536,
      "step": 5974
    },
    {
      "epoch": 0.2045532351934269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6613,
      "step": 5975
    },
    {
      "epoch": 0.2045874700445053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6516,
      "step": 5976
    },
    {
      "epoch": 0.2046217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6596,
      "step": 5977
    },
    {
      "epoch": 0.2046559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6668,
      "step": 5978
    },
    {
      "epoch": 0.2046901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6225,
      "step": 5979
    },
    {
      "epoch": 0.2047244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6431,
      "step": 5980
    },
    {
      "epoch": 0.2047586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6478,
      "step": 5981
    },
    {
      "epoch": 0.2047928791509757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.671,
      "step": 5982
    },
    {
      "epoch": 0.20482711400205408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6425,
      "step": 5983
    },
    {
      "epoch": 0.20486134885313248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.649,
      "step": 5984
    },
    {
      "epoch": 0.20489558370421088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.638,
      "step": 5985
    },
    {
      "epoch": 0.20492981855528927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6519,
      "step": 5986
    },
    {
      "epoch": 0.20496405340636767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6309,
      "step": 5987
    },
    {
      "epoch": 0.2049982882574461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6508,
      "step": 5988
    },
    {
      "epoch": 0.2050325231085245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6187,
      "step": 5989
    },
    {
      "epoch": 0.20506675795960289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6493,
      "step": 5990
    },
    {
      "epoch": 0.20510099281068128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6659,
      "step": 5991
    },
    {
      "epoch": 0.20513522766175968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6512,
      "step": 5992
    },
    {
      "epoch": 0.20516946251283807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6255,
      "step": 5993
    },
    {
      "epoch": 0.20520369736391647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6546,
      "step": 5994
    },
    {
      "epoch": 0.20523793221499487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6454,
      "step": 5995
    },
    {
      "epoch": 0.20527216706607326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6449,
      "step": 5996
    },
    {
      "epoch": 0.20530640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6636,
      "step": 5997
    },
    {
      "epoch": 0.20534063676823006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6382,
      "step": 5998
    },
    {
      "epoch": 0.20537487161930845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6281,
      "step": 5999
    },
    {
      "epoch": 0.20540910647038685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6378,
      "step": 6000
    },
    {
      "epoch": 0.20544334132146524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6269,
      "step": 6001
    },
    {
      "epoch": 0.20547757617254364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6544,
      "step": 6002
    },
    {
      "epoch": 0.20551181102362204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6449,
      "step": 6003
    },
    {
      "epoch": 0.20554604587470043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6495,
      "step": 6004
    },
    {
      "epoch": 0.20558028072577883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6737,
      "step": 6005
    },
    {
      "epoch": 0.20561451557685725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6543,
      "step": 6006
    },
    {
      "epoch": 0.20564875042793565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6411,
      "step": 6007
    },
    {
      "epoch": 0.20568298527901405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6554,
      "step": 6008
    },
    {
      "epoch": 0.20571722013009244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6585,
      "step": 6009
    },
    {
      "epoch": 0.20575145498117084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6317,
      "step": 6010
    },
    {
      "epoch": 0.20578568983224924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6377,
      "step": 6011
    },
    {
      "epoch": 0.20581992468332763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6737,
      "step": 6012
    },
    {
      "epoch": 0.20585415953440603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.63,
      "step": 6013
    },
    {
      "epoch": 0.20588839438548442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6558,
      "step": 6014
    },
    {
      "epoch": 0.20592262923656282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6766,
      "step": 6015
    },
    {
      "epoch": 0.20595686408764122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6548,
      "step": 6016
    },
    {
      "epoch": 0.2059910989387196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6682,
      "step": 6017
    },
    {
      "epoch": 0.206025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6443,
      "step": 6018
    },
    {
      "epoch": 0.2060595686408764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6438,
      "step": 6019
    },
    {
      "epoch": 0.2060938034919548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6386,
      "step": 6020
    },
    {
      "epoch": 0.2061280383430332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6534,
      "step": 6021
    },
    {
      "epoch": 0.2061622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6615,
      "step": 6022
    },
    {
      "epoch": 0.20619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6454,
      "step": 6023
    },
    {
      "epoch": 0.2062307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6631,
      "step": 6024
    },
    {
      "epoch": 0.2062649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6425,
      "step": 6025
    },
    {
      "epoch": 0.2062992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6357,
      "step": 6026
    },
    {
      "epoch": 0.2063334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6281,
      "step": 6027
    },
    {
      "epoch": 0.206367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6547,
      "step": 6028
    },
    {
      "epoch": 0.2064019171516604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6525,
      "step": 6029
    },
    {
      "epoch": 0.2064361520027388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.663,
      "step": 6030
    },
    {
      "epoch": 0.2064703868538172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6345,
      "step": 6031
    },
    {
      "epoch": 0.20650462170489559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6369,
      "step": 6032
    },
    {
      "epoch": 0.20653885655597398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6605,
      "step": 6033
    },
    {
      "epoch": 0.20657309140705238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6723,
      "step": 6034
    },
    {
      "epoch": 0.20660732625813077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.672,
      "step": 6035
    },
    {
      "epoch": 0.20664156110920917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6793,
      "step": 6036
    },
    {
      "epoch": 0.20667579596028757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6267,
      "step": 6037
    },
    {
      "epoch": 0.20671003081136596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6951,
      "step": 6038
    },
    {
      "epoch": 0.20674426566244436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6514,
      "step": 6039
    },
    {
      "epoch": 0.20677850051352276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6122,
      "step": 6040
    },
    {
      "epoch": 0.20681273536460115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6452,
      "step": 6041
    },
    {
      "epoch": 0.20684697021567955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6269,
      "step": 6042
    },
    {
      "epoch": 0.20688120506675797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6557,
      "step": 6043
    },
    {
      "epoch": 0.20691543991783637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6346,
      "step": 6044
    },
    {
      "epoch": 0.20694967476891477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6262,
      "step": 6045
    },
    {
      "epoch": 0.20698390961999316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6571,
      "step": 6046
    },
    {
      "epoch": 0.20701814447107156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6483,
      "step": 6047
    },
    {
      "epoch": 0.20705237932214995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.652,
      "step": 6048
    },
    {
      "epoch": 0.20708661417322835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6464,
      "step": 6049
    },
    {
      "epoch": 0.20712084902430675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6624,
      "step": 6050
    },
    {
      "epoch": 0.20715508387538514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.653,
      "step": 6051
    },
    {
      "epoch": 0.20718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6352,
      "step": 6052
    },
    {
      "epoch": 0.20722355357754194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6411,
      "step": 6053
    },
    {
      "epoch": 0.20725778842862033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6587,
      "step": 6054
    },
    {
      "epoch": 0.20729202327969873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6402,
      "step": 6055
    },
    {
      "epoch": 0.20732625813077712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.64,
      "step": 6056
    },
    {
      "epoch": 0.20736049298185552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.628,
      "step": 6057
    },
    {
      "epoch": 0.20739472783293392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6631,
      "step": 6058
    },
    {
      "epoch": 0.2074289626840123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6574,
      "step": 6059
    },
    {
      "epoch": 0.2074631975350907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6644,
      "step": 6060
    },
    {
      "epoch": 0.20749743238616913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6327,
      "step": 6061
    },
    {
      "epoch": 0.20753166723724753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6535,
      "step": 6062
    },
    {
      "epoch": 0.20756590208832593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6367,
      "step": 6063
    },
    {
      "epoch": 0.20760013693940432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6379,
      "step": 6064
    },
    {
      "epoch": 0.20763437179048272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6627,
      "step": 6065
    },
    {
      "epoch": 0.20766860664156112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6396,
      "step": 6066
    },
    {
      "epoch": 0.2077028414926395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6394,
      "step": 6067
    },
    {
      "epoch": 0.2077370763437179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6546,
      "step": 6068
    },
    {
      "epoch": 0.2077713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6516,
      "step": 6069
    },
    {
      "epoch": 0.2078055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6505,
      "step": 6070
    },
    {
      "epoch": 0.2078397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6242,
      "step": 6071
    },
    {
      "epoch": 0.2078740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6369,
      "step": 6072
    },
    {
      "epoch": 0.2079082505991099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6522,
      "step": 6073
    },
    {
      "epoch": 0.20794248545018829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6464,
      "step": 6074
    },
    {
      "epoch": 0.20797672030126668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6488,
      "step": 6075
    },
    {
      "epoch": 0.20801095515234508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6354,
      "step": 6076
    },
    {
      "epoch": 0.20804519000342347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6266,
      "step": 6077
    },
    {
      "epoch": 0.20807942485450187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6348,
      "step": 6078
    },
    {
      "epoch": 0.20811365970558027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6284,
      "step": 6079
    },
    {
      "epoch": 0.2081478945566587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6571,
      "step": 6080
    },
    {
      "epoch": 0.2081821294077371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6396,
      "step": 6081
    },
    {
      "epoch": 0.20821636425881548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6524,
      "step": 6082
    },
    {
      "epoch": 0.20825059910989388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.641,
      "step": 6083
    },
    {
      "epoch": 0.20828483396097228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.646,
      "step": 6084
    },
    {
      "epoch": 0.20831906881205067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6411,
      "step": 6085
    },
    {
      "epoch": 0.20835330366312907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6443,
      "step": 6086
    },
    {
      "epoch": 0.20838753851420747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6257,
      "step": 6087
    },
    {
      "epoch": 0.20842177336528586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6291,
      "step": 6088
    },
    {
      "epoch": 0.20845600821636426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6471,
      "step": 6089
    },
    {
      "epoch": 0.20849024306744265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6255,
      "step": 6090
    },
    {
      "epoch": 0.20852447791852105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6239,
      "step": 6091
    },
    {
      "epoch": 0.20855871276959945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6377,
      "step": 6092
    },
    {
      "epoch": 0.20859294762067784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6444,
      "step": 6093
    },
    {
      "epoch": 0.20862718247175624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6473,
      "step": 6094
    },
    {
      "epoch": 0.20866141732283464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6266,
      "step": 6095
    },
    {
      "epoch": 0.20869565217391303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6334,
      "step": 6096
    },
    {
      "epoch": 0.20872988702499143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6302,
      "step": 6097
    },
    {
      "epoch": 0.20876412187606985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6289,
      "step": 6098
    },
    {
      "epoch": 0.20879835672714825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6499,
      "step": 6099
    },
    {
      "epoch": 0.20883259157822665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6446,
      "step": 6100
    },
    {
      "epoch": 0.20886682642930504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6549,
      "step": 6101
    },
    {
      "epoch": 0.20890106128038344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6235,
      "step": 6102
    },
    {
      "epoch": 0.20893529613146183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6515,
      "step": 6103
    },
    {
      "epoch": 0.20896953098254023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6358,
      "step": 6104
    },
    {
      "epoch": 0.20900376583361863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6638,
      "step": 6105
    },
    {
      "epoch": 0.20903800068469702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6296,
      "step": 6106
    },
    {
      "epoch": 0.20907223553577542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6381,
      "step": 6107
    },
    {
      "epoch": 0.20910647038685382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6264,
      "step": 6108
    },
    {
      "epoch": 0.2091407052379322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6372,
      "step": 6109
    },
    {
      "epoch": 0.2091749400890106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6431,
      "step": 6110
    },
    {
      "epoch": 0.209209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6512,
      "step": 6111
    },
    {
      "epoch": 0.2092434097911674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.672,
      "step": 6112
    },
    {
      "epoch": 0.2092776446422458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6349,
      "step": 6113
    },
    {
      "epoch": 0.2093118794933242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6574,
      "step": 6114
    },
    {
      "epoch": 0.2093461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6507,
      "step": 6115
    },
    {
      "epoch": 0.20938034919548099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.622,
      "step": 6116
    },
    {
      "epoch": 0.2094145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6424,
      "step": 6117
    },
    {
      "epoch": 0.2094488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6407,
      "step": 6118
    },
    {
      "epoch": 0.2094830537487162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6665,
      "step": 6119
    },
    {
      "epoch": 0.2095172885997946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6424,
      "step": 6120
    },
    {
      "epoch": 0.209551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6235,
      "step": 6121
    },
    {
      "epoch": 0.2095857583019514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6441,
      "step": 6122
    },
    {
      "epoch": 0.2096199931530298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6291,
      "step": 6123
    },
    {
      "epoch": 0.20965422800410818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6474,
      "step": 6124
    },
    {
      "epoch": 0.20968846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.651,
      "step": 6125
    },
    {
      "epoch": 0.20972269770626498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6384,
      "step": 6126
    },
    {
      "epoch": 0.20975693255734337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6399,
      "step": 6127
    },
    {
      "epoch": 0.20979116740842177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6449,
      "step": 6128
    },
    {
      "epoch": 0.20982540225950017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6388,
      "step": 6129
    },
    {
      "epoch": 0.20985963711057856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6535,
      "step": 6130
    },
    {
      "epoch": 0.20989387196165696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6284,
      "step": 6131
    },
    {
      "epoch": 0.20992810681273535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6207,
      "step": 6132
    },
    {
      "epoch": 0.20996234166381375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6203,
      "step": 6133
    },
    {
      "epoch": 0.20999657651489215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6275,
      "step": 6134
    },
    {
      "epoch": 0.21003081136597057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6472,
      "step": 6135
    },
    {
      "epoch": 0.21006504621704897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6518,
      "step": 6136
    },
    {
      "epoch": 0.21009928106812736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6516,
      "step": 6137
    },
    {
      "epoch": 0.21013351591920576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6374,
      "step": 6138
    },
    {
      "epoch": 0.21016775077028416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6417,
      "step": 6139
    },
    {
      "epoch": 0.21020198562136255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6358,
      "step": 6140
    },
    {
      "epoch": 0.21023622047244095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6294,
      "step": 6141
    },
    {
      "epoch": 0.21027045532351935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6301,
      "step": 6142
    },
    {
      "epoch": 0.21030469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6459,
      "step": 6143
    },
    {
      "epoch": 0.21033892502567614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6269,
      "step": 6144
    },
    {
      "epoch": 0.21037315987675453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6585,
      "step": 6145
    },
    {
      "epoch": 0.21040739472783293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6416,
      "step": 6146
    },
    {
      "epoch": 0.21044162957891133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.661,
      "step": 6147
    },
    {
      "epoch": 0.21047586442998972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6323,
      "step": 6148
    },
    {
      "epoch": 0.21051009928106812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6371,
      "step": 6149
    },
    {
      "epoch": 0.21054433413214652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6237,
      "step": 6150
    },
    {
      "epoch": 0.2105785689832249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6362,
      "step": 6151
    },
    {
      "epoch": 0.2106128038343033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6462,
      "step": 6152
    },
    {
      "epoch": 0.21064703868538173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6248,
      "step": 6153
    },
    {
      "epoch": 0.21068127353646013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6322,
      "step": 6154
    },
    {
      "epoch": 0.21071550838753852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6333,
      "step": 6155
    },
    {
      "epoch": 0.21074974323861692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6135,
      "step": 6156
    },
    {
      "epoch": 0.21078397808969532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6131,
      "step": 6157
    },
    {
      "epoch": 0.2108182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6374,
      "step": 6158
    },
    {
      "epoch": 0.2108524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.621,
      "step": 6159
    },
    {
      "epoch": 0.2108866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6322,
      "step": 6160
    },
    {
      "epoch": 0.2109209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6417,
      "step": 6161
    },
    {
      "epoch": 0.2109551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6309,
      "step": 6162
    },
    {
      "epoch": 0.2109893871961657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6394,
      "step": 6163
    },
    {
      "epoch": 0.2110236220472441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6245,
      "step": 6164
    },
    {
      "epoch": 0.2110578568983225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6213,
      "step": 6165
    },
    {
      "epoch": 0.21109209174940088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6309,
      "step": 6166
    },
    {
      "epoch": 0.21112632660047928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6388,
      "step": 6167
    },
    {
      "epoch": 0.21116056145155768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6295,
      "step": 6168
    },
    {
      "epoch": 0.21119479630263607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.621,
      "step": 6169
    },
    {
      "epoch": 0.21122903115371447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6237,
      "step": 6170
    },
    {
      "epoch": 0.21126326600479287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6324,
      "step": 6171
    },
    {
      "epoch": 0.2112975008558713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6297,
      "step": 6172
    },
    {
      "epoch": 0.21133173570694969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6198,
      "step": 6173
    },
    {
      "epoch": 0.21136597055802808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6326,
      "step": 6174
    },
    {
      "epoch": 0.21140020540910648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.639,
      "step": 6175
    },
    {
      "epoch": 0.21143444026018487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6359,
      "step": 6176
    },
    {
      "epoch": 0.21146867511126327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6312,
      "step": 6177
    },
    {
      "epoch": 0.21150290996234167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6203,
      "step": 6178
    },
    {
      "epoch": 0.21153714481342006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6272,
      "step": 6179
    },
    {
      "epoch": 0.21157137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.63,
      "step": 6180
    },
    {
      "epoch": 0.21160561451557686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6492,
      "step": 6181
    },
    {
      "epoch": 0.21163984936665525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6309,
      "step": 6182
    },
    {
      "epoch": 0.21167408421773365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.665,
      "step": 6183
    },
    {
      "epoch": 0.21170831906881205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6395,
      "step": 6184
    },
    {
      "epoch": 0.21174255391989044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.638,
      "step": 6185
    },
    {
      "epoch": 0.21177678877096884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6379,
      "step": 6186
    },
    {
      "epoch": 0.21181102362204723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6228,
      "step": 6187
    },
    {
      "epoch": 0.21184525847312563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6453,
      "step": 6188
    },
    {
      "epoch": 0.21187949332420403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6278,
      "step": 6189
    },
    {
      "epoch": 0.21191372817528245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.62,
      "step": 6190
    },
    {
      "epoch": 0.21194796302636085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6558,
      "step": 6191
    },
    {
      "epoch": 0.21198219787743924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6122,
      "step": 6192
    },
    {
      "epoch": 0.21201643272851764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6358,
      "step": 6193
    },
    {
      "epoch": 0.21205066757959604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6468,
      "step": 6194
    },
    {
      "epoch": 0.21208490243067443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6401,
      "step": 6195
    },
    {
      "epoch": 0.21211913728175283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6308,
      "step": 6196
    },
    {
      "epoch": 0.21215337213283122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6425,
      "step": 6197
    },
    {
      "epoch": 0.21218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6414,
      "step": 6198
    },
    {
      "epoch": 0.21222184183498802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6314,
      "step": 6199
    },
    {
      "epoch": 0.2122560766860664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6461,
      "step": 6200
    },
    {
      "epoch": 0.2122903115371448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6182,
      "step": 6201
    },
    {
      "epoch": 0.2123245463882232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6412,
      "step": 6202
    },
    {
      "epoch": 0.2123587812393016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6141,
      "step": 6203
    },
    {
      "epoch": 0.21239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.63,
      "step": 6204
    },
    {
      "epoch": 0.2124272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6402,
      "step": 6205
    },
    {
      "epoch": 0.2124614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6305,
      "step": 6206
    },
    {
      "epoch": 0.2124957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6612,
      "step": 6207
    },
    {
      "epoch": 0.21252995549469358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6449,
      "step": 6208
    },
    {
      "epoch": 0.212564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6336,
      "step": 6209
    },
    {
      "epoch": 0.2125984251968504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6254,
      "step": 6210
    },
    {
      "epoch": 0.2126326600479288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6362,
      "step": 6211
    },
    {
      "epoch": 0.2126668948990072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6347,
      "step": 6212
    },
    {
      "epoch": 0.2127011297500856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6341,
      "step": 6213
    },
    {
      "epoch": 0.212735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6487,
      "step": 6214
    },
    {
      "epoch": 0.2127695994522424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6374,
      "step": 6215
    },
    {
      "epoch": 0.21280383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6176,
      "step": 6216
    },
    {
      "epoch": 0.21283806915439918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6157,
      "step": 6217
    },
    {
      "epoch": 0.21287230400547758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.634,
      "step": 6218
    },
    {
      "epoch": 0.21290653885655597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6413,
      "step": 6219
    },
    {
      "epoch": 0.21294077370763437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6001,
      "step": 6220
    },
    {
      "epoch": 0.21297500855871276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.615,
      "step": 6221
    },
    {
      "epoch": 0.21300924340979116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6225,
      "step": 6222
    },
    {
      "epoch": 0.21304347826086956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.604,
      "step": 6223
    },
    {
      "epoch": 0.21307771311194795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6228,
      "step": 6224
    },
    {
      "epoch": 0.21311194796302635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6248,
      "step": 6225
    },
    {
      "epoch": 0.21314618281410475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6504,
      "step": 6226
    },
    {
      "epoch": 0.21318041766518317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6091,
      "step": 6227
    },
    {
      "epoch": 0.21321465251626157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6168,
      "step": 6228
    },
    {
      "epoch": 0.21324888736733996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6247,
      "step": 6229
    },
    {
      "epoch": 0.21328312221841836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6036,
      "step": 6230
    },
    {
      "epoch": 0.21331735706949675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6387,
      "step": 6231
    },
    {
      "epoch": 0.21335159192057515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6423,
      "step": 6232
    },
    {
      "epoch": 0.21338582677165355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6361,
      "step": 6233
    },
    {
      "epoch": 0.21342006162273194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6347,
      "step": 6234
    },
    {
      "epoch": 0.21345429647381034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.628,
      "step": 6235
    },
    {
      "epoch": 0.21348853132488874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6112,
      "step": 6236
    },
    {
      "epoch": 0.21352276617596713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6246,
      "step": 6237
    },
    {
      "epoch": 0.21355700102704553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6082,
      "step": 6238
    },
    {
      "epoch": 0.21359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6427,
      "step": 6239
    },
    {
      "epoch": 0.21362547072920232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6289,
      "step": 6240
    },
    {
      "epoch": 0.21365970558028072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6303,
      "step": 6241
    },
    {
      "epoch": 0.21369394043135911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6051,
      "step": 6242
    },
    {
      "epoch": 0.2137281752824375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6267,
      "step": 6243
    },
    {
      "epoch": 0.2137624101335159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6101,
      "step": 6244
    },
    {
      "epoch": 0.21379664498459433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6443,
      "step": 6245
    },
    {
      "epoch": 0.21383087983567273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6132,
      "step": 6246
    },
    {
      "epoch": 0.21386511468675112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6176,
      "step": 6247
    },
    {
      "epoch": 0.21389934953782952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6207,
      "step": 6248
    },
    {
      "epoch": 0.21393358438890792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6388,
      "step": 6249
    },
    {
      "epoch": 0.2139678192399863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6314,
      "step": 6250
    },
    {
      "epoch": 0.2140020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.644,
      "step": 6251
    },
    {
      "epoch": 0.2140362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6358,
      "step": 6252
    },
    {
      "epoch": 0.2140705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5958,
      "step": 6253
    },
    {
      "epoch": 0.2141047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6077,
      "step": 6254
    },
    {
      "epoch": 0.2141389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6225,
      "step": 6255
    },
    {
      "epoch": 0.2141732283464567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6191,
      "step": 6256
    },
    {
      "epoch": 0.2142074631975351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6243,
      "step": 6257
    },
    {
      "epoch": 0.21424169804861348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6286,
      "step": 6258
    },
    {
      "epoch": 0.21427593289969188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6442,
      "step": 6259
    },
    {
      "epoch": 0.21431016775077028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6319,
      "step": 6260
    },
    {
      "epoch": 0.21434440260184867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.625,
      "step": 6261
    },
    {
      "epoch": 0.21437863745292707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6524,
      "step": 6262
    },
    {
      "epoch": 0.21441287230400546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6261,
      "step": 6263
    },
    {
      "epoch": 0.2144471071550839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6438,
      "step": 6264
    },
    {
      "epoch": 0.21448134200616228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6302,
      "step": 6265
    },
    {
      "epoch": 0.21451557685724068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6217,
      "step": 6266
    },
    {
      "epoch": 0.21454981170831908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6262,
      "step": 6267
    },
    {
      "epoch": 0.21458404655939747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6415,
      "step": 6268
    },
    {
      "epoch": 0.21461828141047587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6132,
      "step": 6269
    },
    {
      "epoch": 0.21465251626155427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6146,
      "step": 6270
    },
    {
      "epoch": 0.21468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6145,
      "step": 6271
    },
    {
      "epoch": 0.21472098596371106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6421,
      "step": 6272
    },
    {
      "epoch": 0.21475522081478945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6311,
      "step": 6273
    },
    {
      "epoch": 0.21478945566586785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6417,
      "step": 6274
    },
    {
      "epoch": 0.21482369051694625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6229,
      "step": 6275
    },
    {
      "epoch": 0.21485792536802464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6169,
      "step": 6276
    },
    {
      "epoch": 0.21489216021910304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6213,
      "step": 6277
    },
    {
      "epoch": 0.21492639507018144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6475,
      "step": 6278
    },
    {
      "epoch": 0.21496062992125983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.621,
      "step": 6279
    },
    {
      "epoch": 0.21499486477233823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6374,
      "step": 6280
    },
    {
      "epoch": 0.21502909962341663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6204,
      "step": 6281
    },
    {
      "epoch": 0.21506333447449505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6182,
      "step": 6282
    },
    {
      "epoch": 0.21509756932557345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6394,
      "step": 6283
    },
    {
      "epoch": 0.21513180417665184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.642,
      "step": 6284
    },
    {
      "epoch": 0.21516603902773024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.619,
      "step": 6285
    },
    {
      "epoch": 0.21520027387880863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6235,
      "step": 6286
    },
    {
      "epoch": 0.21523450872988703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6358,
      "step": 6287
    },
    {
      "epoch": 0.21526874358096543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.625,
      "step": 6288
    },
    {
      "epoch": 0.21530297843204382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6579,
      "step": 6289
    },
    {
      "epoch": 0.21533721328312222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6403,
      "step": 6290
    },
    {
      "epoch": 0.21537144813420062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6369,
      "step": 6291
    },
    {
      "epoch": 0.215405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6459,
      "step": 6292
    },
    {
      "epoch": 0.2154399178363574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6178,
      "step": 6293
    },
    {
      "epoch": 0.2154741526874358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6078,
      "step": 6294
    },
    {
      "epoch": 0.2155083875385142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6341,
      "step": 6295
    },
    {
      "epoch": 0.2155426223895926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6653,
      "step": 6296
    },
    {
      "epoch": 0.215576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6255,
      "step": 6297
    },
    {
      "epoch": 0.2156110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6179,
      "step": 6298
    },
    {
      "epoch": 0.2156453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6295,
      "step": 6299
    },
    {
      "epoch": 0.21567956179390618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.624,
      "step": 6300
    },
    {
      "epoch": 0.2157137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6117,
      "step": 6301
    },
    {
      "epoch": 0.215748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6234,
      "step": 6302
    },
    {
      "epoch": 0.2157822663471414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6103,
      "step": 6303
    },
    {
      "epoch": 0.2158165011982198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6134,
      "step": 6304
    },
    {
      "epoch": 0.2158507360492982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6112,
      "step": 6305
    },
    {
      "epoch": 0.2158849709003766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6231,
      "step": 6306
    },
    {
      "epoch": 0.21591920575145498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6281,
      "step": 6307
    },
    {
      "epoch": 0.21595344060253338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6263,
      "step": 6308
    },
    {
      "epoch": 0.21598767545361178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.645,
      "step": 6309
    },
    {
      "epoch": 0.21602191030469017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6258,
      "step": 6310
    },
    {
      "epoch": 0.21605614515576857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6207,
      "step": 6311
    },
    {
      "epoch": 0.21609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6221,
      "step": 6312
    },
    {
      "epoch": 0.21612461485792536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6015,
      "step": 6313
    },
    {
      "epoch": 0.21615884970900376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.616,
      "step": 6314
    },
    {
      "epoch": 0.21619308456008215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6049,
      "step": 6315
    },
    {
      "epoch": 0.21622731941116055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6381,
      "step": 6316
    },
    {
      "epoch": 0.21626155426223895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6238,
      "step": 6317
    },
    {
      "epoch": 0.21629578911331734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6225,
      "step": 6318
    },
    {
      "epoch": 0.21633002396439577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6341,
      "step": 6319
    },
    {
      "epoch": 0.21636425881547416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6286,
      "step": 6320
    },
    {
      "epoch": 0.21639849366655256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6171,
      "step": 6321
    },
    {
      "epoch": 0.21643272851763096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6406,
      "step": 6322
    },
    {
      "epoch": 0.21646696336870935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6276,
      "step": 6323
    },
    {
      "epoch": 0.21650119821978775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6162,
      "step": 6324
    },
    {
      "epoch": 0.21653543307086615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6343,
      "step": 6325
    },
    {
      "epoch": 0.21656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6166,
      "step": 6326
    },
    {
      "epoch": 0.21660390277302294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6273,
      "step": 6327
    },
    {
      "epoch": 0.21663813762410133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6034,
      "step": 6328
    },
    {
      "epoch": 0.21667237247517973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6375,
      "step": 6329
    },
    {
      "epoch": 0.21670660732625813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6275,
      "step": 6330
    },
    {
      "epoch": 0.21674084217733652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6237,
      "step": 6331
    },
    {
      "epoch": 0.21677507702841492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.611,
      "step": 6332
    },
    {
      "epoch": 0.21680931187949332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6099,
      "step": 6333
    },
    {
      "epoch": 0.2168435467305717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6229,
      "step": 6334
    },
    {
      "epoch": 0.2168777815816501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6165,
      "step": 6335
    },
    {
      "epoch": 0.2169120164327285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5985,
      "step": 6336
    },
    {
      "epoch": 0.21694625128380693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6282,
      "step": 6337
    },
    {
      "epoch": 0.21698048613488533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6192,
      "step": 6338
    },
    {
      "epoch": 0.21701472098596372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6166,
      "step": 6339
    },
    {
      "epoch": 0.21704895583704212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6037,
      "step": 6340
    },
    {
      "epoch": 0.21708319068812051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6222,
      "step": 6341
    },
    {
      "epoch": 0.2171174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6054,
      "step": 6342
    },
    {
      "epoch": 0.2171516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6295,
      "step": 6343
    },
    {
      "epoch": 0.2171858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.612,
      "step": 6344
    },
    {
      "epoch": 0.2172201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6161,
      "step": 6345
    },
    {
      "epoch": 0.2172543649435125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6297,
      "step": 6346
    },
    {
      "epoch": 0.2172885997945909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6288,
      "step": 6347
    },
    {
      "epoch": 0.2173228346456693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6073,
      "step": 6348
    },
    {
      "epoch": 0.21735706949674768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6067,
      "step": 6349
    },
    {
      "epoch": 0.21739130434782608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6087,
      "step": 6350
    },
    {
      "epoch": 0.21742553919890448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6231,
      "step": 6351
    },
    {
      "epoch": 0.21745977404998287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6178,
      "step": 6352
    },
    {
      "epoch": 0.21749400890106127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6079,
      "step": 6353
    },
    {
      "epoch": 0.21752824375213967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6057,
      "step": 6354
    },
    {
      "epoch": 0.21756247860321806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6417,
      "step": 6355
    },
    {
      "epoch": 0.2175967134542965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6134,
      "step": 6356
    },
    {
      "epoch": 0.21763094830537488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6385,
      "step": 6357
    },
    {
      "epoch": 0.21766518315645328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6409,
      "step": 6358
    },
    {
      "epoch": 0.21769941800753168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6167,
      "step": 6359
    },
    {
      "epoch": 0.21773365285861007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6013,
      "step": 6360
    },
    {
      "epoch": 0.21776788770968847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6146,
      "step": 6361
    },
    {
      "epoch": 0.21780212256076686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6318,
      "step": 6362
    },
    {
      "epoch": 0.21783635741184526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.627,
      "step": 6363
    },
    {
      "epoch": 0.21787059226292366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6252,
      "step": 6364
    },
    {
      "epoch": 0.21790482711400205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6207,
      "step": 6365
    },
    {
      "epoch": 0.21793906196508045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6203,
      "step": 6366
    },
    {
      "epoch": 0.21797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5971,
      "step": 6367
    },
    {
      "epoch": 0.21800753166723724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6317,
      "step": 6368
    },
    {
      "epoch": 0.21804176651831564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6105,
      "step": 6369
    },
    {
      "epoch": 0.21807600136939403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6342,
      "step": 6370
    },
    {
      "epoch": 0.21811023622047243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5985,
      "step": 6371
    },
    {
      "epoch": 0.21814447107155083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6056,
      "step": 6372
    },
    {
      "epoch": 0.21817870592262922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6141,
      "step": 6373
    },
    {
      "epoch": 0.21821294077370765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.618,
      "step": 6374
    },
    {
      "epoch": 0.21824717562478604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6212,
      "step": 6375
    },
    {
      "epoch": 0.21828141047586444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6363,
      "step": 6376
    },
    {
      "epoch": 0.21831564532694284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6051,
      "step": 6377
    },
    {
      "epoch": 0.21834988017802123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6103,
      "step": 6378
    },
    {
      "epoch": 0.21838411502909963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6189,
      "step": 6379
    },
    {
      "epoch": 0.21841834988017803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.631,
      "step": 6380
    },
    {
      "epoch": 0.21845258473125642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6267,
      "step": 6381
    },
    {
      "epoch": 0.21848681958233482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6247,
      "step": 6382
    },
    {
      "epoch": 0.21852105443341321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.62,
      "step": 6383
    },
    {
      "epoch": 0.2185552892844916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6117,
      "step": 6384
    },
    {
      "epoch": 0.21858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.618,
      "step": 6385
    },
    {
      "epoch": 0.2186237589866484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6066,
      "step": 6386
    },
    {
      "epoch": 0.2186579938377268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6283,
      "step": 6387
    },
    {
      "epoch": 0.2186922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6211,
      "step": 6388
    },
    {
      "epoch": 0.2187264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.586,
      "step": 6389
    },
    {
      "epoch": 0.218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6142,
      "step": 6390
    },
    {
      "epoch": 0.21879493324204038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6181,
      "step": 6391
    },
    {
      "epoch": 0.2188291680931188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6088,
      "step": 6392
    },
    {
      "epoch": 0.2188634029441972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5761,
      "step": 6393
    },
    {
      "epoch": 0.2188976377952756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6159,
      "step": 6394
    },
    {
      "epoch": 0.218931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6212,
      "step": 6395
    },
    {
      "epoch": 0.2189661074974324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6223,
      "step": 6396
    },
    {
      "epoch": 0.2190003423485108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6106,
      "step": 6397
    },
    {
      "epoch": 0.2190345771995892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6085,
      "step": 6398
    },
    {
      "epoch": 0.21906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6142,
      "step": 6399
    },
    {
      "epoch": 0.21910304690174598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5853,
      "step": 6400
    },
    {
      "epoch": 0.21913728175282438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6228,
      "step": 6401
    },
    {
      "epoch": 0.21917151660390277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5935,
      "step": 6402
    },
    {
      "epoch": 0.21920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.62,
      "step": 6403
    },
    {
      "epoch": 0.21923998630605956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5911,
      "step": 6404
    },
    {
      "epoch": 0.21927422115713796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.62,
      "step": 6405
    },
    {
      "epoch": 0.21930845600821636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6195,
      "step": 6406
    },
    {
      "epoch": 0.21934269085929475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6274,
      "step": 6407
    },
    {
      "epoch": 0.21937692571037315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5964,
      "step": 6408
    },
    {
      "epoch": 0.21941116056145155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5994,
      "step": 6409
    },
    {
      "epoch": 0.21944539541252994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6059,
      "step": 6410
    },
    {
      "epoch": 0.21947963026360837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.602,
      "step": 6411
    },
    {
      "epoch": 0.21951386511468676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6061,
      "step": 6412
    },
    {
      "epoch": 0.21954809996576516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6055,
      "step": 6413
    },
    {
      "epoch": 0.21958233481684356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6283,
      "step": 6414
    },
    {
      "epoch": 0.21961656966792195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6326,
      "step": 6415
    },
    {
      "epoch": 0.21965080451900035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6168,
      "step": 6416
    },
    {
      "epoch": 0.21968503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5882,
      "step": 6417
    },
    {
      "epoch": 0.21971927422115714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6054,
      "step": 6418
    },
    {
      "epoch": 0.21975350907223554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6128,
      "step": 6419
    },
    {
      "epoch": 0.21978774392331393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6099,
      "step": 6420
    },
    {
      "epoch": 0.21982197877439233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5846,
      "step": 6421
    },
    {
      "epoch": 0.21985621362547073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6002,
      "step": 6422
    },
    {
      "epoch": 0.21989044847654912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6276,
      "step": 6423
    },
    {
      "epoch": 0.21992468332762752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6019,
      "step": 6424
    },
    {
      "epoch": 0.21995891817870591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.599,
      "step": 6425
    },
    {
      "epoch": 0.2199931530297843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6168,
      "step": 6426
    },
    {
      "epoch": 0.2200273878808627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6134,
      "step": 6427
    },
    {
      "epoch": 0.2200616227319411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6278,
      "step": 6428
    },
    {
      "epoch": 0.22009585758301953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6091,
      "step": 6429
    },
    {
      "epoch": 0.22013009243409792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6239,
      "step": 6430
    },
    {
      "epoch": 0.22016432728517632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6143,
      "step": 6431
    },
    {
      "epoch": 0.22019856213625472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6063,
      "step": 6432
    },
    {
      "epoch": 0.2202327969873331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6335,
      "step": 6433
    },
    {
      "epoch": 0.2202670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6189,
      "step": 6434
    },
    {
      "epoch": 0.2203012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6146,
      "step": 6435
    },
    {
      "epoch": 0.2203355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6098,
      "step": 6436
    },
    {
      "epoch": 0.2203697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6116,
      "step": 6437
    },
    {
      "epoch": 0.2204039712427251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6037,
      "step": 6438
    },
    {
      "epoch": 0.2204382060938035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6182,
      "step": 6439
    },
    {
      "epoch": 0.2204724409448819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6103,
      "step": 6440
    },
    {
      "epoch": 0.22050667579596028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6013,
      "step": 6441
    },
    {
      "epoch": 0.22054091064703868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5886,
      "step": 6442
    },
    {
      "epoch": 0.22057514549811708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6173,
      "step": 6443
    },
    {
      "epoch": 0.22060938034919547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6013,
      "step": 6444
    },
    {
      "epoch": 0.22064361520027387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6093,
      "step": 6445
    },
    {
      "epoch": 0.22067785005135226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6132,
      "step": 6446
    },
    {
      "epoch": 0.22071208490243066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.626,
      "step": 6447
    },
    {
      "epoch": 0.22074631975350908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5946,
      "step": 6448
    },
    {
      "epoch": 0.22078055460458748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6227,
      "step": 6449
    },
    {
      "epoch": 0.22081478945566588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6264,
      "step": 6450
    },
    {
      "epoch": 0.22084902430674427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6176,
      "step": 6451
    },
    {
      "epoch": 0.22088325915782267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6126,
      "step": 6452
    },
    {
      "epoch": 0.22091749400890107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6146,
      "step": 6453
    },
    {
      "epoch": 0.22095172885997946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.619,
      "step": 6454
    },
    {
      "epoch": 0.22098596371105786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6039,
      "step": 6455
    },
    {
      "epoch": 0.22102019856213626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6133,
      "step": 6456
    },
    {
      "epoch": 0.22105443341321465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6094,
      "step": 6457
    },
    {
      "epoch": 0.22108866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.602,
      "step": 6458
    },
    {
      "epoch": 0.22112290311537144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6065,
      "step": 6459
    },
    {
      "epoch": 0.22115713796644984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6241,
      "step": 6460
    },
    {
      "epoch": 0.22119137281752824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6163,
      "step": 6461
    },
    {
      "epoch": 0.22122560766860663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.609,
      "step": 6462
    },
    {
      "epoch": 0.22125984251968503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5964,
      "step": 6463
    },
    {
      "epoch": 0.22129407737076343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6191,
      "step": 6464
    },
    {
      "epoch": 0.22132831222184182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6166,
      "step": 6465
    },
    {
      "epoch": 0.22136254707292025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6167,
      "step": 6466
    },
    {
      "epoch": 0.22139678192399864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6163,
      "step": 6467
    },
    {
      "epoch": 0.22143101677507704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6406,
      "step": 6468
    },
    {
      "epoch": 0.22146525162615543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6051,
      "step": 6469
    },
    {
      "epoch": 0.22149948647723383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6124,
      "step": 6470
    },
    {
      "epoch": 0.22153372132831223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.593,
      "step": 6471
    },
    {
      "epoch": 0.22156795617939062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6107,
      "step": 6472
    },
    {
      "epoch": 0.22160219103046902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6052,
      "step": 6473
    },
    {
      "epoch": 0.22163642588154742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6051,
      "step": 6474
    },
    {
      "epoch": 0.2216706607326258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.611,
      "step": 6475
    },
    {
      "epoch": 0.2217048955837042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6127,
      "step": 6476
    },
    {
      "epoch": 0.2217391304347826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6262,
      "step": 6477
    },
    {
      "epoch": 0.221773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5994,
      "step": 6478
    },
    {
      "epoch": 0.2218076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6085,
      "step": 6479
    },
    {
      "epoch": 0.2218418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5982,
      "step": 6480
    },
    {
      "epoch": 0.2218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6004,
      "step": 6481
    },
    {
      "epoch": 0.2219103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6027,
      "step": 6482
    },
    {
      "epoch": 0.22194453954125298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5905,
      "step": 6483
    },
    {
      "epoch": 0.2219787743923314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5977,
      "step": 6484
    },
    {
      "epoch": 0.2220130092434098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6071,
      "step": 6485
    },
    {
      "epoch": 0.2220472440944882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6215,
      "step": 6486
    },
    {
      "epoch": 0.2220814789455666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6059,
      "step": 6487
    },
    {
      "epoch": 0.222115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5987,
      "step": 6488
    },
    {
      "epoch": 0.2221499486477234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6177,
      "step": 6489
    },
    {
      "epoch": 0.22218418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6037,
      "step": 6490
    },
    {
      "epoch": 0.22221841834988018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5929,
      "step": 6491
    },
    {
      "epoch": 0.22225265320095858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6099,
      "step": 6492
    },
    {
      "epoch": 0.22228688805203697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5791,
      "step": 6493
    },
    {
      "epoch": 0.22232112290311537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.629,
      "step": 6494
    },
    {
      "epoch": 0.22235535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6107,
      "step": 6495
    },
    {
      "epoch": 0.22238959260527216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6087,
      "step": 6496
    },
    {
      "epoch": 0.22242382745635056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6217,
      "step": 6497
    },
    {
      "epoch": 0.22245806230742896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6386,
      "step": 6498
    },
    {
      "epoch": 0.22249229715850735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6167,
      "step": 6499
    },
    {
      "epoch": 0.22252653200958575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6278,
      "step": 6500
    },
    {
      "epoch": 0.22256076686066414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6055,
      "step": 6501
    },
    {
      "epoch": 0.22259500171174254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6372,
      "step": 6502
    },
    {
      "epoch": 0.22262923656282096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6146,
      "step": 6503
    },
    {
      "epoch": 0.22266347141389936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6117,
      "step": 6504
    },
    {
      "epoch": 0.22269770626497776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6002,
      "step": 6505
    },
    {
      "epoch": 0.22273194111605615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6238,
      "step": 6506
    },
    {
      "epoch": 0.22276617596713455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6131,
      "step": 6507
    },
    {
      "epoch": 0.22280041081821295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6214,
      "step": 6508
    },
    {
      "epoch": 0.22283464566929134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5982,
      "step": 6509
    },
    {
      "epoch": 0.22286888052036974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6146,
      "step": 6510
    },
    {
      "epoch": 0.22290311537144814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5942,
      "step": 6511
    },
    {
      "epoch": 0.22293735022252653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6151,
      "step": 6512
    },
    {
      "epoch": 0.22297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6069,
      "step": 6513
    },
    {
      "epoch": 0.22300581992468332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6075,
      "step": 6514
    },
    {
      "epoch": 0.22304005477576172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5992,
      "step": 6515
    },
    {
      "epoch": 0.22307428962684012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6091,
      "step": 6516
    },
    {
      "epoch": 0.2231085244779185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6148,
      "step": 6517
    },
    {
      "epoch": 0.2231427593289969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6141,
      "step": 6518
    },
    {
      "epoch": 0.2231769941800753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6068,
      "step": 6519
    },
    {
      "epoch": 0.2232112290311537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.615,
      "step": 6520
    },
    {
      "epoch": 0.22324546388223213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6233,
      "step": 6521
    },
    {
      "epoch": 0.22327969873331052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6104,
      "step": 6522
    },
    {
      "epoch": 0.22331393358438892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5889,
      "step": 6523
    },
    {
      "epoch": 0.22334816843546731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5988,
      "step": 6524
    },
    {
      "epoch": 0.2233824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6091,
      "step": 6525
    },
    {
      "epoch": 0.2234166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.599,
      "step": 6526
    },
    {
      "epoch": 0.2234508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6096,
      "step": 6527
    },
    {
      "epoch": 0.2234851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.622,
      "step": 6528
    },
    {
      "epoch": 0.2235193426908593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6275,
      "step": 6529
    },
    {
      "epoch": 0.2235535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6089,
      "step": 6530
    },
    {
      "epoch": 0.2235878123930161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6028,
      "step": 6531
    },
    {
      "epoch": 0.22362204724409449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6045,
      "step": 6532
    },
    {
      "epoch": 0.22365628209517288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6133,
      "step": 6533
    },
    {
      "epoch": 0.22369051694625128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6186,
      "step": 6534
    },
    {
      "epoch": 0.22372475179732967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5794,
      "step": 6535
    },
    {
      "epoch": 0.22375898664840807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6073,
      "step": 6536
    },
    {
      "epoch": 0.22379322149948647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6294,
      "step": 6537
    },
    {
      "epoch": 0.22382745635056486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.612,
      "step": 6538
    },
    {
      "epoch": 0.22386169120164326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6013,
      "step": 6539
    },
    {
      "epoch": 0.22389592605272168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6085,
      "step": 6540
    },
    {
      "epoch": 0.22393016090380008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6193,
      "step": 6541
    },
    {
      "epoch": 0.22396439575487848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.598,
      "step": 6542
    },
    {
      "epoch": 0.22399863060595687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6297,
      "step": 6543
    },
    {
      "epoch": 0.22403286545703527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6121,
      "step": 6544
    },
    {
      "epoch": 0.22406710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6026,
      "step": 6545
    },
    {
      "epoch": 0.22410133515919206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6123,
      "step": 6546
    },
    {
      "epoch": 0.22413557001027046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5843,
      "step": 6547
    },
    {
      "epoch": 0.22416980486134885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6025,
      "step": 6548
    },
    {
      "epoch": 0.22420403971242725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6156,
      "step": 6549
    },
    {
      "epoch": 0.22423827456350565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6317,
      "step": 6550
    },
    {
      "epoch": 0.22427250941458404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6021,
      "step": 6551
    },
    {
      "epoch": 0.22430674426566244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6217,
      "step": 6552
    },
    {
      "epoch": 0.22434097911674084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6236,
      "step": 6553
    },
    {
      "epoch": 0.22437521396781923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6099,
      "step": 6554
    },
    {
      "epoch": 0.22440944881889763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5858,
      "step": 6555
    },
    {
      "epoch": 0.22444368366997602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6071,
      "step": 6556
    },
    {
      "epoch": 0.22447791852105442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5916,
      "step": 6557
    },
    {
      "epoch": 0.22451215337213284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5877,
      "step": 6558
    },
    {
      "epoch": 0.22454638822321124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5978,
      "step": 6559
    },
    {
      "epoch": 0.22458062307428964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5819,
      "step": 6560
    },
    {
      "epoch": 0.22461485792536803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6048,
      "step": 6561
    },
    {
      "epoch": 0.22464909277644643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5842,
      "step": 6562
    },
    {
      "epoch": 0.22468332762752483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6101,
      "step": 6563
    },
    {
      "epoch": 0.22471756247860322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6026,
      "step": 6564
    },
    {
      "epoch": 0.22475179732968162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5841,
      "step": 6565
    },
    {
      "epoch": 0.22478603218076001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.598,
      "step": 6566
    },
    {
      "epoch": 0.2248202670318384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5996,
      "step": 6567
    },
    {
      "epoch": 0.2248545018829168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6024,
      "step": 6568
    },
    {
      "epoch": 0.2248887367339952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5875,
      "step": 6569
    },
    {
      "epoch": 0.2249229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5974,
      "step": 6570
    },
    {
      "epoch": 0.224957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6104,
      "step": 6571
    },
    {
      "epoch": 0.2249914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6028,
      "step": 6572
    },
    {
      "epoch": 0.2250256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5846,
      "step": 6573
    },
    {
      "epoch": 0.22505991098938719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6005,
      "step": 6574
    },
    {
      "epoch": 0.22509414584046558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6092,
      "step": 6575
    },
    {
      "epoch": 0.225128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6223,
      "step": 6576
    },
    {
      "epoch": 0.2251626155426224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6128,
      "step": 6577
    },
    {
      "epoch": 0.2251968503937008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5802,
      "step": 6578
    },
    {
      "epoch": 0.2252310852447792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6052,
      "step": 6579
    },
    {
      "epoch": 0.2252653200958576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.593,
      "step": 6580
    },
    {
      "epoch": 0.225299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6032,
      "step": 6581
    },
    {
      "epoch": 0.22533378979801438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6002,
      "step": 6582
    },
    {
      "epoch": 0.22536802464909278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6032,
      "step": 6583
    },
    {
      "epoch": 0.22540225950017118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6068,
      "step": 6584
    },
    {
      "epoch": 0.22543649435124957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5828,
      "step": 6585
    },
    {
      "epoch": 0.22547072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5877,
      "step": 6586
    },
    {
      "epoch": 0.22550496405340636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6013,
      "step": 6587
    },
    {
      "epoch": 0.22553919890448476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5592,
      "step": 6588
    },
    {
      "epoch": 0.22557343375556316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.599,
      "step": 6589
    },
    {
      "epoch": 0.22560766860664155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5922,
      "step": 6590
    },
    {
      "epoch": 0.22564190345771995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5982,
      "step": 6591
    },
    {
      "epoch": 0.22567613830879835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6049,
      "step": 6592
    },
    {
      "epoch": 0.22571037315987674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5884,
      "step": 6593
    },
    {
      "epoch": 0.22574460801095514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6118,
      "step": 6594
    },
    {
      "epoch": 0.22577884286203356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6008,
      "step": 6595
    },
    {
      "epoch": 0.22581307771311196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.596,
      "step": 6596
    },
    {
      "epoch": 0.22584731256419036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.597,
      "step": 6597
    },
    {
      "epoch": 0.22588154741526875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5897,
      "step": 6598
    },
    {
      "epoch": 0.22591578226634715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.593,
      "step": 6599
    },
    {
      "epoch": 0.22595001711742554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6108,
      "step": 6600
    },
    {
      "epoch": 0.22598425196850394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5934,
      "step": 6601
    },
    {
      "epoch": 0.22601848681958234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5944,
      "step": 6602
    },
    {
      "epoch": 0.22605272167066073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6004,
      "step": 6603
    },
    {
      "epoch": 0.22608695652173913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6031,
      "step": 6604
    },
    {
      "epoch": 0.22612119137281753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6027,
      "step": 6605
    },
    {
      "epoch": 0.22615542622389592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6041,
      "step": 6606
    },
    {
      "epoch": 0.22618966107497432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5882,
      "step": 6607
    },
    {
      "epoch": 0.22622389592605271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5835,
      "step": 6608
    },
    {
      "epoch": 0.2262581307771311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6012,
      "step": 6609
    },
    {
      "epoch": 0.2262923656282095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5788,
      "step": 6610
    },
    {
      "epoch": 0.2263266004792879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6107,
      "step": 6611
    },
    {
      "epoch": 0.2263608353303663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5991,
      "step": 6612
    },
    {
      "epoch": 0.22639507018144472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5881,
      "step": 6613
    },
    {
      "epoch": 0.22642930503252312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6067,
      "step": 6614
    },
    {
      "epoch": 0.22646353988360152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5845,
      "step": 6615
    },
    {
      "epoch": 0.2264977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5903,
      "step": 6616
    },
    {
      "epoch": 0.2265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5957,
      "step": 6617
    },
    {
      "epoch": 0.2265662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.601,
      "step": 6618
    },
    {
      "epoch": 0.2266004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6198,
      "step": 6619
    },
    {
      "epoch": 0.2266347141389935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6081,
      "step": 6620
    },
    {
      "epoch": 0.2266689489900719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5924,
      "step": 6621
    },
    {
      "epoch": 0.2267031838411503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6186,
      "step": 6622
    },
    {
      "epoch": 0.2267374186922287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6021,
      "step": 6623
    },
    {
      "epoch": 0.22677165354330708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5762,
      "step": 6624
    },
    {
      "epoch": 0.22680588839438548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5857,
      "step": 6625
    },
    {
      "epoch": 0.22684012324546388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.636,
      "step": 6626
    },
    {
      "epoch": 0.22687435809654227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5858,
      "step": 6627
    },
    {
      "epoch": 0.22690859294762067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6071,
      "step": 6628
    },
    {
      "epoch": 0.22694282779869906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6259,
      "step": 6629
    },
    {
      "epoch": 0.22697706264977746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5848,
      "step": 6630
    },
    {
      "epoch": 0.22701129750085586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6012,
      "step": 6631
    },
    {
      "epoch": 0.22704553235193428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5835,
      "step": 6632
    },
    {
      "epoch": 0.22707976720301268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.596,
      "step": 6633
    },
    {
      "epoch": 0.22711400205409107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5808,
      "step": 6634
    },
    {
      "epoch": 0.22714823690516947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6085,
      "step": 6635
    },
    {
      "epoch": 0.22718247175624787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5949,
      "step": 6636
    },
    {
      "epoch": 0.22721670660732626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5954,
      "step": 6637
    },
    {
      "epoch": 0.22725094145840466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5935,
      "step": 6638
    },
    {
      "epoch": 0.22728517630948306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5887,
      "step": 6639
    },
    {
      "epoch": 0.22731941116056145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5953,
      "step": 6640
    },
    {
      "epoch": 0.22735364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6212,
      "step": 6641
    },
    {
      "epoch": 0.22738788086271824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5983,
      "step": 6642
    },
    {
      "epoch": 0.22742211571379664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5829,
      "step": 6643
    },
    {
      "epoch": 0.22745635056487504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5962,
      "step": 6644
    },
    {
      "epoch": 0.22749058541595343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6001,
      "step": 6645
    },
    {
      "epoch": 0.22752482026703183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.601,
      "step": 6646
    },
    {
      "epoch": 0.22755905511811023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5825,
      "step": 6647
    },
    {
      "epoch": 0.22759328996918862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6084,
      "step": 6648
    },
    {
      "epoch": 0.22762752482026702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6121,
      "step": 6649
    },
    {
      "epoch": 0.22766175967134544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.618,
      "step": 6650
    },
    {
      "epoch": 0.22769599452242384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6091,
      "step": 6651
    },
    {
      "epoch": 0.22773022937350224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5836,
      "step": 6652
    },
    {
      "epoch": 0.22776446422458063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5985,
      "step": 6653
    },
    {
      "epoch": 0.22779869907565903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6023,
      "step": 6654
    },
    {
      "epoch": 0.22783293392673742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6084,
      "step": 6655
    },
    {
      "epoch": 0.22786716877781582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5989,
      "step": 6656
    },
    {
      "epoch": 0.22790140362889422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5894,
      "step": 6657
    },
    {
      "epoch": 0.2279356384799726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5778,
      "step": 6658
    },
    {
      "epoch": 0.227969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6051,
      "step": 6659
    },
    {
      "epoch": 0.2280041081821294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5729,
      "step": 6660
    },
    {
      "epoch": 0.2280383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.586,
      "step": 6661
    },
    {
      "epoch": 0.2280725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5744,
      "step": 6662
    },
    {
      "epoch": 0.2281068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5993,
      "step": 6663
    },
    {
      "epoch": 0.228141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5925,
      "step": 6664
    },
    {
      "epoch": 0.2281752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6112,
      "step": 6665
    },
    {
      "epoch": 0.22820951728859978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5966,
      "step": 6666
    },
    {
      "epoch": 0.22824375213967818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6041,
      "step": 6667
    },
    {
      "epoch": 0.2282779869907566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6026,
      "step": 6668
    },
    {
      "epoch": 0.228312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6032,
      "step": 6669
    },
    {
      "epoch": 0.2283464566929134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6079,
      "step": 6670
    },
    {
      "epoch": 0.2283806915439918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5881,
      "step": 6671
    },
    {
      "epoch": 0.2284149263950702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5882,
      "step": 6672
    },
    {
      "epoch": 0.22844916124614859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5863,
      "step": 6673
    },
    {
      "epoch": 0.22848339609722698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.586,
      "step": 6674
    },
    {
      "epoch": 0.22851763094830538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5919,
      "step": 6675
    },
    {
      "epoch": 0.22855186579938377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.601,
      "step": 6676
    },
    {
      "epoch": 0.22858610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5845,
      "step": 6677
    },
    {
      "epoch": 0.22862033550154057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6035,
      "step": 6678
    },
    {
      "epoch": 0.22865457035261896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6176,
      "step": 6679
    },
    {
      "epoch": 0.22868880520369736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6093,
      "step": 6680
    },
    {
      "epoch": 0.22872304005477576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6005,
      "step": 6681
    },
    {
      "epoch": 0.22875727490585415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5757,
      "step": 6682
    },
    {
      "epoch": 0.22879150975693255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5706,
      "step": 6683
    },
    {
      "epoch": 0.22882574460801094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5814,
      "step": 6684
    },
    {
      "epoch": 0.22885997945908934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.596,
      "step": 6685
    },
    {
      "epoch": 0.22889421431016774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5866,
      "step": 6686
    },
    {
      "epoch": 0.22892844916124616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5805,
      "step": 6687
    },
    {
      "epoch": 0.22896268401232456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5994,
      "step": 6688
    },
    {
      "epoch": 0.22899691886340295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5794,
      "step": 6689
    },
    {
      "epoch": 0.22903115371448135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5921,
      "step": 6690
    },
    {
      "epoch": 0.22906538856555975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5784,
      "step": 6691
    },
    {
      "epoch": 0.22909962341663814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5859,
      "step": 6692
    },
    {
      "epoch": 0.22913385826771654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5852,
      "step": 6693
    },
    {
      "epoch": 0.22916809311879494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5987,
      "step": 6694
    },
    {
      "epoch": 0.22920232796987333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5905,
      "step": 6695
    },
    {
      "epoch": 0.22923656282095173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5979,
      "step": 6696
    },
    {
      "epoch": 0.22927079767203012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5955,
      "step": 6697
    },
    {
      "epoch": 0.22930503252310852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5688,
      "step": 6698
    },
    {
      "epoch": 0.22933926737418692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6169,
      "step": 6699
    },
    {
      "epoch": 0.2293735022252653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5994,
      "step": 6700
    },
    {
      "epoch": 0.2294077370763437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5795,
      "step": 6701
    },
    {
      "epoch": 0.2294419719274221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5998,
      "step": 6702
    },
    {
      "epoch": 0.2294762067785005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5896,
      "step": 6703
    },
    {
      "epoch": 0.2295104416295789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6025,
      "step": 6704
    },
    {
      "epoch": 0.22954467648065732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5954,
      "step": 6705
    },
    {
      "epoch": 0.22957891133173572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5983,
      "step": 6706
    },
    {
      "epoch": 0.22961314618281412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5882,
      "step": 6707
    },
    {
      "epoch": 0.2296473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5789,
      "step": 6708
    },
    {
      "epoch": 0.2296816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5649,
      "step": 6709
    },
    {
      "epoch": 0.2297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.573,
      "step": 6710
    },
    {
      "epoch": 0.2297500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5887,
      "step": 6711
    },
    {
      "epoch": 0.2297843204382061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6037,
      "step": 6712
    },
    {
      "epoch": 0.2298185552892845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5906,
      "step": 6713
    },
    {
      "epoch": 0.2298527901403629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5872,
      "step": 6714
    },
    {
      "epoch": 0.22988702499144129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.591,
      "step": 6715
    },
    {
      "epoch": 0.22992125984251968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6033,
      "step": 6716
    },
    {
      "epoch": 0.22995549469359808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5994,
      "step": 6717
    },
    {
      "epoch": 0.22998972954467647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5875,
      "step": 6718
    },
    {
      "epoch": 0.23002396439575487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6093,
      "step": 6719
    },
    {
      "epoch": 0.23005819924683327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5778,
      "step": 6720
    },
    {
      "epoch": 0.23009243409791166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5976,
      "step": 6721
    },
    {
      "epoch": 0.23012666894899006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5777,
      "step": 6722
    },
    {
      "epoch": 0.23016090380006846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5846,
      "step": 6723
    },
    {
      "epoch": 0.23019513865114688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5904,
      "step": 6724
    },
    {
      "epoch": 0.23022937350222528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5985,
      "step": 6725
    },
    {
      "epoch": 0.23026360835330367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5748,
      "step": 6726
    },
    {
      "epoch": 0.23029784320438207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5888,
      "step": 6727
    },
    {
      "epoch": 0.23033207805546047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5879,
      "step": 6728
    },
    {
      "epoch": 0.23036631290653886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5885,
      "step": 6729
    },
    {
      "epoch": 0.23040054775761726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6037,
      "step": 6730
    },
    {
      "epoch": 0.23043478260869565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5947,
      "step": 6731
    },
    {
      "epoch": 0.23046901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5983,
      "step": 6732
    },
    {
      "epoch": 0.23050325231085245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.586,
      "step": 6733
    },
    {
      "epoch": 0.23053748716193084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.585,
      "step": 6734
    },
    {
      "epoch": 0.23057172201300924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5898,
      "step": 6735
    },
    {
      "epoch": 0.23060595686408764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.543,
      "step": 6736
    },
    {
      "epoch": 0.23064019171516603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5704,
      "step": 6737
    },
    {
      "epoch": 0.23067442656624443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.588,
      "step": 6738
    },
    {
      "epoch": 0.23070866141732282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5672,
      "step": 6739
    },
    {
      "epoch": 0.23074289626840122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5758,
      "step": 6740
    },
    {
      "epoch": 0.23077713111947962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5717,
      "step": 6741
    },
    {
      "epoch": 0.23081136597055804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5932,
      "step": 6742
    },
    {
      "epoch": 0.23084560082163644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5855,
      "step": 6743
    },
    {
      "epoch": 0.23087983567271483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5925,
      "step": 6744
    },
    {
      "epoch": 0.23091407052379323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5955,
      "step": 6745
    },
    {
      "epoch": 0.23094830537487163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6179,
      "step": 6746
    },
    {
      "epoch": 0.23098254022595002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5756,
      "step": 6747
    },
    {
      "epoch": 0.23101677507702842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5727,
      "step": 6748
    },
    {
      "epoch": 0.23105100992810682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.581,
      "step": 6749
    },
    {
      "epoch": 0.2310852447791852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5623,
      "step": 6750
    },
    {
      "epoch": 0.2311194796302636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5931,
      "step": 6751
    },
    {
      "epoch": 0.231153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.583,
      "step": 6752
    },
    {
      "epoch": 0.2311879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5971,
      "step": 6753
    },
    {
      "epoch": 0.2312221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5784,
      "step": 6754
    },
    {
      "epoch": 0.2312564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5581,
      "step": 6755
    },
    {
      "epoch": 0.2312906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5821,
      "step": 6756
    },
    {
      "epoch": 0.23132488873673399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5994,
      "step": 6757
    },
    {
      "epoch": 0.23135912358781238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5699,
      "step": 6758
    },
    {
      "epoch": 0.23139335843889078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5846,
      "step": 6759
    },
    {
      "epoch": 0.2314275932899692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.608,
      "step": 6760
    },
    {
      "epoch": 0.2314618281410476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5808,
      "step": 6761
    },
    {
      "epoch": 0.231496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5699,
      "step": 6762
    },
    {
      "epoch": 0.2315302978432044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5896,
      "step": 6763
    },
    {
      "epoch": 0.2315645326942828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5843,
      "step": 6764
    },
    {
      "epoch": 0.23159876754536118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5813,
      "step": 6765
    },
    {
      "epoch": 0.23163300239643958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5905,
      "step": 6766
    },
    {
      "epoch": 0.23166723724751798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5852,
      "step": 6767
    },
    {
      "epoch": 0.23170147209859637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5857,
      "step": 6768
    },
    {
      "epoch": 0.23173570694967477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6077,
      "step": 6769
    },
    {
      "epoch": 0.23176994180075317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6075,
      "step": 6770
    },
    {
      "epoch": 0.23180417665183156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5894,
      "step": 6771
    },
    {
      "epoch": 0.23183841150290996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5641,
      "step": 6772
    },
    {
      "epoch": 0.23187264635398835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5662,
      "step": 6773
    },
    {
      "epoch": 0.23190688120506675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5836,
      "step": 6774
    },
    {
      "epoch": 0.23194111605614515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5906,
      "step": 6775
    },
    {
      "epoch": 0.23197535090722354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6019,
      "step": 6776
    },
    {
      "epoch": 0.23200958575830194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5775,
      "step": 6777
    },
    {
      "epoch": 0.23204382060938034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5874,
      "step": 6778
    },
    {
      "epoch": 0.23207805546045876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5925,
      "step": 6779
    },
    {
      "epoch": 0.23211229031153716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5864,
      "step": 6780
    },
    {
      "epoch": 0.23214652516261555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5956,
      "step": 6781
    },
    {
      "epoch": 0.23218076001369395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5992,
      "step": 6782
    },
    {
      "epoch": 0.23221499486477234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5782,
      "step": 6783
    },
    {
      "epoch": 0.23224922971585074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5881,
      "step": 6784
    },
    {
      "epoch": 0.23228346456692914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.594,
      "step": 6785
    },
    {
      "epoch": 0.23231769941800753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5996,
      "step": 6786
    },
    {
      "epoch": 0.23235193426908593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5834,
      "step": 6787
    },
    {
      "epoch": 0.23238616912016433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5789,
      "step": 6788
    },
    {
      "epoch": 0.23242040397124272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.603,
      "step": 6789
    },
    {
      "epoch": 0.23245463882232112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5636,
      "step": 6790
    },
    {
      "epoch": 0.23248887367339952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5529,
      "step": 6791
    },
    {
      "epoch": 0.2325231085244779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.596,
      "step": 6792
    },
    {
      "epoch": 0.2325573433755563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5664,
      "step": 6793
    },
    {
      "epoch": 0.2325915782266347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5885,
      "step": 6794
    },
    {
      "epoch": 0.2326258130777131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5857,
      "step": 6795
    },
    {
      "epoch": 0.2326600479287915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5828,
      "step": 6796
    },
    {
      "epoch": 0.23269428277986992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.609,
      "step": 6797
    },
    {
      "epoch": 0.23272851763094832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6044,
      "step": 6798
    },
    {
      "epoch": 0.2327627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5841,
      "step": 6799
    },
    {
      "epoch": 0.2327969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5814,
      "step": 6800
    },
    {
      "epoch": 0.2328312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5909,
      "step": 6801
    },
    {
      "epoch": 0.2328654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5716,
      "step": 6802
    },
    {
      "epoch": 0.2328996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5987,
      "step": 6803
    },
    {
      "epoch": 0.2329339267374187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5891,
      "step": 6804
    },
    {
      "epoch": 0.2329681615884971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5869,
      "step": 6805
    },
    {
      "epoch": 0.2330023964395755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5601,
      "step": 6806
    },
    {
      "epoch": 0.23303663129065388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.57,
      "step": 6807
    },
    {
      "epoch": 0.23307086614173228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5643,
      "step": 6808
    },
    {
      "epoch": 0.23310510099281068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5787,
      "step": 6809
    },
    {
      "epoch": 0.23313933584388907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5807,
      "step": 6810
    },
    {
      "epoch": 0.23317357069496747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5919,
      "step": 6811
    },
    {
      "epoch": 0.23320780554604587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5943,
      "step": 6812
    },
    {
      "epoch": 0.23324204039712426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5847,
      "step": 6813
    },
    {
      "epoch": 0.23327627524820266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5959,
      "step": 6814
    },
    {
      "epoch": 0.23331051009928105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5697,
      "step": 6815
    },
    {
      "epoch": 0.23334474495035948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5866,
      "step": 6816
    },
    {
      "epoch": 0.23337897980143787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5744,
      "step": 6817
    },
    {
      "epoch": 0.23341321465251627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5869,
      "step": 6818
    },
    {
      "epoch": 0.23344744950359467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6034,
      "step": 6819
    },
    {
      "epoch": 0.23348168435467306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5689,
      "step": 6820
    },
    {
      "epoch": 0.23351591920575146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5878,
      "step": 6821
    },
    {
      "epoch": 0.23355015405682986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5549,
      "step": 6822
    },
    {
      "epoch": 0.23358438890790825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6012,
      "step": 6823
    },
    {
      "epoch": 0.23361862375898665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5849,
      "step": 6824
    },
    {
      "epoch": 0.23365285861006505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5784,
      "step": 6825
    },
    {
      "epoch": 0.23368709346114344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5822,
      "step": 6826
    },
    {
      "epoch": 0.23372132831222184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6183,
      "step": 6827
    },
    {
      "epoch": 0.23375556316330023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5849,
      "step": 6828
    },
    {
      "epoch": 0.23378979801437863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.563,
      "step": 6829
    },
    {
      "epoch": 0.23382403286545703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5886,
      "step": 6830
    },
    {
      "epoch": 0.23385826771653542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.559,
      "step": 6831
    },
    {
      "epoch": 0.23389250256761382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5624,
      "step": 6832
    },
    {
      "epoch": 0.23392673741869222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.588,
      "step": 6833
    },
    {
      "epoch": 0.23396097226977064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5781,
      "step": 6834
    },
    {
      "epoch": 0.23399520712084904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5756,
      "step": 6835
    },
    {
      "epoch": 0.23402944197192743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.587,
      "step": 6836
    },
    {
      "epoch": 0.23406367682300583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5711,
      "step": 6837
    },
    {
      "epoch": 0.23409791167408422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5871,
      "step": 6838
    },
    {
      "epoch": 0.23413214652516262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6001,
      "step": 6839
    },
    {
      "epoch": 0.23416638137624102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5857,
      "step": 6840
    },
    {
      "epoch": 0.2342006162273194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5682,
      "step": 6841
    },
    {
      "epoch": 0.2342348510783978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5913,
      "step": 6842
    },
    {
      "epoch": 0.2342690859294762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5828,
      "step": 6843
    },
    {
      "epoch": 0.2343033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5751,
      "step": 6844
    },
    {
      "epoch": 0.234337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6041,
      "step": 6845
    },
    {
      "epoch": 0.2343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5761,
      "step": 6846
    },
    {
      "epoch": 0.2344060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5883,
      "step": 6847
    },
    {
      "epoch": 0.2344402601848682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5881,
      "step": 6848
    },
    {
      "epoch": 0.23447449503594658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5808,
      "step": 6849
    },
    {
      "epoch": 0.23450872988702498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5928,
      "step": 6850
    },
    {
      "epoch": 0.23454296473810338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5744,
      "step": 6851
    },
    {
      "epoch": 0.2345771995891818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5784,
      "step": 6852
    },
    {
      "epoch": 0.2346114344402602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5614,
      "step": 6853
    },
    {
      "epoch": 0.2346456692913386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5764,
      "step": 6854
    },
    {
      "epoch": 0.234679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5752,
      "step": 6855
    },
    {
      "epoch": 0.23471413899349539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.585,
      "step": 6856
    },
    {
      "epoch": 0.23474837384457378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5529,
      "step": 6857
    },
    {
      "epoch": 0.23478260869565218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5867,
      "step": 6858
    },
    {
      "epoch": 0.23481684354673057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5803,
      "step": 6859
    },
    {
      "epoch": 0.23485107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.575,
      "step": 6860
    },
    {
      "epoch": 0.23488531324888737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5781,
      "step": 6861
    },
    {
      "epoch": 0.23491954809996576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5748,
      "step": 6862
    },
    {
      "epoch": 0.23495378295104416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5443,
      "step": 6863
    },
    {
      "epoch": 0.23498801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6016,
      "step": 6864
    },
    {
      "epoch": 0.23502225265320095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5703,
      "step": 6865
    },
    {
      "epoch": 0.23505648750427935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5564,
      "step": 6866
    },
    {
      "epoch": 0.23509072235535775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5897,
      "step": 6867
    },
    {
      "epoch": 0.23512495720643614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6037,
      "step": 6868
    },
    {
      "epoch": 0.23515919205751454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5819,
      "step": 6869
    },
    {
      "epoch": 0.23519342690859293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5808,
      "step": 6870
    },
    {
      "epoch": 0.23522766175967136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5589,
      "step": 6871
    },
    {
      "epoch": 0.23526189661074975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.556,
      "step": 6872
    },
    {
      "epoch": 0.23529613146182815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.591,
      "step": 6873
    },
    {
      "epoch": 0.23533036631290655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5614,
      "step": 6874
    },
    {
      "epoch": 0.23536460116398494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5597,
      "step": 6875
    },
    {
      "epoch": 0.23539883601506334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5507,
      "step": 6876
    },
    {
      "epoch": 0.23543307086614174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5918,
      "step": 6877
    },
    {
      "epoch": 0.23546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.578,
      "step": 6878
    },
    {
      "epoch": 0.23550154056829853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5761,
      "step": 6879
    },
    {
      "epoch": 0.23553577541937692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5644,
      "step": 6880
    },
    {
      "epoch": 0.23557001027045532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5813,
      "step": 6881
    },
    {
      "epoch": 0.23560424512153372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5687,
      "step": 6882
    },
    {
      "epoch": 0.2356384799726121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5794,
      "step": 6883
    },
    {
      "epoch": 0.2356727148236905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5792,
      "step": 6884
    },
    {
      "epoch": 0.2357069496747689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5764,
      "step": 6885
    },
    {
      "epoch": 0.2357411845258473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5825,
      "step": 6886
    },
    {
      "epoch": 0.2357754193769257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5434,
      "step": 6887
    },
    {
      "epoch": 0.2358096542280041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5758,
      "step": 6888
    },
    {
      "epoch": 0.23584388907908252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5713,
      "step": 6889
    },
    {
      "epoch": 0.23587812393016092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5922,
      "step": 6890
    },
    {
      "epoch": 0.2359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5792,
      "step": 6891
    },
    {
      "epoch": 0.2359465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5736,
      "step": 6892
    },
    {
      "epoch": 0.2359808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5667,
      "step": 6893
    },
    {
      "epoch": 0.2360150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5769,
      "step": 6894
    },
    {
      "epoch": 0.2360492981855529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5693,
      "step": 6895
    },
    {
      "epoch": 0.2360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5752,
      "step": 6896
    },
    {
      "epoch": 0.2361177678877097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5634,
      "step": 6897
    },
    {
      "epoch": 0.23615200273878809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.569,
      "step": 6898
    },
    {
      "epoch": 0.23618623758986648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5941,
      "step": 6899
    },
    {
      "epoch": 0.23622047244094488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5622,
      "step": 6900
    },
    {
      "epoch": 0.23625470729202327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5269,
      "step": 6901
    },
    {
      "epoch": 0.23628894214310167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5765,
      "step": 6902
    },
    {
      "epoch": 0.23632317699418007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5866,
      "step": 6903
    },
    {
      "epoch": 0.23635741184525846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5506,
      "step": 6904
    },
    {
      "epoch": 0.23639164669633686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5708,
      "step": 6905
    },
    {
      "epoch": 0.23642588154741526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5584,
      "step": 6906
    },
    {
      "epoch": 0.23646011639849365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5557,
      "step": 6907
    },
    {
      "epoch": 0.23649435124957208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6053,
      "step": 6908
    },
    {
      "epoch": 0.23652858610065047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.56,
      "step": 6909
    },
    {
      "epoch": 0.23656282095172887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5507,
      "step": 6910
    },
    {
      "epoch": 0.23659705580280727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5855,
      "step": 6911
    },
    {
      "epoch": 0.23663129065388566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5743,
      "step": 6912
    },
    {
      "epoch": 0.23666552550496406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5778,
      "step": 6913
    },
    {
      "epoch": 0.23669976035604245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5692,
      "step": 6914
    },
    {
      "epoch": 0.23673399520712085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.576,
      "step": 6915
    },
    {
      "epoch": 0.23676823005819925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5534,
      "step": 6916
    },
    {
      "epoch": 0.23680246490927764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5586,
      "step": 6917
    },
    {
      "epoch": 0.23683669976035604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5722,
      "step": 6918
    },
    {
      "epoch": 0.23687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5809,
      "step": 6919
    },
    {
      "epoch": 0.23690516946251283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5828,
      "step": 6920
    },
    {
      "epoch": 0.23693940431359123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5755,
      "step": 6921
    },
    {
      "epoch": 0.23697363916466962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5624,
      "step": 6922
    },
    {
      "epoch": 0.23700787401574802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5443,
      "step": 6923
    },
    {
      "epoch": 0.23704210886682642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5688,
      "step": 6924
    },
    {
      "epoch": 0.2370763437179048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5569,
      "step": 6925
    },
    {
      "epoch": 0.23711057856898324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.569,
      "step": 6926
    },
    {
      "epoch": 0.23714481342006163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5592,
      "step": 6927
    },
    {
      "epoch": 0.23717904827114003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5562,
      "step": 6928
    },
    {
      "epoch": 0.23721328312221843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5818,
      "step": 6929
    },
    {
      "epoch": 0.23724751797329682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5765,
      "step": 6930
    },
    {
      "epoch": 0.23728175282437522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5466,
      "step": 6931
    },
    {
      "epoch": 0.23731598767545362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5711,
      "step": 6932
    },
    {
      "epoch": 0.237350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.567,
      "step": 6933
    },
    {
      "epoch": 0.2373844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5562,
      "step": 6934
    },
    {
      "epoch": 0.2374186922286888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5748,
      "step": 6935
    },
    {
      "epoch": 0.2374529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5418,
      "step": 6936
    },
    {
      "epoch": 0.2374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5639,
      "step": 6937
    },
    {
      "epoch": 0.237521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5978,
      "step": 6938
    },
    {
      "epoch": 0.2375556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5686,
      "step": 6939
    },
    {
      "epoch": 0.2375898664840808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5867,
      "step": 6940
    },
    {
      "epoch": 0.23762410133515918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5612,
      "step": 6941
    },
    {
      "epoch": 0.23765833618623758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5835,
      "step": 6942
    },
    {
      "epoch": 0.23769257103731598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5654,
      "step": 6943
    },
    {
      "epoch": 0.2377268058883944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5595,
      "step": 6944
    },
    {
      "epoch": 0.2377610407394728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6004,
      "step": 6945
    },
    {
      "epoch": 0.2377952755905512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5736,
      "step": 6946
    },
    {
      "epoch": 0.2378295104416296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5686,
      "step": 6947
    },
    {
      "epoch": 0.23786374529270798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5619,
      "step": 6948
    },
    {
      "epoch": 0.23789798014378638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5826,
      "step": 6949
    },
    {
      "epoch": 0.23793221499486478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5548,
      "step": 6950
    },
    {
      "epoch": 0.23796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5521,
      "step": 6951
    },
    {
      "epoch": 0.23800068469702157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5685,
      "step": 6952
    },
    {
      "epoch": 0.23803491954809997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5569,
      "step": 6953
    },
    {
      "epoch": 0.23806915439917836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5595,
      "step": 6954
    },
    {
      "epoch": 0.23810338925025676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5866,
      "step": 6955
    },
    {
      "epoch": 0.23813762410133515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5761,
      "step": 6956
    },
    {
      "epoch": 0.23817185895241355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5695,
      "step": 6957
    },
    {
      "epoch": 0.23820609380349195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5752,
      "step": 6958
    },
    {
      "epoch": 0.23824032865457034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5383,
      "step": 6959
    },
    {
      "epoch": 0.23827456350564874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5889,
      "step": 6960
    },
    {
      "epoch": 0.23830879835672714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5692,
      "step": 6961
    },
    {
      "epoch": 0.23834303320780553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5845,
      "step": 6962
    },
    {
      "epoch": 0.23837726805888396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5594,
      "step": 6963
    },
    {
      "epoch": 0.23841150290996235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.56,
      "step": 6964
    },
    {
      "epoch": 0.23844573776104075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5705,
      "step": 6965
    },
    {
      "epoch": 0.23847997261211915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5799,
      "step": 6966
    },
    {
      "epoch": 0.23851420746319754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5437,
      "step": 6967
    },
    {
      "epoch": 0.23854844231427594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5764,
      "step": 6968
    },
    {
      "epoch": 0.23858267716535433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5735,
      "step": 6969
    },
    {
      "epoch": 0.23861691201643273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5735,
      "step": 6970
    },
    {
      "epoch": 0.23865114686751113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5954,
      "step": 6971
    },
    {
      "epoch": 0.23868538171858952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5523,
      "step": 6972
    },
    {
      "epoch": 0.23871961656966792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5676,
      "step": 6973
    },
    {
      "epoch": 0.23875385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5686,
      "step": 6974
    },
    {
      "epoch": 0.2387880862718247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.586,
      "step": 6975
    },
    {
      "epoch": 0.2388223211229031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5464,
      "step": 6976
    },
    {
      "epoch": 0.2388565559739815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5739,
      "step": 6977
    },
    {
      "epoch": 0.2388907908250599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5662,
      "step": 6978
    },
    {
      "epoch": 0.2389250256761383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.56,
      "step": 6979
    },
    {
      "epoch": 0.2389592605272167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5658,
      "step": 6980
    },
    {
      "epoch": 0.23899349537829512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5719,
      "step": 6981
    },
    {
      "epoch": 0.23902773022937351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5618,
      "step": 6982
    },
    {
      "epoch": 0.2390619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.574,
      "step": 6983
    },
    {
      "epoch": 0.2390961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5646,
      "step": 6984
    },
    {
      "epoch": 0.2391304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5502,
      "step": 6985
    },
    {
      "epoch": 0.2391646696336871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5638,
      "step": 6986
    },
    {
      "epoch": 0.2391989044847655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5567,
      "step": 6987
    },
    {
      "epoch": 0.2392331393358439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5565,
      "step": 6988
    },
    {
      "epoch": 0.2392673741869223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5617,
      "step": 6989
    },
    {
      "epoch": 0.23930160903800068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5692,
      "step": 6990
    },
    {
      "epoch": 0.23933584388907908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5743,
      "step": 6991
    },
    {
      "epoch": 0.23937007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5739,
      "step": 6992
    },
    {
      "epoch": 0.23940431359123587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.567,
      "step": 6993
    },
    {
      "epoch": 0.23943854844231427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5856,
      "step": 6994
    },
    {
      "epoch": 0.23947278329339267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5658,
      "step": 6995
    },
    {
      "epoch": 0.23950701814447106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5845,
      "step": 6996
    },
    {
      "epoch": 0.23954125299554946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5916,
      "step": 6997
    },
    {
      "epoch": 0.23957548784662785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5833,
      "step": 6998
    },
    {
      "epoch": 0.23960972269770628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5681,
      "step": 6999
    },
    {
      "epoch": 0.23964395754878468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5794,
      "step": 7000
    },
    {
      "epoch": 0.23967819239986307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5798,
      "step": 7001
    },
    {
      "epoch": 0.23971242725094147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.564,
      "step": 7002
    },
    {
      "epoch": 0.23974666210201986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5667,
      "step": 7003
    },
    {
      "epoch": 0.23978089695309826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5662,
      "step": 7004
    },
    {
      "epoch": 0.23981513180417666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5687,
      "step": 7005
    },
    {
      "epoch": 0.23984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5601,
      "step": 7006
    },
    {
      "epoch": 0.23988360150633345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5596,
      "step": 7007
    },
    {
      "epoch": 0.23991783635741185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5724,
      "step": 7008
    },
    {
      "epoch": 0.23995207120849024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5639,
      "step": 7009
    },
    {
      "epoch": 0.23998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5576,
      "step": 7010
    },
    {
      "epoch": 0.24002054091064703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.552,
      "step": 7011
    },
    {
      "epoch": 0.24005477576172543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5758,
      "step": 7012
    },
    {
      "epoch": 0.24008901061280383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5566,
      "step": 7013
    },
    {
      "epoch": 0.24012324546388222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5814,
      "step": 7014
    },
    {
      "epoch": 0.24015748031496062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.568,
      "step": 7015
    },
    {
      "epoch": 0.24019171516603902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.573,
      "step": 7016
    },
    {
      "epoch": 0.2402259500171174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5471,
      "step": 7017
    },
    {
      "epoch": 0.24026018486819584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5637,
      "step": 7018
    },
    {
      "epoch": 0.24029441971927423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5693,
      "step": 7019
    },
    {
      "epoch": 0.24032865457035263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5707,
      "step": 7020
    },
    {
      "epoch": 0.24036288942143103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5648,
      "step": 7021
    },
    {
      "epoch": 0.24039712427250942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5604,
      "step": 7022
    },
    {
      "epoch": 0.24043135912358782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5631,
      "step": 7023
    },
    {
      "epoch": 0.24046559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5731,
      "step": 7024
    },
    {
      "epoch": 0.2404998288257446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5611,
      "step": 7025
    },
    {
      "epoch": 0.240534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5282,
      "step": 7026
    },
    {
      "epoch": 0.2405682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5587,
      "step": 7027
    },
    {
      "epoch": 0.2406025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5842,
      "step": 7028
    },
    {
      "epoch": 0.2406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.601,
      "step": 7029
    },
    {
      "epoch": 0.2406710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.554,
      "step": 7030
    },
    {
      "epoch": 0.240705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5467,
      "step": 7031
    },
    {
      "epoch": 0.24073947278329338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5641,
      "step": 7032
    },
    {
      "epoch": 0.24077370763437178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.578,
      "step": 7033
    },
    {
      "epoch": 0.24080794248545018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5656,
      "step": 7034
    },
    {
      "epoch": 0.24084217733652857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5641,
      "step": 7035
    },
    {
      "epoch": 0.240876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5374,
      "step": 7036
    },
    {
      "epoch": 0.2409106470386854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5758,
      "step": 7037
    },
    {
      "epoch": 0.2409448818897638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5792,
      "step": 7038
    },
    {
      "epoch": 0.2409791167408422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5709,
      "step": 7039
    },
    {
      "epoch": 0.24101335159192058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.587,
      "step": 7040
    },
    {
      "epoch": 0.24104758644299898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5574,
      "step": 7041
    },
    {
      "epoch": 0.24108182129407738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5626,
      "step": 7042
    },
    {
      "epoch": 0.24111605614515577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.582,
      "step": 7043
    },
    {
      "epoch": 0.24115029099623417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5551,
      "step": 7044
    },
    {
      "epoch": 0.24118452584731256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5601,
      "step": 7045
    },
    {
      "epoch": 0.24121876069839096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5492,
      "step": 7046
    },
    {
      "epoch": 0.24125299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5431,
      "step": 7047
    },
    {
      "epoch": 0.24128723040054775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5668,
      "step": 7048
    },
    {
      "epoch": 0.24132146525162615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.564,
      "step": 7049
    },
    {
      "epoch": 0.24135570010270455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5493,
      "step": 7050
    },
    {
      "epoch": 0.24138993495378294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5649,
      "step": 7051
    },
    {
      "epoch": 0.24142416980486134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.549,
      "step": 7052
    },
    {
      "epoch": 0.24145840465593973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5436,
      "step": 7053
    },
    {
      "epoch": 0.24149263950701813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5561,
      "step": 7054
    },
    {
      "epoch": 0.24152687435809655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5592,
      "step": 7055
    },
    {
      "epoch": 0.24156110920917495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5714,
      "step": 7056
    },
    {
      "epoch": 0.24159534406025335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5415,
      "step": 7057
    },
    {
      "epoch": 0.24162957891133174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5743,
      "step": 7058
    },
    {
      "epoch": 0.24166381376241014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5846,
      "step": 7059
    },
    {
      "epoch": 0.24169804861348854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5527,
      "step": 7060
    },
    {
      "epoch": 0.24173228346456693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.551,
      "step": 7061
    },
    {
      "epoch": 0.24176651831564533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5575,
      "step": 7062
    },
    {
      "epoch": 0.24180075316672373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5662,
      "step": 7063
    },
    {
      "epoch": 0.24183498801780212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.558,
      "step": 7064
    },
    {
      "epoch": 0.24186922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5755,
      "step": 7065
    },
    {
      "epoch": 0.24190345771995891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5455,
      "step": 7066
    },
    {
      "epoch": 0.2419376925710373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5574,
      "step": 7067
    },
    {
      "epoch": 0.2419719274221157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5702,
      "step": 7068
    },
    {
      "epoch": 0.2420061622731941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5753,
      "step": 7069
    },
    {
      "epoch": 0.2420403971242725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5528,
      "step": 7070
    },
    {
      "epoch": 0.2420746319753509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5629,
      "step": 7071
    },
    {
      "epoch": 0.2421088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5652,
      "step": 7072
    },
    {
      "epoch": 0.24214310167750772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5274,
      "step": 7073
    },
    {
      "epoch": 0.2421773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5318,
      "step": 7074
    },
    {
      "epoch": 0.2422115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.582,
      "step": 7075
    },
    {
      "epoch": 0.2422458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5586,
      "step": 7076
    },
    {
      "epoch": 0.2422800410818213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5654,
      "step": 7077
    },
    {
      "epoch": 0.2423142759328997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5644,
      "step": 7078
    },
    {
      "epoch": 0.2423485107839781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5484,
      "step": 7079
    },
    {
      "epoch": 0.2423827456350565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5746,
      "step": 7080
    },
    {
      "epoch": 0.2424169804861349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5753,
      "step": 7081
    },
    {
      "epoch": 0.24245121533721328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5803,
      "step": 7082
    },
    {
      "epoch": 0.24248545018829168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5508,
      "step": 7083
    },
    {
      "epoch": 0.24251968503937008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5639,
      "step": 7084
    },
    {
      "epoch": 0.24255391989044847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5563,
      "step": 7085
    },
    {
      "epoch": 0.24258815474152687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5781,
      "step": 7086
    },
    {
      "epoch": 0.24262238959260526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5401,
      "step": 7087
    },
    {
      "epoch": 0.24265662444368366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5777,
      "step": 7088
    },
    {
      "epoch": 0.24269085929476206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5419,
      "step": 7089
    },
    {
      "epoch": 0.24272509414584045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5665,
      "step": 7090
    },
    {
      "epoch": 0.24275932899691888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5465,
      "step": 7091
    },
    {
      "epoch": 0.24279356384799727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5596,
      "step": 7092
    },
    {
      "epoch": 0.24282779869907567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5722,
      "step": 7093
    },
    {
      "epoch": 0.24286203355015407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5412,
      "step": 7094
    },
    {
      "epoch": 0.24289626840123246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5562,
      "step": 7095
    },
    {
      "epoch": 0.24293050325231086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5605,
      "step": 7096
    },
    {
      "epoch": 0.24296473810338926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5813,
      "step": 7097
    },
    {
      "epoch": 0.24299897295446765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5652,
      "step": 7098
    },
    {
      "epoch": 0.24303320780554605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5373,
      "step": 7099
    },
    {
      "epoch": 0.24306744265662444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5631,
      "step": 7100
    },
    {
      "epoch": 0.24310167750770284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5504,
      "step": 7101
    },
    {
      "epoch": 0.24313591235878124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5576,
      "step": 7102
    },
    {
      "epoch": 0.24317014720985963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5454,
      "step": 7103
    },
    {
      "epoch": 0.24320438206093803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5566,
      "step": 7104
    },
    {
      "epoch": 0.24323861691201643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5612,
      "step": 7105
    },
    {
      "epoch": 0.24327285176309482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5405,
      "step": 7106
    },
    {
      "epoch": 0.24330708661417322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5695,
      "step": 7107
    },
    {
      "epoch": 0.24334132146525161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5777,
      "step": 7108
    },
    {
      "epoch": 0.24337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5452,
      "step": 7109
    },
    {
      "epoch": 0.24340979116740843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5715,
      "step": 7110
    },
    {
      "epoch": 0.24344402601848683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5788,
      "step": 7111
    },
    {
      "epoch": 0.24347826086956523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5598,
      "step": 7112
    },
    {
      "epoch": 0.24351249572064362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5653,
      "step": 7113
    },
    {
      "epoch": 0.24354673057172202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5559,
      "step": 7114
    },
    {
      "epoch": 0.24358096542280042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5623,
      "step": 7115
    },
    {
      "epoch": 0.2436152002738788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5625,
      "step": 7116
    },
    {
      "epoch": 0.2436494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5565,
      "step": 7117
    },
    {
      "epoch": 0.2436836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5674,
      "step": 7118
    },
    {
      "epoch": 0.243717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5729,
      "step": 7119
    },
    {
      "epoch": 0.2437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5598,
      "step": 7120
    },
    {
      "epoch": 0.2437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5637,
      "step": 7121
    },
    {
      "epoch": 0.2438206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5595,
      "step": 7122
    },
    {
      "epoch": 0.2438548442314276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5721,
      "step": 7123
    },
    {
      "epoch": 0.24388907908250598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5611,
      "step": 7124
    },
    {
      "epoch": 0.24392331393358438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5521,
      "step": 7125
    },
    {
      "epoch": 0.24395754878466278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.546,
      "step": 7126
    },
    {
      "epoch": 0.24399178363574117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5716,
      "step": 7127
    },
    {
      "epoch": 0.2440260184868196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5552,
      "step": 7128
    },
    {
      "epoch": 0.244060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.571,
      "step": 7129
    },
    {
      "epoch": 0.2440944881889764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5584,
      "step": 7130
    },
    {
      "epoch": 0.24412872304005478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5668,
      "step": 7131
    },
    {
      "epoch": 0.24416295789113318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5441,
      "step": 7132
    },
    {
      "epoch": 0.24419719274221158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.575,
      "step": 7133
    },
    {
      "epoch": 0.24423142759328997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5599,
      "step": 7134
    },
    {
      "epoch": 0.24426566244436837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5766,
      "step": 7135
    },
    {
      "epoch": 0.24429989729544677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5693,
      "step": 7136
    },
    {
      "epoch": 0.24433413214652516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5702,
      "step": 7137
    },
    {
      "epoch": 0.24436836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5571,
      "step": 7138
    },
    {
      "epoch": 0.24440260184868196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5549,
      "step": 7139
    },
    {
      "epoch": 0.24443683669976035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5521,
      "step": 7140
    },
    {
      "epoch": 0.24447107155083875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5562,
      "step": 7141
    },
    {
      "epoch": 0.24450530640191714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5606,
      "step": 7142
    },
    {
      "epoch": 0.24453954125299554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5282,
      "step": 7143
    },
    {
      "epoch": 0.24457377610407394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5579,
      "step": 7144
    },
    {
      "epoch": 0.24460801095515233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5553,
      "step": 7145
    },
    {
      "epoch": 0.24464224580623073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5564,
      "step": 7146
    },
    {
      "epoch": 0.24467648065730915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5661,
      "step": 7147
    },
    {
      "epoch": 0.24471071550838755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5531,
      "step": 7148
    },
    {
      "epoch": 0.24474495035946595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5576,
      "step": 7149
    },
    {
      "epoch": 0.24477918521054434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5799,
      "step": 7150
    },
    {
      "epoch": 0.24481342006162274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5366,
      "step": 7151
    },
    {
      "epoch": 0.24484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5444,
      "step": 7152
    },
    {
      "epoch": 0.24488188976377953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5385,
      "step": 7153
    },
    {
      "epoch": 0.24491612461485793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5553,
      "step": 7154
    },
    {
      "epoch": 0.24495035946593632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5242,
      "step": 7155
    },
    {
      "epoch": 0.24498459431701472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5726,
      "step": 7156
    },
    {
      "epoch": 0.24501882916809312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5546,
      "step": 7157
    },
    {
      "epoch": 0.2450530640191715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5607,
      "step": 7158
    },
    {
      "epoch": 0.2450872988702499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5525,
      "step": 7159
    },
    {
      "epoch": 0.2451215337213283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5644,
      "step": 7160
    },
    {
      "epoch": 0.2451557685724067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5524,
      "step": 7161
    },
    {
      "epoch": 0.2451900034234851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5458,
      "step": 7162
    },
    {
      "epoch": 0.2452242382745635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.556,
      "step": 7163
    },
    {
      "epoch": 0.2452584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.556,
      "step": 7164
    },
    {
      "epoch": 0.24529270797672031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5582,
      "step": 7165
    },
    {
      "epoch": 0.2453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5601,
      "step": 7166
    },
    {
      "epoch": 0.2453611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5538,
      "step": 7167
    },
    {
      "epoch": 0.2453954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5493,
      "step": 7168
    },
    {
      "epoch": 0.2454296473810339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5601,
      "step": 7169
    },
    {
      "epoch": 0.2454638822321123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5786,
      "step": 7170
    },
    {
      "epoch": 0.2454981170831907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.57,
      "step": 7171
    },
    {
      "epoch": 0.2455323519342691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5519,
      "step": 7172
    },
    {
      "epoch": 0.24556658678534748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5712,
      "step": 7173
    },
    {
      "epoch": 0.24560082163642588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5508,
      "step": 7174
    },
    {
      "epoch": 0.24563505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5448,
      "step": 7175
    },
    {
      "epoch": 0.24566929133858267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5551,
      "step": 7176
    },
    {
      "epoch": 0.24570352618966107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5629,
      "step": 7177
    },
    {
      "epoch": 0.24573776104073947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5471,
      "step": 7178
    },
    {
      "epoch": 0.24577199589181786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5516,
      "step": 7179
    },
    {
      "epoch": 0.24580623074289626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5578,
      "step": 7180
    },
    {
      "epoch": 0.24584046559397466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.56,
      "step": 7181
    },
    {
      "epoch": 0.24587470044505305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5683,
      "step": 7182
    },
    {
      "epoch": 0.24590893529613148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5673,
      "step": 7183
    },
    {
      "epoch": 0.24594317014720987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5585,
      "step": 7184
    },
    {
      "epoch": 0.24597740499828827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5614,
      "step": 7185
    },
    {
      "epoch": 0.24601163984936666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5462,
      "step": 7186
    },
    {
      "epoch": 0.24604587470044506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5773,
      "step": 7187
    },
    {
      "epoch": 0.24608010955152346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.538,
      "step": 7188
    },
    {
      "epoch": 0.24611434440260185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5485,
      "step": 7189
    },
    {
      "epoch": 0.24614857925368025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5678,
      "step": 7190
    },
    {
      "epoch": 0.24618281410475865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5551,
      "step": 7191
    },
    {
      "epoch": 0.24621704895583704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5561,
      "step": 7192
    },
    {
      "epoch": 0.24625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5583,
      "step": 7193
    },
    {
      "epoch": 0.24628551865799383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5903,
      "step": 7194
    },
    {
      "epoch": 0.24631975350907223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5516,
      "step": 7195
    },
    {
      "epoch": 0.24635398836015063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5438,
      "step": 7196
    },
    {
      "epoch": 0.24638822321122902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5694,
      "step": 7197
    },
    {
      "epoch": 0.24642245806230742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5296,
      "step": 7198
    },
    {
      "epoch": 0.24645669291338582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.536,
      "step": 7199
    },
    {
      "epoch": 0.2464909277644642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5541,
      "step": 7200
    },
    {
      "epoch": 0.2465251626155426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5556,
      "step": 7201
    },
    {
      "epoch": 0.24655939746662103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5455,
      "step": 7202
    },
    {
      "epoch": 0.24659363231769943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5731,
      "step": 7203
    },
    {
      "epoch": 0.24662786716877783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5299,
      "step": 7204
    },
    {
      "epoch": 0.24666210201985622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5456,
      "step": 7205
    },
    {
      "epoch": 0.24669633687093462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5529,
      "step": 7206
    },
    {
      "epoch": 0.24673057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5632,
      "step": 7207
    },
    {
      "epoch": 0.2467648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5441,
      "step": 7208
    },
    {
      "epoch": 0.2467990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5412,
      "step": 7209
    },
    {
      "epoch": 0.2468332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5553,
      "step": 7210
    },
    {
      "epoch": 0.2468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5334,
      "step": 7211
    },
    {
      "epoch": 0.246901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5238,
      "step": 7212
    },
    {
      "epoch": 0.2469359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5753,
      "step": 7213
    },
    {
      "epoch": 0.2469702156795618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5383,
      "step": 7214
    },
    {
      "epoch": 0.24700445053064018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5479,
      "step": 7215
    },
    {
      "epoch": 0.24703868538171858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.551,
      "step": 7216
    },
    {
      "epoch": 0.24707292023279698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5501,
      "step": 7217
    },
    {
      "epoch": 0.24710715508387537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5518,
      "step": 7218
    },
    {
      "epoch": 0.24714138993495377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5712,
      "step": 7219
    },
    {
      "epoch": 0.2471756247860322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5449,
      "step": 7220
    },
    {
      "epoch": 0.2472098596371106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5428,
      "step": 7221
    },
    {
      "epoch": 0.247244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5561,
      "step": 7222
    },
    {
      "epoch": 0.24727832933926738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5521,
      "step": 7223
    },
    {
      "epoch": 0.24731256419034578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5191,
      "step": 7224
    },
    {
      "epoch": 0.24734679904142418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.535,
      "step": 7225
    },
    {
      "epoch": 0.24738103389250257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5474,
      "step": 7226
    },
    {
      "epoch": 0.24741526874358097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5549,
      "step": 7227
    },
    {
      "epoch": 0.24744950359465936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5597,
      "step": 7228
    },
    {
      "epoch": 0.24748373844573776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5469,
      "step": 7229
    },
    {
      "epoch": 0.24751797329681616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5662,
      "step": 7230
    },
    {
      "epoch": 0.24755220814789455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5377,
      "step": 7231
    },
    {
      "epoch": 0.24758644299897295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5445,
      "step": 7232
    },
    {
      "epoch": 0.24762067785005135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5726,
      "step": 7233
    },
    {
      "epoch": 0.24765491270112974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5423,
      "step": 7234
    },
    {
      "epoch": 0.24768914755220814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5277,
      "step": 7235
    },
    {
      "epoch": 0.24772338240328654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.526,
      "step": 7236
    },
    {
      "epoch": 0.24775761725436493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5526,
      "step": 7237
    },
    {
      "epoch": 0.24779185210544333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5487,
      "step": 7238
    },
    {
      "epoch": 0.24782608695652175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5703,
      "step": 7239
    },
    {
      "epoch": 0.24786032180760015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5463,
      "step": 7240
    },
    {
      "epoch": 0.24789455665867854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5473,
      "step": 7241
    },
    {
      "epoch": 0.24792879150975694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5399,
      "step": 7242
    },
    {
      "epoch": 0.24796302636083534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5576,
      "step": 7243
    },
    {
      "epoch": 0.24799726121191373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5166,
      "step": 7244
    },
    {
      "epoch": 0.24803149606299213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5403,
      "step": 7245
    },
    {
      "epoch": 0.24806573091407053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5444,
      "step": 7246
    },
    {
      "epoch": 0.24809996576514892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5583,
      "step": 7247
    },
    {
      "epoch": 0.24813420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5305,
      "step": 7248
    },
    {
      "epoch": 0.24816843546730571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5516,
      "step": 7249
    },
    {
      "epoch": 0.2482026703183841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5542,
      "step": 7250
    },
    {
      "epoch": 0.2482369051694625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5401,
      "step": 7251
    },
    {
      "epoch": 0.2482711400205409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.551,
      "step": 7252
    },
    {
      "epoch": 0.2483053748716193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5306,
      "step": 7253
    },
    {
      "epoch": 0.2483396097226977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5448,
      "step": 7254
    },
    {
      "epoch": 0.2483738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5296,
      "step": 7255
    },
    {
      "epoch": 0.2484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5565,
      "step": 7256
    },
    {
      "epoch": 0.2484423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5557,
      "step": 7257
    },
    {
      "epoch": 0.2484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5447,
      "step": 7258
    },
    {
      "epoch": 0.2485107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5549,
      "step": 7259
    },
    {
      "epoch": 0.2485450188291681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5346,
      "step": 7260
    },
    {
      "epoch": 0.2485792536802465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5384,
      "step": 7261
    },
    {
      "epoch": 0.2486134885313249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5441,
      "step": 7262
    },
    {
      "epoch": 0.2486477233824033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5409,
      "step": 7263
    },
    {
      "epoch": 0.2486819582334817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5512,
      "step": 7264
    },
    {
      "epoch": 0.24871619308456008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5394,
      "step": 7265
    },
    {
      "epoch": 0.24875042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.551,
      "step": 7266
    },
    {
      "epoch": 0.24878466278671688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.569,
      "step": 7267
    },
    {
      "epoch": 0.24881889763779527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.559,
      "step": 7268
    },
    {
      "epoch": 0.24885313248887367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.544,
      "step": 7269
    },
    {
      "epoch": 0.24888736733995206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5473,
      "step": 7270
    },
    {
      "epoch": 0.24892160219103046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5517,
      "step": 7271
    },
    {
      "epoch": 0.24895583704210886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5455,
      "step": 7272
    },
    {
      "epoch": 0.24899007189318725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5393,
      "step": 7273
    },
    {
      "epoch": 0.24902430674426565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5467,
      "step": 7274
    },
    {
      "epoch": 0.24905854159534407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5565,
      "step": 7275
    },
    {
      "epoch": 0.24909277644642247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5461,
      "step": 7276
    },
    {
      "epoch": 0.24912701129750087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5387,
      "step": 7277
    },
    {
      "epoch": 0.24916124614857926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5476,
      "step": 7278
    },
    {
      "epoch": 0.24919548099965766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5353,
      "step": 7279
    },
    {
      "epoch": 0.24922971585073606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5572,
      "step": 7280
    },
    {
      "epoch": 0.24926395070181445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5689,
      "step": 7281
    },
    {
      "epoch": 0.24929818555289285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5364,
      "step": 7282
    },
    {
      "epoch": 0.24933242040397124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5608,
      "step": 7283
    },
    {
      "epoch": 0.24936665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5648,
      "step": 7284
    },
    {
      "epoch": 0.24940089010612804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5359,
      "step": 7285
    },
    {
      "epoch": 0.24943512495720643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5409,
      "step": 7286
    },
    {
      "epoch": 0.24946935980828483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5511,
      "step": 7287
    },
    {
      "epoch": 0.24950359465936323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5355,
      "step": 7288
    },
    {
      "epoch": 0.24953782951044162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5542,
      "step": 7289
    },
    {
      "epoch": 0.24957206436152002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5339,
      "step": 7290
    },
    {
      "epoch": 0.24960629921259841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5631,
      "step": 7291
    },
    {
      "epoch": 0.2496405340636768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5455,
      "step": 7292
    },
    {
      "epoch": 0.2496747689147552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5303,
      "step": 7293
    },
    {
      "epoch": 0.24970900376583363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5383,
      "step": 7294
    },
    {
      "epoch": 0.24974323861691203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5495,
      "step": 7295
    },
    {
      "epoch": 0.24977747346799042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.534,
      "step": 7296
    },
    {
      "epoch": 0.24981170831906882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5264,
      "step": 7297
    },
    {
      "epoch": 0.24984594317014722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5521,
      "step": 7298
    },
    {
      "epoch": 0.2498801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5159,
      "step": 7299
    },
    {
      "epoch": 0.249914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5226,
      "step": 7300
    },
    {
      "epoch": 0.2499486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.525,
      "step": 7301
    },
    {
      "epoch": 0.2499828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5202,
      "step": 7302
    },
    {
      "epoch": 0.2500171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5088,
      "step": 7303
    },
    {
      "epoch": 0.2500513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.554,
      "step": 7304
    },
    {
      "epoch": 0.250085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5421,
      "step": 7305
    },
    {
      "epoch": 0.2501198219787744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5347,
      "step": 7306
    },
    {
      "epoch": 0.2501540568298528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5474,
      "step": 7307
    },
    {
      "epoch": 0.2501882916809312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5346,
      "step": 7308
    },
    {
      "epoch": 0.2502225265320096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5432,
      "step": 7309
    },
    {
      "epoch": 0.250256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.552,
      "step": 7310
    },
    {
      "epoch": 0.2502909962341664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5374,
      "step": 7311
    },
    {
      "epoch": 0.2503252310852448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5503,
      "step": 7312
    },
    {
      "epoch": 0.2503594659363232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5324,
      "step": 7313
    },
    {
      "epoch": 0.2503937007874016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5527,
      "step": 7314
    },
    {
      "epoch": 0.25042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5433,
      "step": 7315
    },
    {
      "epoch": 0.2504621704895584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.56,
      "step": 7316
    },
    {
      "epoch": 0.2504964053406368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5487,
      "step": 7317
    },
    {
      "epoch": 0.25053064019171517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5498,
      "step": 7318
    },
    {
      "epoch": 0.25056487504279357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5469,
      "step": 7319
    },
    {
      "epoch": 0.25059910989387196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5497,
      "step": 7320
    },
    {
      "epoch": 0.25063334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5484,
      "step": 7321
    },
    {
      "epoch": 0.25066757959602876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5311,
      "step": 7322
    },
    {
      "epoch": 0.25070181444710715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5459,
      "step": 7323
    },
    {
      "epoch": 0.25073604929818555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5366,
      "step": 7324
    },
    {
      "epoch": 0.25077028414926394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5388,
      "step": 7325
    },
    {
      "epoch": 0.25080451900034234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5294,
      "step": 7326
    },
    {
      "epoch": 0.25083875385142074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5399,
      "step": 7327
    },
    {
      "epoch": 0.25087298870249913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5374,
      "step": 7328
    },
    {
      "epoch": 0.25090722355357753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5366,
      "step": 7329
    },
    {
      "epoch": 0.2509414584046559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5589,
      "step": 7330
    },
    {
      "epoch": 0.2509756932557343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.561,
      "step": 7331
    },
    {
      "epoch": 0.2510099281068127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5309,
      "step": 7332
    },
    {
      "epoch": 0.2510441629578911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5537,
      "step": 7333
    },
    {
      "epoch": 0.2510783978089695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.53,
      "step": 7334
    },
    {
      "epoch": 0.2511126326600479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.525,
      "step": 7335
    },
    {
      "epoch": 0.2511468675111263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5474,
      "step": 7336
    },
    {
      "epoch": 0.2511811023622047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5239,
      "step": 7337
    },
    {
      "epoch": 0.2512153372132831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5504,
      "step": 7338
    },
    {
      "epoch": 0.2512495720643615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5197,
      "step": 7339
    },
    {
      "epoch": 0.25128380691543994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5469,
      "step": 7340
    },
    {
      "epoch": 0.25131804176651834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5553,
      "step": 7341
    },
    {
      "epoch": 0.25135227661759674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5469,
      "step": 7342
    },
    {
      "epoch": 0.25138651146867513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5483,
      "step": 7343
    },
    {
      "epoch": 0.25142074631975353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5499,
      "step": 7344
    },
    {
      "epoch": 0.2514549811708319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5347,
      "step": 7345
    },
    {
      "epoch": 0.2514892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5412,
      "step": 7346
    },
    {
      "epoch": 0.2515234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5444,
      "step": 7347
    },
    {
      "epoch": 0.2515576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5502,
      "step": 7348
    },
    {
      "epoch": 0.2515919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5454,
      "step": 7349
    },
    {
      "epoch": 0.2516261554262239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5404,
      "step": 7350
    },
    {
      "epoch": 0.2516603902773023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.545,
      "step": 7351
    },
    {
      "epoch": 0.2516946251283807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5399,
      "step": 7352
    },
    {
      "epoch": 0.2517288599794591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5288,
      "step": 7353
    },
    {
      "epoch": 0.2517630948305375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5485,
      "step": 7354
    },
    {
      "epoch": 0.2517973296816159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5396,
      "step": 7355
    },
    {
      "epoch": 0.2518315645326943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5533,
      "step": 7356
    },
    {
      "epoch": 0.2518657993837727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.56,
      "step": 7357
    },
    {
      "epoch": 0.2519000342348511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5322,
      "step": 7358
    },
    {
      "epoch": 0.2519342690859295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5578,
      "step": 7359
    },
    {
      "epoch": 0.25196850393700787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5297,
      "step": 7360
    },
    {
      "epoch": 0.25200273878808627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.515,
      "step": 7361
    },
    {
      "epoch": 0.25203697363916466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5271,
      "step": 7362
    },
    {
      "epoch": 0.25207120849024306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5532,
      "step": 7363
    },
    {
      "epoch": 0.25210544334132146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5462,
      "step": 7364
    },
    {
      "epoch": 0.25213967819239985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.539,
      "step": 7365
    },
    {
      "epoch": 0.25217391304347825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.523,
      "step": 7366
    },
    {
      "epoch": 0.25220814789455664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5449,
      "step": 7367
    },
    {
      "epoch": 0.25224238274563504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5557,
      "step": 7368
    },
    {
      "epoch": 0.25227661759671344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5463,
      "step": 7369
    },
    {
      "epoch": 0.25231085244779183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5393,
      "step": 7370
    },
    {
      "epoch": 0.25234508729887023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5468,
      "step": 7371
    },
    {
      "epoch": 0.2523793221499486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5552,
      "step": 7372
    },
    {
      "epoch": 0.252413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.544,
      "step": 7373
    },
    {
      "epoch": 0.2524477918521054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5305,
      "step": 7374
    },
    {
      "epoch": 0.2524820267031838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5481,
      "step": 7375
    },
    {
      "epoch": 0.2525162615542622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5386,
      "step": 7376
    },
    {
      "epoch": 0.25255049640534066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5525,
      "step": 7377
    },
    {
      "epoch": 0.25258473125641906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5264,
      "step": 7378
    },
    {
      "epoch": 0.25261896610749746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5368,
      "step": 7379
    },
    {
      "epoch": 0.25265320095857585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.543,
      "step": 7380
    },
    {
      "epoch": 0.25268743580965425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5263,
      "step": 7381
    },
    {
      "epoch": 0.25272167066073264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5418,
      "step": 7382
    },
    {
      "epoch": 0.25275590551181104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5487,
      "step": 7383
    },
    {
      "epoch": 0.25279014036288944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5438,
      "step": 7384
    },
    {
      "epoch": 0.25282437521396783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5407,
      "step": 7385
    },
    {
      "epoch": 0.25285861006504623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5364,
      "step": 7386
    },
    {
      "epoch": 0.2528928449161246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5311,
      "step": 7387
    },
    {
      "epoch": 0.252927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5338,
      "step": 7388
    },
    {
      "epoch": 0.2529613146182814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5292,
      "step": 7389
    },
    {
      "epoch": 0.2529955494693598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5278,
      "step": 7390
    },
    {
      "epoch": 0.2530297843204382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5255,
      "step": 7391
    },
    {
      "epoch": 0.2530640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5404,
      "step": 7392
    },
    {
      "epoch": 0.253098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5479,
      "step": 7393
    },
    {
      "epoch": 0.2531324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5602,
      "step": 7394
    },
    {
      "epoch": 0.2531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5356,
      "step": 7395
    },
    {
      "epoch": 0.2532009585758302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.552,
      "step": 7396
    },
    {
      "epoch": 0.2532351934269086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5415,
      "step": 7397
    },
    {
      "epoch": 0.253269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5583,
      "step": 7398
    },
    {
      "epoch": 0.2533036631290654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5133,
      "step": 7399
    },
    {
      "epoch": 0.2533378979801438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5418,
      "step": 7400
    },
    {
      "epoch": 0.2533721328312222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5462,
      "step": 7401
    },
    {
      "epoch": 0.25340636768230057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5315,
      "step": 7402
    },
    {
      "epoch": 0.25344060253337897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5422,
      "step": 7403
    },
    {
      "epoch": 0.25347483738445736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.537,
      "step": 7404
    },
    {
      "epoch": 0.25350907223553576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5252,
      "step": 7405
    },
    {
      "epoch": 0.25354330708661416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5396,
      "step": 7406
    },
    {
      "epoch": 0.25357754193769255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5353,
      "step": 7407
    },
    {
      "epoch": 0.25361177678877095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5404,
      "step": 7408
    },
    {
      "epoch": 0.25364601163984934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5492,
      "step": 7409
    },
    {
      "epoch": 0.25368024649092774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5374,
      "step": 7410
    },
    {
      "epoch": 0.25371448134200614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.54,
      "step": 7411
    },
    {
      "epoch": 0.25374871619308453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5438,
      "step": 7412
    },
    {
      "epoch": 0.253782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.537,
      "step": 7413
    },
    {
      "epoch": 0.2538171858952414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5307,
      "step": 7414
    },
    {
      "epoch": 0.2538514207463198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5332,
      "step": 7415
    },
    {
      "epoch": 0.2538856555973982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5192,
      "step": 7416
    },
    {
      "epoch": 0.25391989044847657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4957,
      "step": 7417
    },
    {
      "epoch": 0.25395412529955497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.521,
      "step": 7418
    },
    {
      "epoch": 0.25398836015063336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5457,
      "step": 7419
    },
    {
      "epoch": 0.25402259500171176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5307,
      "step": 7420
    },
    {
      "epoch": 0.25405682985279016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5469,
      "step": 7421
    },
    {
      "epoch": 0.25409106470386855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5064,
      "step": 7422
    },
    {
      "epoch": 0.25412529955494695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5331,
      "step": 7423
    },
    {
      "epoch": 0.25415953440602534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5546,
      "step": 7424
    },
    {
      "epoch": 0.25419376925710374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5192,
      "step": 7425
    },
    {
      "epoch": 0.25422800410818214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5372,
      "step": 7426
    },
    {
      "epoch": 0.25426223895926053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5462,
      "step": 7427
    },
    {
      "epoch": 0.25429647381033893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5341,
      "step": 7428
    },
    {
      "epoch": 0.2543307086614173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5363,
      "step": 7429
    },
    {
      "epoch": 0.2543649435124957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5569,
      "step": 7430
    },
    {
      "epoch": 0.2543991783635741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.53,
      "step": 7431
    },
    {
      "epoch": 0.2544334132146525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5541,
      "step": 7432
    },
    {
      "epoch": 0.2544676480657309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5656,
      "step": 7433
    },
    {
      "epoch": 0.2545018829168093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5503,
      "step": 7434
    },
    {
      "epoch": 0.2545361177678877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5429,
      "step": 7435
    },
    {
      "epoch": 0.2545703526189661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.543,
      "step": 7436
    },
    {
      "epoch": 0.2546045874700445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.565,
      "step": 7437
    },
    {
      "epoch": 0.2546388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.531,
      "step": 7438
    },
    {
      "epoch": 0.2546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5528,
      "step": 7439
    },
    {
      "epoch": 0.2547072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5323,
      "step": 7440
    },
    {
      "epoch": 0.2547415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5266,
      "step": 7441
    },
    {
      "epoch": 0.2547757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5375,
      "step": 7442
    },
    {
      "epoch": 0.2548099965765149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5511,
      "step": 7443
    },
    {
      "epoch": 0.25484423142759327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5192,
      "step": 7444
    },
    {
      "epoch": 0.25487846627867167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5332,
      "step": 7445
    },
    {
      "epoch": 0.25491270112975006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5421,
      "step": 7446
    },
    {
      "epoch": 0.25494693598082846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5277,
      "step": 7447
    },
    {
      "epoch": 0.25498117083190686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5515,
      "step": 7448
    },
    {
      "epoch": 0.25501540568298525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5564,
      "step": 7449
    },
    {
      "epoch": 0.2550496405340637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5313,
      "step": 7450
    },
    {
      "epoch": 0.2550838753851421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5292,
      "step": 7451
    },
    {
      "epoch": 0.2551181102362205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5339,
      "step": 7452
    },
    {
      "epoch": 0.2551523450872989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5274,
      "step": 7453
    },
    {
      "epoch": 0.2551865799383773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5548,
      "step": 7454
    },
    {
      "epoch": 0.2552208147894557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.518,
      "step": 7455
    },
    {
      "epoch": 0.2552550496405341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5428,
      "step": 7456
    },
    {
      "epoch": 0.2552892844916125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5479,
      "step": 7457
    },
    {
      "epoch": 0.2553235193426909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5222,
      "step": 7458
    },
    {
      "epoch": 0.25535775419376927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5554,
      "step": 7459
    },
    {
      "epoch": 0.25539198904484767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5286,
      "step": 7460
    },
    {
      "epoch": 0.25542622389592606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5211,
      "step": 7461
    },
    {
      "epoch": 0.25546045874700446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5472,
      "step": 7462
    },
    {
      "epoch": 0.25549469359808286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5401,
      "step": 7463
    },
    {
      "epoch": 0.25552892844916125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5184,
      "step": 7464
    },
    {
      "epoch": 0.25556316330023965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5389,
      "step": 7465
    },
    {
      "epoch": 0.25559739815131804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5437,
      "step": 7466
    },
    {
      "epoch": 0.25563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5467,
      "step": 7467
    },
    {
      "epoch": 0.25566586785347484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5298,
      "step": 7468
    },
    {
      "epoch": 0.25570010270455323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5085,
      "step": 7469
    },
    {
      "epoch": 0.25573433755563163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5411,
      "step": 7470
    },
    {
      "epoch": 0.25576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5494,
      "step": 7471
    },
    {
      "epoch": 0.2558028072577884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5327,
      "step": 7472
    },
    {
      "epoch": 0.2558370421088668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5236,
      "step": 7473
    },
    {
      "epoch": 0.2558712769599452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5007,
      "step": 7474
    },
    {
      "epoch": 0.2559055118110236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5315,
      "step": 7475
    },
    {
      "epoch": 0.255939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5128,
      "step": 7476
    },
    {
      "epoch": 0.2559739815131804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5114,
      "step": 7477
    },
    {
      "epoch": 0.2560082163642588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5366,
      "step": 7478
    },
    {
      "epoch": 0.2560424512153372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5423,
      "step": 7479
    },
    {
      "epoch": 0.2560766860664156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5443,
      "step": 7480
    },
    {
      "epoch": 0.256110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5374,
      "step": 7481
    },
    {
      "epoch": 0.2561451557685724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5294,
      "step": 7482
    },
    {
      "epoch": 0.2561793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5321,
      "step": 7483
    },
    {
      "epoch": 0.2562136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5195,
      "step": 7484
    },
    {
      "epoch": 0.2562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.552,
      "step": 7485
    },
    {
      "epoch": 0.25628209517288597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5394,
      "step": 7486
    },
    {
      "epoch": 0.2563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5357,
      "step": 7487
    },
    {
      "epoch": 0.2563505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5273,
      "step": 7488
    },
    {
      "epoch": 0.2563847997261212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5205,
      "step": 7489
    },
    {
      "epoch": 0.2564190345771996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5144,
      "step": 7490
    },
    {
      "epoch": 0.256453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5112,
      "step": 7491
    },
    {
      "epoch": 0.2564875042793564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.531,
      "step": 7492
    },
    {
      "epoch": 0.2565217391304348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.521,
      "step": 7493
    },
    {
      "epoch": 0.2565559739815132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5154,
      "step": 7494
    },
    {
      "epoch": 0.2565902088325916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5531,
      "step": 7495
    },
    {
      "epoch": 0.25662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5366,
      "step": 7496
    },
    {
      "epoch": 0.2566586785347484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5308,
      "step": 7497
    },
    {
      "epoch": 0.2566929133858268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5181,
      "step": 7498
    },
    {
      "epoch": 0.2567271482369052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.53,
      "step": 7499
    },
    {
      "epoch": 0.2567613830879836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5426,
      "step": 7500
    },
    {
      "epoch": 0.25679561793906197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5346,
      "step": 7501
    },
    {
      "epoch": 0.25682985279014037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5286,
      "step": 7502
    },
    {
      "epoch": 0.25686408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5218,
      "step": 7503
    },
    {
      "epoch": 0.25689832249229716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5205,
      "step": 7504
    },
    {
      "epoch": 0.25693255734337556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5396,
      "step": 7505
    },
    {
      "epoch": 0.25696679219445395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5354,
      "step": 7506
    },
    {
      "epoch": 0.25700102704553235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5175,
      "step": 7507
    },
    {
      "epoch": 0.25703526189661074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5433,
      "step": 7508
    },
    {
      "epoch": 0.25706949674768914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5297,
      "step": 7509
    },
    {
      "epoch": 0.25710373159876754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5243,
      "step": 7510
    },
    {
      "epoch": 0.25713796644984593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.521,
      "step": 7511
    },
    {
      "epoch": 0.25717220130092433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5142,
      "step": 7512
    },
    {
      "epoch": 0.2572064361520027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5437,
      "step": 7513
    },
    {
      "epoch": 0.2572406710030811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.51,
      "step": 7514
    },
    {
      "epoch": 0.2572749058541595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5178,
      "step": 7515
    },
    {
      "epoch": 0.2573091407052379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5397,
      "step": 7516
    },
    {
      "epoch": 0.2573433755563163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5338,
      "step": 7517
    },
    {
      "epoch": 0.2573776104073947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5377,
      "step": 7518
    },
    {
      "epoch": 0.2574118452584731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5276,
      "step": 7519
    },
    {
      "epoch": 0.2574460801095515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5178,
      "step": 7520
    },
    {
      "epoch": 0.2574803149606299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5562,
      "step": 7521
    },
    {
      "epoch": 0.2575145498117083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5369,
      "step": 7522
    },
    {
      "epoch": 0.2575487846627867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5155,
      "step": 7523
    },
    {
      "epoch": 0.25758301951386514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5493,
      "step": 7524
    },
    {
      "epoch": 0.25761725436494354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5257,
      "step": 7525
    },
    {
      "epoch": 0.25765148921602193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5194,
      "step": 7526
    },
    {
      "epoch": 0.25768572406710033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.528,
      "step": 7527
    },
    {
      "epoch": 0.2577199589181787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5201,
      "step": 7528
    },
    {
      "epoch": 0.2577541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5397,
      "step": 7529
    },
    {
      "epoch": 0.2577884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5433,
      "step": 7530
    },
    {
      "epoch": 0.2578226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.536,
      "step": 7531
    },
    {
      "epoch": 0.2578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5365,
      "step": 7532
    },
    {
      "epoch": 0.2578911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5137,
      "step": 7533
    },
    {
      "epoch": 0.2579253680246491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5136,
      "step": 7534
    },
    {
      "epoch": 0.2579596028757275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5406,
      "step": 7535
    },
    {
      "epoch": 0.2579938377268059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5291,
      "step": 7536
    },
    {
      "epoch": 0.2580280725778843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5319,
      "step": 7537
    },
    {
      "epoch": 0.2580623074289627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.533,
      "step": 7538
    },
    {
      "epoch": 0.2580965422800411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5336,
      "step": 7539
    },
    {
      "epoch": 0.2581307771311195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5227,
      "step": 7540
    },
    {
      "epoch": 0.2581650119821979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5294,
      "step": 7541
    },
    {
      "epoch": 0.2581992468332763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5139,
      "step": 7542
    },
    {
      "epoch": 0.25823348168435467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5035,
      "step": 7543
    },
    {
      "epoch": 0.25826771653543307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5188,
      "step": 7544
    },
    {
      "epoch": 0.25830195138651146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5324,
      "step": 7545
    },
    {
      "epoch": 0.25833618623758986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5235,
      "step": 7546
    },
    {
      "epoch": 0.25837042108866826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.53,
      "step": 7547
    },
    {
      "epoch": 0.25840465593974665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5082,
      "step": 7548
    },
    {
      "epoch": 0.25843889079082505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5487,
      "step": 7549
    },
    {
      "epoch": 0.25847312564190345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5302,
      "step": 7550
    },
    {
      "epoch": 0.25850736049298184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5354,
      "step": 7551
    },
    {
      "epoch": 0.25854159534406024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5355,
      "step": 7552
    },
    {
      "epoch": 0.25857583019513863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5258,
      "step": 7553
    },
    {
      "epoch": 0.25861006504621703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5448,
      "step": 7554
    },
    {
      "epoch": 0.2586442998972954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5222,
      "step": 7555
    },
    {
      "epoch": 0.2586785347483738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5174,
      "step": 7556
    },
    {
      "epoch": 0.2587127695994522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5065,
      "step": 7557
    },
    {
      "epoch": 0.2587470044505306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5295,
      "step": 7558
    },
    {
      "epoch": 0.258781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.528,
      "step": 7559
    },
    {
      "epoch": 0.2588154741526874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5526,
      "step": 7560
    },
    {
      "epoch": 0.25884970900376586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.529,
      "step": 7561
    },
    {
      "epoch": 0.25888394385484426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.51,
      "step": 7562
    },
    {
      "epoch": 0.25891817870592265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5281,
      "step": 7563
    },
    {
      "epoch": 0.25895241355700105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5272,
      "step": 7564
    },
    {
      "epoch": 0.25898664840807945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5496,
      "step": 7565
    },
    {
      "epoch": 0.25902088325915784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5303,
      "step": 7566
    },
    {
      "epoch": 0.25905511811023624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5089,
      "step": 7567
    },
    {
      "epoch": 0.25908935296131463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5269,
      "step": 7568
    },
    {
      "epoch": 0.25912358781239303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5288,
      "step": 7569
    },
    {
      "epoch": 0.2591578226634714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5372,
      "step": 7570
    },
    {
      "epoch": 0.2591920575145498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5327,
      "step": 7571
    },
    {
      "epoch": 0.2592262923656282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5534,
      "step": 7572
    },
    {
      "epoch": 0.2592605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5344,
      "step": 7573
    },
    {
      "epoch": 0.259294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5244,
      "step": 7574
    },
    {
      "epoch": 0.2593289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5353,
      "step": 7575
    },
    {
      "epoch": 0.2593632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5255,
      "step": 7576
    },
    {
      "epoch": 0.2593974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5267,
      "step": 7577
    },
    {
      "epoch": 0.2594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5331,
      "step": 7578
    },
    {
      "epoch": 0.259465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5048,
      "step": 7579
    },
    {
      "epoch": 0.2595001711742554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5167,
      "step": 7580
    },
    {
      "epoch": 0.2595344060253338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4996,
      "step": 7581
    },
    {
      "epoch": 0.2595686408764122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5368,
      "step": 7582
    },
    {
      "epoch": 0.2596028757274906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5256,
      "step": 7583
    },
    {
      "epoch": 0.259637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5292,
      "step": 7584
    },
    {
      "epoch": 0.25967134542964737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5189,
      "step": 7585
    },
    {
      "epoch": 0.25970558028072577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5222,
      "step": 7586
    },
    {
      "epoch": 0.25973981513180416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5246,
      "step": 7587
    },
    {
      "epoch": 0.25977404998288256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5265,
      "step": 7588
    },
    {
      "epoch": 0.25980828483396096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.533,
      "step": 7589
    },
    {
      "epoch": 0.25984251968503935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5321,
      "step": 7590
    },
    {
      "epoch": 0.25987675453611775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5553,
      "step": 7591
    },
    {
      "epoch": 0.25991098938719615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5129,
      "step": 7592
    },
    {
      "epoch": 0.25994522423827454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5189,
      "step": 7593
    },
    {
      "epoch": 0.25997945908935294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5414,
      "step": 7594
    },
    {
      "epoch": 0.26001369394043133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5057,
      "step": 7595
    },
    {
      "epoch": 0.26004792879150973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5301,
      "step": 7596
    },
    {
      "epoch": 0.2600821636425882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5304,
      "step": 7597
    },
    {
      "epoch": 0.2601163984936666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.509,
      "step": 7598
    },
    {
      "epoch": 0.260150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5352,
      "step": 7599
    },
    {
      "epoch": 0.26018486819582337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5344,
      "step": 7600
    },
    {
      "epoch": 0.26021910304690177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5267,
      "step": 7601
    },
    {
      "epoch": 0.26025333789798016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5158,
      "step": 7602
    },
    {
      "epoch": 0.26028757274905856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.515,
      "step": 7603
    },
    {
      "epoch": 0.26032180760013696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5127,
      "step": 7604
    },
    {
      "epoch": 0.26035604245121535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5296,
      "step": 7605
    },
    {
      "epoch": 0.26039027730229375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5321,
      "step": 7606
    },
    {
      "epoch": 0.26042451215337215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4924,
      "step": 7607
    },
    {
      "epoch": 0.26045874700445054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5184,
      "step": 7608
    },
    {
      "epoch": 0.26049298185552894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5186,
      "step": 7609
    },
    {
      "epoch": 0.26052721670660733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5058,
      "step": 7610
    },
    {
      "epoch": 0.26056145155768573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5302,
      "step": 7611
    },
    {
      "epoch": 0.2605956864087641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5256,
      "step": 7612
    },
    {
      "epoch": 0.2606299212598425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5132,
      "step": 7613
    },
    {
      "epoch": 0.2606641561109209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5416,
      "step": 7614
    },
    {
      "epoch": 0.2606983909619993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.516,
      "step": 7615
    },
    {
      "epoch": 0.2607326258130777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5194,
      "step": 7616
    },
    {
      "epoch": 0.2607668606641561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5208,
      "step": 7617
    },
    {
      "epoch": 0.2608010955152345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5191,
      "step": 7618
    },
    {
      "epoch": 0.2608353303663129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5225,
      "step": 7619
    },
    {
      "epoch": 0.2608695652173913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5111,
      "step": 7620
    },
    {
      "epoch": 0.2609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.517,
      "step": 7621
    },
    {
      "epoch": 0.2609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5379,
      "step": 7622
    },
    {
      "epoch": 0.2609722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5111,
      "step": 7623
    },
    {
      "epoch": 0.2610065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5009,
      "step": 7624
    },
    {
      "epoch": 0.2610407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5108,
      "step": 7625
    },
    {
      "epoch": 0.2610749743238617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5316,
      "step": 7626
    },
    {
      "epoch": 0.26110920917494007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5278,
      "step": 7627
    },
    {
      "epoch": 0.26114344402601847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5203,
      "step": 7628
    },
    {
      "epoch": 0.26117767887709686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5009,
      "step": 7629
    },
    {
      "epoch": 0.26121191372817526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5253,
      "step": 7630
    },
    {
      "epoch": 0.26124614857925366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5202,
      "step": 7631
    },
    {
      "epoch": 0.26128038343033205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5401,
      "step": 7632
    },
    {
      "epoch": 0.26131461828141045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.497,
      "step": 7633
    },
    {
      "epoch": 0.2613488531324889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5249,
      "step": 7634
    },
    {
      "epoch": 0.2613830879835673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5234,
      "step": 7635
    },
    {
      "epoch": 0.2614173228346457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5208,
      "step": 7636
    },
    {
      "epoch": 0.2614515576857241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5104,
      "step": 7637
    },
    {
      "epoch": 0.2614857925368025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5067,
      "step": 7638
    },
    {
      "epoch": 0.2615200273878809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5311,
      "step": 7639
    },
    {
      "epoch": 0.2615542622389593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5327,
      "step": 7640
    },
    {
      "epoch": 0.2615884970900377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5321,
      "step": 7641
    },
    {
      "epoch": 0.26162273194111607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5126,
      "step": 7642
    },
    {
      "epoch": 0.26165696679219447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5131,
      "step": 7643
    },
    {
      "epoch": 0.26169120164327286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5136,
      "step": 7644
    },
    {
      "epoch": 0.26172543649435126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5109,
      "step": 7645
    },
    {
      "epoch": 0.26175967134542966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5211,
      "step": 7646
    },
    {
      "epoch": 0.26179390619650805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5437,
      "step": 7647
    },
    {
      "epoch": 0.26182814104758645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5199,
      "step": 7648
    },
    {
      "epoch": 0.26186237589866485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5338,
      "step": 7649
    },
    {
      "epoch": 0.26189661074974324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5415,
      "step": 7650
    },
    {
      "epoch": 0.26193084560082164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5009,
      "step": 7651
    },
    {
      "epoch": 0.26196508045190003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.525,
      "step": 7652
    },
    {
      "epoch": 0.26199931530297843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5159,
      "step": 7653
    },
    {
      "epoch": 0.2620335501540568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5228,
      "step": 7654
    },
    {
      "epoch": 0.2620677850051352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5253,
      "step": 7655
    },
    {
      "epoch": 0.2621020198562136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.531,
      "step": 7656
    },
    {
      "epoch": 0.262136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5175,
      "step": 7657
    },
    {
      "epoch": 0.2621704895583704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5168,
      "step": 7658
    },
    {
      "epoch": 0.2622047244094488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5194,
      "step": 7659
    },
    {
      "epoch": 0.2622389592605272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5261,
      "step": 7660
    },
    {
      "epoch": 0.2622731941116056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5167,
      "step": 7661
    },
    {
      "epoch": 0.262307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5392,
      "step": 7662
    },
    {
      "epoch": 0.2623416638137624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5078,
      "step": 7663
    },
    {
      "epoch": 0.2623758986648408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.518,
      "step": 7664
    },
    {
      "epoch": 0.2624101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5103,
      "step": 7665
    },
    {
      "epoch": 0.2624443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5302,
      "step": 7666
    },
    {
      "epoch": 0.262478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5135,
      "step": 7667
    },
    {
      "epoch": 0.2625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5325,
      "step": 7668
    },
    {
      "epoch": 0.26254707292023277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5309,
      "step": 7669
    },
    {
      "epoch": 0.26258130777131117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5156,
      "step": 7670
    },
    {
      "epoch": 0.2626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5222,
      "step": 7671
    },
    {
      "epoch": 0.262649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5134,
      "step": 7672
    },
    {
      "epoch": 0.2626840123245464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5107,
      "step": 7673
    },
    {
      "epoch": 0.2627182471756248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5098,
      "step": 7674
    },
    {
      "epoch": 0.2627524820267032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5471,
      "step": 7675
    },
    {
      "epoch": 0.2627867168777816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5178,
      "step": 7676
    },
    {
      "epoch": 0.26282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5498,
      "step": 7677
    },
    {
      "epoch": 0.2628551865799384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5169,
      "step": 7678
    },
    {
      "epoch": 0.2628894214310168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5447,
      "step": 7679
    },
    {
      "epoch": 0.2629236562820952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5119,
      "step": 7680
    },
    {
      "epoch": 0.2629578911331736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5184,
      "step": 7681
    },
    {
      "epoch": 0.262992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5322,
      "step": 7682
    },
    {
      "epoch": 0.2630263608353304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5197,
      "step": 7683
    },
    {
      "epoch": 0.26306059568640877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5172,
      "step": 7684
    },
    {
      "epoch": 0.26309483053748717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5197,
      "step": 7685
    },
    {
      "epoch": 0.26312906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5216,
      "step": 7686
    },
    {
      "epoch": 0.26316330023964396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5075,
      "step": 7687
    },
    {
      "epoch": 0.26319753509072236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5129,
      "step": 7688
    },
    {
      "epoch": 0.26323176994180075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5241,
      "step": 7689
    },
    {
      "epoch": 0.26326600479287915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5174,
      "step": 7690
    },
    {
      "epoch": 0.26330023964395755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4985,
      "step": 7691
    },
    {
      "epoch": 0.26333447449503594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5296,
      "step": 7692
    },
    {
      "epoch": 0.26336870934611434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5314,
      "step": 7693
    },
    {
      "epoch": 0.26340294419719273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4997,
      "step": 7694
    },
    {
      "epoch": 0.26343717904827113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5067,
      "step": 7695
    },
    {
      "epoch": 0.2634714138993495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5044,
      "step": 7696
    },
    {
      "epoch": 0.2635056487504279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5285,
      "step": 7697
    },
    {
      "epoch": 0.2635398836015063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.507,
      "step": 7698
    },
    {
      "epoch": 0.2635741184525847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5105,
      "step": 7699
    },
    {
      "epoch": 0.2636083533036631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5269,
      "step": 7700
    },
    {
      "epoch": 0.2636425881547415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5225,
      "step": 7701
    },
    {
      "epoch": 0.2636768230058199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5179,
      "step": 7702
    },
    {
      "epoch": 0.2637110578568983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5151,
      "step": 7703
    },
    {
      "epoch": 0.2637452927079767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5068,
      "step": 7704
    },
    {
      "epoch": 0.2637795275590551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5324,
      "step": 7705
    },
    {
      "epoch": 0.2638137624101335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5308,
      "step": 7706
    },
    {
      "epoch": 0.2638479972612119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.505,
      "step": 7707
    },
    {
      "epoch": 0.26388223211229034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5254,
      "step": 7708
    },
    {
      "epoch": 0.26391646696336873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5093,
      "step": 7709
    },
    {
      "epoch": 0.26395070181444713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5103,
      "step": 7710
    },
    {
      "epoch": 0.2639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5213,
      "step": 7711
    },
    {
      "epoch": 0.2640191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.512,
      "step": 7712
    },
    {
      "epoch": 0.2640534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5109,
      "step": 7713
    },
    {
      "epoch": 0.2640876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5083,
      "step": 7714
    },
    {
      "epoch": 0.2641218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5023,
      "step": 7715
    },
    {
      "epoch": 0.2641561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5224,
      "step": 7716
    },
    {
      "epoch": 0.2641903457719959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5183,
      "step": 7717
    },
    {
      "epoch": 0.2642245806230743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4755,
      "step": 7718
    },
    {
      "epoch": 0.2642588154741527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5096,
      "step": 7719
    },
    {
      "epoch": 0.2642930503252311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5135,
      "step": 7720
    },
    {
      "epoch": 0.2643272851763095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5418,
      "step": 7721
    },
    {
      "epoch": 0.2643615200273879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5205,
      "step": 7722
    },
    {
      "epoch": 0.2643957548784663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5098,
      "step": 7723
    },
    {
      "epoch": 0.2644299897295447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5318,
      "step": 7724
    },
    {
      "epoch": 0.2644642245806231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5137,
      "step": 7725
    },
    {
      "epoch": 0.26449845943170147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5139,
      "step": 7726
    },
    {
      "epoch": 0.26453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5492,
      "step": 7727
    },
    {
      "epoch": 0.26456692913385826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.51,
      "step": 7728
    },
    {
      "epoch": 0.26460116398493666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5078,
      "step": 7729
    },
    {
      "epoch": 0.26463539883601506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5116,
      "step": 7730
    },
    {
      "epoch": 0.26466963368709345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5067,
      "step": 7731
    },
    {
      "epoch": 0.26470386853817185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5137,
      "step": 7732
    },
    {
      "epoch": 0.26473810338925025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5167,
      "step": 7733
    },
    {
      "epoch": 0.26477233824032864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5017,
      "step": 7734
    },
    {
      "epoch": 0.26480657309140704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5268,
      "step": 7735
    },
    {
      "epoch": 0.26484080794248543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5088,
      "step": 7736
    },
    {
      "epoch": 0.26487504279356383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5029,
      "step": 7737
    },
    {
      "epoch": 0.2649092776446422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5141,
      "step": 7738
    },
    {
      "epoch": 0.2649435124957206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4904,
      "step": 7739
    },
    {
      "epoch": 0.264977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5155,
      "step": 7740
    },
    {
      "epoch": 0.2650119821978774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5208,
      "step": 7741
    },
    {
      "epoch": 0.2650462170489558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5094,
      "step": 7742
    },
    {
      "epoch": 0.2650804519000342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4828,
      "step": 7743
    },
    {
      "epoch": 0.26511468675111266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5436,
      "step": 7744
    },
    {
      "epoch": 0.26514892160219106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5002,
      "step": 7745
    },
    {
      "epoch": 0.26518315645326945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.512,
      "step": 7746
    },
    {
      "epoch": 0.26521739130434785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5128,
      "step": 7747
    },
    {
      "epoch": 0.26525162615542625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5029,
      "step": 7748
    },
    {
      "epoch": 0.26528586100650464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5193,
      "step": 7749
    },
    {
      "epoch": 0.26532009585758304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.52,
      "step": 7750
    },
    {
      "epoch": 0.26535433070866143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4907,
      "step": 7751
    },
    {
      "epoch": 0.26538856555973983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5416,
      "step": 7752
    },
    {
      "epoch": 0.2654228004108182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.501,
      "step": 7753
    },
    {
      "epoch": 0.2654570352618966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5334,
      "step": 7754
    },
    {
      "epoch": 0.265491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5053,
      "step": 7755
    },
    {
      "epoch": 0.2655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5137,
      "step": 7756
    },
    {
      "epoch": 0.2655597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5015,
      "step": 7757
    },
    {
      "epoch": 0.2655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5011,
      "step": 7758
    },
    {
      "epoch": 0.2656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5154,
      "step": 7759
    },
    {
      "epoch": 0.265662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5231,
      "step": 7760
    },
    {
      "epoch": 0.2656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5217,
      "step": 7761
    },
    {
      "epoch": 0.2657309140705238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5236,
      "step": 7762
    },
    {
      "epoch": 0.2657651489216022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5202,
      "step": 7763
    },
    {
      "epoch": 0.2657993837726806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.513,
      "step": 7764
    },
    {
      "epoch": 0.265833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5166,
      "step": 7765
    },
    {
      "epoch": 0.2658678534748374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5202,
      "step": 7766
    },
    {
      "epoch": 0.2659020883259158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5084,
      "step": 7767
    },
    {
      "epoch": 0.26593632317699417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5053,
      "step": 7768
    },
    {
      "epoch": 0.26597055802807257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5231,
      "step": 7769
    },
    {
      "epoch": 0.26600479287915096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5061,
      "step": 7770
    },
    {
      "epoch": 0.26603902773022936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5091,
      "step": 7771
    },
    {
      "epoch": 0.26607326258130776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5509,
      "step": 7772
    },
    {
      "epoch": 0.26610749743238615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5068,
      "step": 7773
    },
    {
      "epoch": 0.26614173228346455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5183,
      "step": 7774
    },
    {
      "epoch": 0.26617596713454295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5096,
      "step": 7775
    },
    {
      "epoch": 0.26621020198562134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5235,
      "step": 7776
    },
    {
      "epoch": 0.26624443683669974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5114,
      "step": 7777
    },
    {
      "epoch": 0.26627867168777813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5413,
      "step": 7778
    },
    {
      "epoch": 0.26631290653885653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5316,
      "step": 7779
    },
    {
      "epoch": 0.2663471413899349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5109,
      "step": 7780
    },
    {
      "epoch": 0.2663813762410134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5293,
      "step": 7781
    },
    {
      "epoch": 0.2664156110920918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5242,
      "step": 7782
    },
    {
      "epoch": 0.26644984594317017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5253,
      "step": 7783
    },
    {
      "epoch": 0.26648408079424857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.489,
      "step": 7784
    },
    {
      "epoch": 0.26651831564532696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4791,
      "step": 7785
    },
    {
      "epoch": 0.26655255049640536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5162,
      "step": 7786
    },
    {
      "epoch": 0.26658678534748376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5252,
      "step": 7787
    },
    {
      "epoch": 0.26662102019856215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5267,
      "step": 7788
    },
    {
      "epoch": 0.26665525504964055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5062,
      "step": 7789
    },
    {
      "epoch": 0.26668948990071895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5086,
      "step": 7790
    },
    {
      "epoch": 0.26672372475179734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5054,
      "step": 7791
    },
    {
      "epoch": 0.26675795960287574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5062,
      "step": 7792
    },
    {
      "epoch": 0.26679219445395413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5446,
      "step": 7793
    },
    {
      "epoch": 0.26682642930503253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5172,
      "step": 7794
    },
    {
      "epoch": 0.2668606641561109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5147,
      "step": 7795
    },
    {
      "epoch": 0.2668948990071893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4956,
      "step": 7796
    },
    {
      "epoch": 0.2669291338582677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.518,
      "step": 7797
    },
    {
      "epoch": 0.2669633687093461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5205,
      "step": 7798
    },
    {
      "epoch": 0.2669976035604245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5026,
      "step": 7799
    },
    {
      "epoch": 0.2670318384115029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5012,
      "step": 7800
    },
    {
      "epoch": 0.2670660732625813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5225,
      "step": 7801
    },
    {
      "epoch": 0.2671003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4874,
      "step": 7802
    },
    {
      "epoch": 0.2671345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5148,
      "step": 7803
    },
    {
      "epoch": 0.2671687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5092,
      "step": 7804
    },
    {
      "epoch": 0.2672030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5144,
      "step": 7805
    },
    {
      "epoch": 0.2672372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4949,
      "step": 7806
    },
    {
      "epoch": 0.2672714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5235,
      "step": 7807
    },
    {
      "epoch": 0.2673057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5064,
      "step": 7808
    },
    {
      "epoch": 0.2673399520712085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5157,
      "step": 7809
    },
    {
      "epoch": 0.26737418692228687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5274,
      "step": 7810
    },
    {
      "epoch": 0.26740842177336527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5071,
      "step": 7811
    },
    {
      "epoch": 0.26744265662444366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.514,
      "step": 7812
    },
    {
      "epoch": 0.26747689147552206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5126,
      "step": 7813
    },
    {
      "epoch": 0.26751112632660046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4971,
      "step": 7814
    },
    {
      "epoch": 0.26754536117767885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5156,
      "step": 7815
    },
    {
      "epoch": 0.26757959602875725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.497,
      "step": 7816
    },
    {
      "epoch": 0.26761383087983565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5073,
      "step": 7817
    },
    {
      "epoch": 0.2676480657309141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5139,
      "step": 7818
    },
    {
      "epoch": 0.2676823005819925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5051,
      "step": 7819
    },
    {
      "epoch": 0.2677165354330709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4829,
      "step": 7820
    },
    {
      "epoch": 0.2677507702841493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5178,
      "step": 7821
    },
    {
      "epoch": 0.2677850051352277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5164,
      "step": 7822
    },
    {
      "epoch": 0.2678192399863061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5061,
      "step": 7823
    },
    {
      "epoch": 0.2678534748373845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5152,
      "step": 7824
    },
    {
      "epoch": 0.26788770968846287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4964,
      "step": 7825
    },
    {
      "epoch": 0.26792194453954127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5126,
      "step": 7826
    },
    {
      "epoch": 0.26795617939061966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.536,
      "step": 7827
    },
    {
      "epoch": 0.26799041424169806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4985,
      "step": 7828
    },
    {
      "epoch": 0.26802464909277646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5114,
      "step": 7829
    },
    {
      "epoch": 0.26805888394385485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5089,
      "step": 7830
    },
    {
      "epoch": 0.26809311879493325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5107,
      "step": 7831
    },
    {
      "epoch": 0.26812735364601165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4967,
      "step": 7832
    },
    {
      "epoch": 0.26816158849709004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5087,
      "step": 7833
    },
    {
      "epoch": 0.26819582334816844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4947,
      "step": 7834
    },
    {
      "epoch": 0.26823005819924683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4954,
      "step": 7835
    },
    {
      "epoch": 0.26826429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5247,
      "step": 7836
    },
    {
      "epoch": 0.2682985279014036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5018,
      "step": 7837
    },
    {
      "epoch": 0.268332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4987,
      "step": 7838
    },
    {
      "epoch": 0.2683669976035604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5288,
      "step": 7839
    },
    {
      "epoch": 0.2684012324546388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5025,
      "step": 7840
    },
    {
      "epoch": 0.2684354673057172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.493,
      "step": 7841
    },
    {
      "epoch": 0.2684697021567956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4766,
      "step": 7842
    },
    {
      "epoch": 0.268503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5123,
      "step": 7843
    },
    {
      "epoch": 0.2685381718589524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4824,
      "step": 7844
    },
    {
      "epoch": 0.2685724067100308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5109,
      "step": 7845
    },
    {
      "epoch": 0.2686066415611092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4959,
      "step": 7846
    },
    {
      "epoch": 0.2686408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4992,
      "step": 7847
    },
    {
      "epoch": 0.268675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5361,
      "step": 7848
    },
    {
      "epoch": 0.2687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5111,
      "step": 7849
    },
    {
      "epoch": 0.2687435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4848,
      "step": 7850
    },
    {
      "epoch": 0.2687778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5046,
      "step": 7851
    },
    {
      "epoch": 0.26881205066757957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5008,
      "step": 7852
    },
    {
      "epoch": 0.26884628551865797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5051,
      "step": 7853
    },
    {
      "epoch": 0.26888052036973636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5013,
      "step": 7854
    },
    {
      "epoch": 0.2689147552208148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5009,
      "step": 7855
    },
    {
      "epoch": 0.2689489900718932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5149,
      "step": 7856
    },
    {
      "epoch": 0.2689832249229716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5155,
      "step": 7857
    },
    {
      "epoch": 0.26901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5071,
      "step": 7858
    },
    {
      "epoch": 0.2690516946251284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5128,
      "step": 7859
    },
    {
      "epoch": 0.2690859294762068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5229,
      "step": 7860
    },
    {
      "epoch": 0.2691201643272852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5164,
      "step": 7861
    },
    {
      "epoch": 0.2691543991783636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4949,
      "step": 7862
    },
    {
      "epoch": 0.269188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5115,
      "step": 7863
    },
    {
      "epoch": 0.2692228688805204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4943,
      "step": 7864
    },
    {
      "epoch": 0.2692571037315988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5038,
      "step": 7865
    },
    {
      "epoch": 0.2692913385826772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5026,
      "step": 7866
    },
    {
      "epoch": 0.26932557343375557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5198,
      "step": 7867
    },
    {
      "epoch": 0.26935980828483397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5196,
      "step": 7868
    },
    {
      "epoch": 0.26939404313591236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5081,
      "step": 7869
    },
    {
      "epoch": 0.26942827798699076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5062,
      "step": 7870
    },
    {
      "epoch": 0.26946251283806916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.484,
      "step": 7871
    },
    {
      "epoch": 0.26949674768914755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5062,
      "step": 7872
    },
    {
      "epoch": 0.26953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4995,
      "step": 7873
    },
    {
      "epoch": 0.26956521739130435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5131,
      "step": 7874
    },
    {
      "epoch": 0.26959945224238274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5092,
      "step": 7875
    },
    {
      "epoch": 0.26963368709346114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.499,
      "step": 7876
    },
    {
      "epoch": 0.26966792194453953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.51,
      "step": 7877
    },
    {
      "epoch": 0.26970215679561793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4859,
      "step": 7878
    },
    {
      "epoch": 0.2697363916466963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5142,
      "step": 7879
    },
    {
      "epoch": 0.2697706264977747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5033,
      "step": 7880
    },
    {
      "epoch": 0.2698048613488531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.512,
      "step": 7881
    },
    {
      "epoch": 0.2698390961999315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4924,
      "step": 7882
    },
    {
      "epoch": 0.2698733310510099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4939,
      "step": 7883
    },
    {
      "epoch": 0.2699075659020883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5159,
      "step": 7884
    },
    {
      "epoch": 0.2699418007531667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.501,
      "step": 7885
    },
    {
      "epoch": 0.2699760356042451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5148,
      "step": 7886
    },
    {
      "epoch": 0.2700102704553235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5068,
      "step": 7887
    },
    {
      "epoch": 0.2700445053064019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5053,
      "step": 7888
    },
    {
      "epoch": 0.2700787401574803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5236,
      "step": 7889
    },
    {
      "epoch": 0.2701129750085587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.516,
      "step": 7890
    },
    {
      "epoch": 0.2701472098596371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5014,
      "step": 7891
    },
    {
      "epoch": 0.27018144471071553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5072,
      "step": 7892
    },
    {
      "epoch": 0.27021567956179393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5135,
      "step": 7893
    },
    {
      "epoch": 0.2702499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5155,
      "step": 7894
    },
    {
      "epoch": 0.2702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5051,
      "step": 7895
    },
    {
      "epoch": 0.2703183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.463,
      "step": 7896
    },
    {
      "epoch": 0.2703526189661075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5101,
      "step": 7897
    },
    {
      "epoch": 0.2703868538171859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4982,
      "step": 7898
    },
    {
      "epoch": 0.2704210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5012,
      "step": 7899
    },
    {
      "epoch": 0.2704553235193427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4974,
      "step": 7900
    },
    {
      "epoch": 0.2704895583704211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5059,
      "step": 7901
    },
    {
      "epoch": 0.2705237932214995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5085,
      "step": 7902
    },
    {
      "epoch": 0.2705580280725779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5011,
      "step": 7903
    },
    {
      "epoch": 0.2705922629236563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5087,
      "step": 7904
    },
    {
      "epoch": 0.2706264977747347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5244,
      "step": 7905
    },
    {
      "epoch": 0.2706607326258131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4949,
      "step": 7906
    },
    {
      "epoch": 0.2706949674768915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5161,
      "step": 7907
    },
    {
      "epoch": 0.2707292023279699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5,
      "step": 7908
    },
    {
      "epoch": 0.27076343717904827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4949,
      "step": 7909
    },
    {
      "epoch": 0.27079767203012667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5158,
      "step": 7910
    },
    {
      "epoch": 0.27083190688120506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5211,
      "step": 7911
    },
    {
      "epoch": 0.27086614173228346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5031,
      "step": 7912
    },
    {
      "epoch": 0.27090037658336186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4935,
      "step": 7913
    },
    {
      "epoch": 0.27093461143444025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4885,
      "step": 7914
    },
    {
      "epoch": 0.27096884628551865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5072,
      "step": 7915
    },
    {
      "epoch": 0.27100308113659705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5279,
      "step": 7916
    },
    {
      "epoch": 0.27103731598767544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4805,
      "step": 7917
    },
    {
      "epoch": 0.27107155083875384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4855,
      "step": 7918
    },
    {
      "epoch": 0.27110578568983223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4996,
      "step": 7919
    },
    {
      "epoch": 0.27114002054091063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5129,
      "step": 7920
    },
    {
      "epoch": 0.271174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5008,
      "step": 7921
    },
    {
      "epoch": 0.2712084902430674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5026,
      "step": 7922
    },
    {
      "epoch": 0.2712427250941458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5025,
      "step": 7923
    },
    {
      "epoch": 0.2712769599452242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.504,
      "step": 7924
    },
    {
      "epoch": 0.2713111947963026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4892,
      "step": 7925
    },
    {
      "epoch": 0.271345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5038,
      "step": 7926
    },
    {
      "epoch": 0.2713796644984594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.498,
      "step": 7927
    },
    {
      "epoch": 0.27141389934953786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4965,
      "step": 7928
    },
    {
      "epoch": 0.27144813420061625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.497,
      "step": 7929
    },
    {
      "epoch": 0.27148236905169465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4919,
      "step": 7930
    },
    {
      "epoch": 0.27151660390277305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4926,
      "step": 7931
    },
    {
      "epoch": 0.27155083875385144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4907,
      "step": 7932
    },
    {
      "epoch": 0.27158507360492984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5028,
      "step": 7933
    },
    {
      "epoch": 0.27161930845600823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5068,
      "step": 7934
    },
    {
      "epoch": 0.27165354330708663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5018,
      "step": 7935
    },
    {
      "epoch": 0.271687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5042,
      "step": 7936
    },
    {
      "epoch": 0.2717220130092434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4971,
      "step": 7937
    },
    {
      "epoch": 0.2717562478603218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.499,
      "step": 7938
    },
    {
      "epoch": 0.2717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4886,
      "step": 7939
    },
    {
      "epoch": 0.2718247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5284,
      "step": 7940
    },
    {
      "epoch": 0.271858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5147,
      "step": 7941
    },
    {
      "epoch": 0.2718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4946,
      "step": 7942
    },
    {
      "epoch": 0.2719274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4876,
      "step": 7943
    },
    {
      "epoch": 0.2719616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4975,
      "step": 7944
    },
    {
      "epoch": 0.2719958918178706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5039,
      "step": 7945
    },
    {
      "epoch": 0.272030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5355,
      "step": 7946
    },
    {
      "epoch": 0.2720643615200274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5053,
      "step": 7947
    },
    {
      "epoch": 0.2720985963711058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5103,
      "step": 7948
    },
    {
      "epoch": 0.2721328312221842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5046,
      "step": 7949
    },
    {
      "epoch": 0.2721670660732626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4999,
      "step": 7950
    },
    {
      "epoch": 0.27220130092434097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5163,
      "step": 7951
    },
    {
      "epoch": 0.27223553577541937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4791,
      "step": 7952
    },
    {
      "epoch": 0.27226977062649776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5093,
      "step": 7953
    },
    {
      "epoch": 0.27230400547757616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4859,
      "step": 7954
    },
    {
      "epoch": 0.27233824032865456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4948,
      "step": 7955
    },
    {
      "epoch": 0.27237247517973295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.505,
      "step": 7956
    },
    {
      "epoch": 0.27240671003081135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5263,
      "step": 7957
    },
    {
      "epoch": 0.27244094488188975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4885,
      "step": 7958
    },
    {
      "epoch": 0.27247517973296814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4818,
      "step": 7959
    },
    {
      "epoch": 0.27250941458404654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4905,
      "step": 7960
    },
    {
      "epoch": 0.27254364943512494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4868,
      "step": 7961
    },
    {
      "epoch": 0.27257788428620333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4933,
      "step": 7962
    },
    {
      "epoch": 0.2726121191372817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5057,
      "step": 7963
    },
    {
      "epoch": 0.2726463539883601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5087,
      "step": 7964
    },
    {
      "epoch": 0.2726805888394386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4947,
      "step": 7965
    },
    {
      "epoch": 0.27271482369051697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4962,
      "step": 7966
    },
    {
      "epoch": 0.27274905854159537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5042,
      "step": 7967
    },
    {
      "epoch": 0.27278329339267376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4985,
      "step": 7968
    },
    {
      "epoch": 0.27281752824375216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4957,
      "step": 7969
    },
    {
      "epoch": 0.27285176309483056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5183,
      "step": 7970
    },
    {
      "epoch": 0.27288599794590895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4881,
      "step": 7971
    },
    {
      "epoch": 0.27292023279698735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5064,
      "step": 7972
    },
    {
      "epoch": 0.27295446764806575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5183,
      "step": 7973
    },
    {
      "epoch": 0.27298870249914414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.51,
      "step": 7974
    },
    {
      "epoch": 0.27302293735022254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.508,
      "step": 7975
    },
    {
      "epoch": 0.27305717220130094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4923,
      "step": 7976
    },
    {
      "epoch": 0.27309140705237933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.499,
      "step": 7977
    },
    {
      "epoch": 0.2731256419034577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5043,
      "step": 7978
    },
    {
      "epoch": 0.2731598767545361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.478,
      "step": 7979
    },
    {
      "epoch": 0.2731941116056145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4886,
      "step": 7980
    },
    {
      "epoch": 0.2732283464566929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5191,
      "step": 7981
    },
    {
      "epoch": 0.2732625813077713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4799,
      "step": 7982
    },
    {
      "epoch": 0.2732968161588497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5103,
      "step": 7983
    },
    {
      "epoch": 0.2733310510099281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4984,
      "step": 7984
    },
    {
      "epoch": 0.2733652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5018,
      "step": 7985
    },
    {
      "epoch": 0.2733995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4893,
      "step": 7986
    },
    {
      "epoch": 0.2734337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4877,
      "step": 7987
    },
    {
      "epoch": 0.2734679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5046,
      "step": 7988
    },
    {
      "epoch": 0.2735022252653201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4968,
      "step": 7989
    },
    {
      "epoch": 0.2735364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4982,
      "step": 7990
    },
    {
      "epoch": 0.2735706949674769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.493,
      "step": 7991
    },
    {
      "epoch": 0.2736049298185553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.505,
      "step": 7992
    },
    {
      "epoch": 0.27363916466963367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4988,
      "step": 7993
    },
    {
      "epoch": 0.27367339952071207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4861,
      "step": 7994
    },
    {
      "epoch": 0.27370763437179046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4974,
      "step": 7995
    },
    {
      "epoch": 0.27374186922286886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4756,
      "step": 7996
    },
    {
      "epoch": 0.27377610407394726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4908,
      "step": 7997
    },
    {
      "epoch": 0.27381033892502565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5091,
      "step": 7998
    },
    {
      "epoch": 0.27384457377610405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.508,
      "step": 7999
    },
    {
      "epoch": 0.27387880862718245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4923,
      "step": 8000
    },
    {
      "epoch": 0.27391304347826084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4818,
      "step": 8001
    },
    {
      "epoch": 0.2739472783293393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4921,
      "step": 8002
    },
    {
      "epoch": 0.2739815131804177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4886,
      "step": 8003
    },
    {
      "epoch": 0.2740157480314961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4847,
      "step": 8004
    },
    {
      "epoch": 0.2740499828825745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5078,
      "step": 8005
    },
    {
      "epoch": 0.2740842177336529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.504,
      "step": 8006
    },
    {
      "epoch": 0.2741184525847313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5047,
      "step": 8007
    },
    {
      "epoch": 0.27415268743580967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4901,
      "step": 8008
    },
    {
      "epoch": 0.27418692228688807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5144,
      "step": 8009
    },
    {
      "epoch": 0.27422115713796646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5153,
      "step": 8010
    },
    {
      "epoch": 0.27425539198904486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4813,
      "step": 8011
    },
    {
      "epoch": 0.27428962684012326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4876,
      "step": 8012
    },
    {
      "epoch": 0.27432386169120165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4778,
      "step": 8013
    },
    {
      "epoch": 0.27435809654228005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4846,
      "step": 8014
    },
    {
      "epoch": 0.27439233139335845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5086,
      "step": 8015
    },
    {
      "epoch": 0.27442656624443684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5057,
      "step": 8016
    },
    {
      "epoch": 0.27446080109551524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4966,
      "step": 8017
    },
    {
      "epoch": 0.27449503594659364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5134,
      "step": 8018
    },
    {
      "epoch": 0.27452927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4851,
      "step": 8019
    },
    {
      "epoch": 0.2745635056487504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4861,
      "step": 8020
    },
    {
      "epoch": 0.2745977404998288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4984,
      "step": 8021
    },
    {
      "epoch": 0.2746319753509072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5109,
      "step": 8022
    },
    {
      "epoch": 0.2746662102019856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4965,
      "step": 8023
    },
    {
      "epoch": 0.274700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4713,
      "step": 8024
    },
    {
      "epoch": 0.2747346799041424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5036,
      "step": 8025
    },
    {
      "epoch": 0.2747689147552208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4948,
      "step": 8026
    },
    {
      "epoch": 0.2748031496062992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4872,
      "step": 8027
    },
    {
      "epoch": 0.2748373844573776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4889,
      "step": 8028
    },
    {
      "epoch": 0.274871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4855,
      "step": 8029
    },
    {
      "epoch": 0.2749058541595344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4796,
      "step": 8030
    },
    {
      "epoch": 0.2749400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.493,
      "step": 8031
    },
    {
      "epoch": 0.2749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4868,
      "step": 8032
    },
    {
      "epoch": 0.2750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4963,
      "step": 8033
    },
    {
      "epoch": 0.275042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4847,
      "step": 8034
    },
    {
      "epoch": 0.2750770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5032,
      "step": 8035
    },
    {
      "epoch": 0.27511126326600477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.525,
      "step": 8036
    },
    {
      "epoch": 0.27514549811708316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4782,
      "step": 8037
    },
    {
      "epoch": 0.27517973296816156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4963,
      "step": 8038
    },
    {
      "epoch": 0.27521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4894,
      "step": 8039
    },
    {
      "epoch": 0.2752482026703184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4742,
      "step": 8040
    },
    {
      "epoch": 0.2752824375213968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4999,
      "step": 8041
    },
    {
      "epoch": 0.2753166723724752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5015,
      "step": 8042
    },
    {
      "epoch": 0.2753509072235536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.491,
      "step": 8043
    },
    {
      "epoch": 0.275385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4902,
      "step": 8044
    },
    {
      "epoch": 0.2754193769257104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5041,
      "step": 8045
    },
    {
      "epoch": 0.2754536117767888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4853,
      "step": 8046
    },
    {
      "epoch": 0.2754878466278672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.507,
      "step": 8047
    },
    {
      "epoch": 0.2755220814789456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4912,
      "step": 8048
    },
    {
      "epoch": 0.275556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5011,
      "step": 8049
    },
    {
      "epoch": 0.2755905511811024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4896,
      "step": 8050
    },
    {
      "epoch": 0.27562478603218077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5108,
      "step": 8051
    },
    {
      "epoch": 0.27565902088325916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4907,
      "step": 8052
    },
    {
      "epoch": 0.27569325573433756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4838,
      "step": 8053
    },
    {
      "epoch": 0.27572749058541596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4906,
      "step": 8054
    },
    {
      "epoch": 0.27576172543649435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4822,
      "step": 8055
    },
    {
      "epoch": 0.27579596028757275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4885,
      "step": 8056
    },
    {
      "epoch": 0.27583019513865115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4993,
      "step": 8057
    },
    {
      "epoch": 0.27586442998972954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4932,
      "step": 8058
    },
    {
      "epoch": 0.27589866484080794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4854,
      "step": 8059
    },
    {
      "epoch": 0.27593289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.487,
      "step": 8060
    },
    {
      "epoch": 0.27596713454296473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4759,
      "step": 8061
    },
    {
      "epoch": 0.27600136939404313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4887,
      "step": 8062
    },
    {
      "epoch": 0.2760356042451215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.484,
      "step": 8063
    },
    {
      "epoch": 0.2760698390961999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5001,
      "step": 8064
    },
    {
      "epoch": 0.2761040739472783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4986,
      "step": 8065
    },
    {
      "epoch": 0.2761383087983567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5073,
      "step": 8066
    },
    {
      "epoch": 0.2761725436494351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4796,
      "step": 8067
    },
    {
      "epoch": 0.2762067785005135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4827,
      "step": 8068
    },
    {
      "epoch": 0.2762410133515919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5046,
      "step": 8069
    },
    {
      "epoch": 0.2762752482026703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4951,
      "step": 8070
    },
    {
      "epoch": 0.2763094830537487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4801,
      "step": 8071
    },
    {
      "epoch": 0.2763437179048271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5026,
      "step": 8072
    },
    {
      "epoch": 0.2763779527559055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4817,
      "step": 8073
    },
    {
      "epoch": 0.2764121876069839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5086,
      "step": 8074
    },
    {
      "epoch": 0.2764464224580623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4937,
      "step": 8075
    },
    {
      "epoch": 0.27648065730914073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5169,
      "step": 8076
    },
    {
      "epoch": 0.27651489216021913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4962,
      "step": 8077
    },
    {
      "epoch": 0.2765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4743,
      "step": 8078
    },
    {
      "epoch": 0.2765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.462,
      "step": 8079
    },
    {
      "epoch": 0.2766175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5198,
      "step": 8080
    },
    {
      "epoch": 0.2766518315645327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4943,
      "step": 8081
    },
    {
      "epoch": 0.2766860664156111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4933,
      "step": 8082
    },
    {
      "epoch": 0.2767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.498,
      "step": 8083
    },
    {
      "epoch": 0.2767545361177679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4851,
      "step": 8084
    },
    {
      "epoch": 0.2767887709688463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5171,
      "step": 8085
    },
    {
      "epoch": 0.2768230058199247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4988,
      "step": 8086
    },
    {
      "epoch": 0.2768572406710031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4874,
      "step": 8087
    },
    {
      "epoch": 0.2768914755220815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4896,
      "step": 8088
    },
    {
      "epoch": 0.2769257103731599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4717,
      "step": 8089
    },
    {
      "epoch": 0.2769599452242383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4895,
      "step": 8090
    },
    {
      "epoch": 0.2769941800753167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4987,
      "step": 8091
    },
    {
      "epoch": 0.2770284149263951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4956,
      "step": 8092
    },
    {
      "epoch": 0.27706264977747347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4952,
      "step": 8093
    },
    {
      "epoch": 0.27709688462855186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4982,
      "step": 8094
    },
    {
      "epoch": 0.27713111947963026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4932,
      "step": 8095
    },
    {
      "epoch": 0.27716535433070866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4784,
      "step": 8096
    },
    {
      "epoch": 0.27719958918178705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4881,
      "step": 8097
    },
    {
      "epoch": 0.27723382403286545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4859,
      "step": 8098
    },
    {
      "epoch": 0.27726805888394385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4843,
      "step": 8099
    },
    {
      "epoch": 0.27730229373502224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.504,
      "step": 8100
    },
    {
      "epoch": 0.27733652858610064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4996,
      "step": 8101
    },
    {
      "epoch": 0.27737076343717904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5001,
      "step": 8102
    },
    {
      "epoch": 0.27740499828825743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4754,
      "step": 8103
    },
    {
      "epoch": 0.27743923313933583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4597,
      "step": 8104
    },
    {
      "epoch": 0.2774734679904142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5103,
      "step": 8105
    },
    {
      "epoch": 0.2775077028414926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4769,
      "step": 8106
    },
    {
      "epoch": 0.277541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4938,
      "step": 8107
    },
    {
      "epoch": 0.2775761725436494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4877,
      "step": 8108
    },
    {
      "epoch": 0.2776104073947278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.498,
      "step": 8109
    },
    {
      "epoch": 0.2776446422458062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.501,
      "step": 8110
    },
    {
      "epoch": 0.2776788770968846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4946,
      "step": 8111
    },
    {
      "epoch": 0.27771311194796305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4934,
      "step": 8112
    },
    {
      "epoch": 0.27774734679904145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4794,
      "step": 8113
    },
    {
      "epoch": 0.27778158165011985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4842,
      "step": 8114
    },
    {
      "epoch": 0.27781581650119824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.497,
      "step": 8115
    },
    {
      "epoch": 0.27785005135227664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4915,
      "step": 8116
    },
    {
      "epoch": 0.27788428620335504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4807,
      "step": 8117
    },
    {
      "epoch": 0.27791852105443343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4904,
      "step": 8118
    },
    {
      "epoch": 0.27795275590551183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5211,
      "step": 8119
    },
    {
      "epoch": 0.2779869907565902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5104,
      "step": 8120
    },
    {
      "epoch": 0.2780212256076686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4802,
      "step": 8121
    },
    {
      "epoch": 0.278055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4955,
      "step": 8122
    },
    {
      "epoch": 0.2780896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4932,
      "step": 8123
    },
    {
      "epoch": 0.2781239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4874,
      "step": 8124
    },
    {
      "epoch": 0.2781581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4733,
      "step": 8125
    },
    {
      "epoch": 0.2781923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4877,
      "step": 8126
    },
    {
      "epoch": 0.278226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4922,
      "step": 8127
    },
    {
      "epoch": 0.2782608695652174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5,
      "step": 8128
    },
    {
      "epoch": 0.2782951044162958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4855,
      "step": 8129
    },
    {
      "epoch": 0.2783293392673742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4824,
      "step": 8130
    },
    {
      "epoch": 0.2783635741184526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4793,
      "step": 8131
    },
    {
      "epoch": 0.278397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4813,
      "step": 8132
    },
    {
      "epoch": 0.2784320438206094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5048,
      "step": 8133
    },
    {
      "epoch": 0.2784662786716878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4737,
      "step": 8134
    },
    {
      "epoch": 0.27850051352276617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.477,
      "step": 8135
    },
    {
      "epoch": 0.27853474837384457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4839,
      "step": 8136
    },
    {
      "epoch": 0.27856898322492296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4989,
      "step": 8137
    },
    {
      "epoch": 0.27860321807600136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5009,
      "step": 8138
    },
    {
      "epoch": 0.27863745292707975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4791,
      "step": 8139
    },
    {
      "epoch": 0.27867168777815815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5134,
      "step": 8140
    },
    {
      "epoch": 0.27870592262923655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4783,
      "step": 8141
    },
    {
      "epoch": 0.27874015748031494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5101,
      "step": 8142
    },
    {
      "epoch": 0.27877439233139334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4974,
      "step": 8143
    },
    {
      "epoch": 0.27880862718247174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4985,
      "step": 8144
    },
    {
      "epoch": 0.27884286203355013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4689,
      "step": 8145
    },
    {
      "epoch": 0.27887709688462853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5087,
      "step": 8146
    },
    {
      "epoch": 0.2789113317357069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5086,
      "step": 8147
    },
    {
      "epoch": 0.2789455665867853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5086,
      "step": 8148
    },
    {
      "epoch": 0.2789798014378638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4819,
      "step": 8149
    },
    {
      "epoch": 0.27901403628894217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4837,
      "step": 8150
    },
    {
      "epoch": 0.27904827114002057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4829,
      "step": 8151
    },
    {
      "epoch": 0.27908250599109896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5007,
      "step": 8152
    },
    {
      "epoch": 0.27911674084217736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4944,
      "step": 8153
    },
    {
      "epoch": 0.27915097569325575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5025,
      "step": 8154
    },
    {
      "epoch": 0.27918521054433415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4884,
      "step": 8155
    },
    {
      "epoch": 0.27921944539541255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4774,
      "step": 8156
    },
    {
      "epoch": 0.27925368024649094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4773,
      "step": 8157
    },
    {
      "epoch": 0.27928791509756934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4833,
      "step": 8158
    },
    {
      "epoch": 0.27932214994864774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4951,
      "step": 8159
    },
    {
      "epoch": 0.27935638479972613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4863,
      "step": 8160
    },
    {
      "epoch": 0.27939061965080453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4946,
      "step": 8161
    },
    {
      "epoch": 0.2794248545018829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4862,
      "step": 8162
    },
    {
      "epoch": 0.2794590893529613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4724,
      "step": 8163
    },
    {
      "epoch": 0.2794933242040397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 8164
    },
    {
      "epoch": 0.2795275590551181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.501,
      "step": 8165
    },
    {
      "epoch": 0.2795617939061965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5047,
      "step": 8166
    },
    {
      "epoch": 0.2795960287572749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.496,
      "step": 8167
    },
    {
      "epoch": 0.2796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4963,
      "step": 8168
    },
    {
      "epoch": 0.2796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4857,
      "step": 8169
    },
    {
      "epoch": 0.2796987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4835,
      "step": 8170
    },
    {
      "epoch": 0.2797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5092,
      "step": 8171
    },
    {
      "epoch": 0.2797672030126669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.476,
      "step": 8172
    },
    {
      "epoch": 0.2798014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.493,
      "step": 8173
    },
    {
      "epoch": 0.2798356727148237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4912,
      "step": 8174
    },
    {
      "epoch": 0.2798699075659021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4849,
      "step": 8175
    },
    {
      "epoch": 0.2799041424169805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4763,
      "step": 8176
    },
    {
      "epoch": 0.27993837726805887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4897,
      "step": 8177
    },
    {
      "epoch": 0.27997261211913727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4845,
      "step": 8178
    },
    {
      "epoch": 0.28000684697021566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4873,
      "step": 8179
    },
    {
      "epoch": 0.28004108182129406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4841,
      "step": 8180
    },
    {
      "epoch": 0.28007531667237245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.502,
      "step": 8181
    },
    {
      "epoch": 0.28010955152345085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4829,
      "step": 8182
    },
    {
      "epoch": 0.28014378637452925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4976,
      "step": 8183
    },
    {
      "epoch": 0.28017802122560764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4769,
      "step": 8184
    },
    {
      "epoch": 0.28021225607668604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4893,
      "step": 8185
    },
    {
      "epoch": 0.2802464909277645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4817,
      "step": 8186
    },
    {
      "epoch": 0.2802807257788429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.481,
      "step": 8187
    },
    {
      "epoch": 0.2803149606299213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5032,
      "step": 8188
    },
    {
      "epoch": 0.2803491954809997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4927,
      "step": 8189
    },
    {
      "epoch": 0.2803834303320781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.449,
      "step": 8190
    },
    {
      "epoch": 0.2804176651831565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4816,
      "step": 8191
    },
    {
      "epoch": 0.28045190003423487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4951,
      "step": 8192
    },
    {
      "epoch": 0.28048613488531327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4785,
      "step": 8193
    },
    {
      "epoch": 0.28052036973639166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4826,
      "step": 8194
    },
    {
      "epoch": 0.28055460458747006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4858,
      "step": 8195
    },
    {
      "epoch": 0.28058883943854845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4794,
      "step": 8196
    },
    {
      "epoch": 0.28062307428962685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4912,
      "step": 8197
    },
    {
      "epoch": 0.28065730914070525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4801,
      "step": 8198
    },
    {
      "epoch": 0.28069154399178364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4691,
      "step": 8199
    },
    {
      "epoch": 0.28072577884286204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5082,
      "step": 8200
    },
    {
      "epoch": 0.28076001369394044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5038,
      "step": 8201
    },
    {
      "epoch": 0.28079424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4958,
      "step": 8202
    },
    {
      "epoch": 0.28082848339609723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4668,
      "step": 8203
    },
    {
      "epoch": 0.2808627182471756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4744,
      "step": 8204
    },
    {
      "epoch": 0.280896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4877,
      "step": 8205
    },
    {
      "epoch": 0.2809311879493324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4772,
      "step": 8206
    },
    {
      "epoch": 0.2809654228004108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4893,
      "step": 8207
    },
    {
      "epoch": 0.2809996576514892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4945,
      "step": 8208
    },
    {
      "epoch": 0.2810338925025676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4612,
      "step": 8209
    },
    {
      "epoch": 0.281068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4794,
      "step": 8210
    },
    {
      "epoch": 0.2811023622047244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4714,
      "step": 8211
    },
    {
      "epoch": 0.2811365970558028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5063,
      "step": 8212
    },
    {
      "epoch": 0.2811708319068812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5136,
      "step": 8213
    },
    {
      "epoch": 0.2812050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4779,
      "step": 8214
    },
    {
      "epoch": 0.281239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.501,
      "step": 8215
    },
    {
      "epoch": 0.2812735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4582,
      "step": 8216
    },
    {
      "epoch": 0.2813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4907,
      "step": 8217
    },
    {
      "epoch": 0.2813420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4745,
      "step": 8218
    },
    {
      "epoch": 0.28137624101335157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4822,
      "step": 8219
    },
    {
      "epoch": 0.28141047586442997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4709,
      "step": 8220
    },
    {
      "epoch": 0.28144471071550836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4737,
      "step": 8221
    },
    {
      "epoch": 0.28147894556658676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.472,
      "step": 8222
    },
    {
      "epoch": 0.2815131804176652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4873,
      "step": 8223
    },
    {
      "epoch": 0.2815474152687436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4685,
      "step": 8224
    },
    {
      "epoch": 0.281581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4696,
      "step": 8225
    },
    {
      "epoch": 0.2816158849709004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4826,
      "step": 8226
    },
    {
      "epoch": 0.2816501198219788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5064,
      "step": 8227
    },
    {
      "epoch": 0.2816843546730572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4712,
      "step": 8228
    },
    {
      "epoch": 0.2817185895241356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4684,
      "step": 8229
    },
    {
      "epoch": 0.281752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.497,
      "step": 8230
    },
    {
      "epoch": 0.2817870592262924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4812,
      "step": 8231
    },
    {
      "epoch": 0.2818212940773708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4711,
      "step": 8232
    },
    {
      "epoch": 0.2818555289284492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5108,
      "step": 8233
    },
    {
      "epoch": 0.28188976377952757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4675,
      "step": 8234
    },
    {
      "epoch": 0.28192399863060597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 8235
    },
    {
      "epoch": 0.28195823348168436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4714,
      "step": 8236
    },
    {
      "epoch": 0.28199246833276276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.465,
      "step": 8237
    },
    {
      "epoch": 0.28202670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4776,
      "step": 8238
    },
    {
      "epoch": 0.28206093803491955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4677,
      "step": 8239
    },
    {
      "epoch": 0.28209517288599795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4707,
      "step": 8240
    },
    {
      "epoch": 0.28212940773707634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4781,
      "step": 8241
    },
    {
      "epoch": 0.28216364258815474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4711,
      "step": 8242
    },
    {
      "epoch": 0.28219787743923314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4905,
      "step": 8243
    },
    {
      "epoch": 0.28223211229031153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.488,
      "step": 8244
    },
    {
      "epoch": 0.28226634714138993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4948,
      "step": 8245
    },
    {
      "epoch": 0.2823005819924683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4879,
      "step": 8246
    },
    {
      "epoch": 0.2823348168435467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4617,
      "step": 8247
    },
    {
      "epoch": 0.2823690516946251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4973,
      "step": 8248
    },
    {
      "epoch": 0.2824032865457035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.488,
      "step": 8249
    },
    {
      "epoch": 0.2824375213967819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4841,
      "step": 8250
    },
    {
      "epoch": 0.2824717562478603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4728,
      "step": 8251
    },
    {
      "epoch": 0.2825059910989387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4813,
      "step": 8252
    },
    {
      "epoch": 0.2825402259500171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4888,
      "step": 8253
    },
    {
      "epoch": 0.2825744608010955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4797,
      "step": 8254
    },
    {
      "epoch": 0.2826086956521739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.466,
      "step": 8255
    },
    {
      "epoch": 0.2826429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4923,
      "step": 8256
    },
    {
      "epoch": 0.2826771653543307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4813,
      "step": 8257
    },
    {
      "epoch": 0.2827114002054091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4793,
      "step": 8258
    },
    {
      "epoch": 0.28274563505648753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4778,
      "step": 8259
    },
    {
      "epoch": 0.28277986990756593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4837,
      "step": 8260
    },
    {
      "epoch": 0.2828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4907,
      "step": 8261
    },
    {
      "epoch": 0.2828483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4809,
      "step": 8262
    },
    {
      "epoch": 0.2828825744608011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4846,
      "step": 8263
    },
    {
      "epoch": 0.2829168093118795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4832,
      "step": 8264
    },
    {
      "epoch": 0.2829510441629579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5013,
      "step": 8265
    },
    {
      "epoch": 0.2829852790140363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.467,
      "step": 8266
    },
    {
      "epoch": 0.2830195138651147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4741,
      "step": 8267
    },
    {
      "epoch": 0.2830537487161931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4802,
      "step": 8268
    },
    {
      "epoch": 0.2830879835672715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4768,
      "step": 8269
    },
    {
      "epoch": 0.2831222184183499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4847,
      "step": 8270
    },
    {
      "epoch": 0.2831564532694283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4903,
      "step": 8271
    },
    {
      "epoch": 0.2831906881205067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.486,
      "step": 8272
    },
    {
      "epoch": 0.2832249229715851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4528,
      "step": 8273
    },
    {
      "epoch": 0.2832591578226635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4977,
      "step": 8274
    },
    {
      "epoch": 0.2832933926737419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4742,
      "step": 8275
    },
    {
      "epoch": 0.28332762752482027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4736,
      "step": 8276
    },
    {
      "epoch": 0.28336186237589867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4859,
      "step": 8277
    },
    {
      "epoch": 0.28339609722697706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4954,
      "step": 8278
    },
    {
      "epoch": 0.28343033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5007,
      "step": 8279
    },
    {
      "epoch": 0.28346456692913385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4946,
      "step": 8280
    },
    {
      "epoch": 0.28349880178021225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4885,
      "step": 8281
    },
    {
      "epoch": 0.28353303663129065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4915,
      "step": 8282
    },
    {
      "epoch": 0.28356727148236904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4622,
      "step": 8283
    },
    {
      "epoch": 0.28360150633344744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4754,
      "step": 8284
    },
    {
      "epoch": 0.28363574118452584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4716,
      "step": 8285
    },
    {
      "epoch": 0.28366997603560423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4876,
      "step": 8286
    },
    {
      "epoch": 0.28370421088668263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4728,
      "step": 8287
    },
    {
      "epoch": 0.283738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5014,
      "step": 8288
    },
    {
      "epoch": 0.2837726805888394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4719,
      "step": 8289
    },
    {
      "epoch": 0.2838069154399178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4858,
      "step": 8290
    },
    {
      "epoch": 0.2838411502909962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4773,
      "step": 8291
    },
    {
      "epoch": 0.2838753851420746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4637,
      "step": 8292
    },
    {
      "epoch": 0.283909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.477,
      "step": 8293
    },
    {
      "epoch": 0.2839438548442314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4836,
      "step": 8294
    },
    {
      "epoch": 0.2839780896953098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4757,
      "step": 8295
    },
    {
      "epoch": 0.28401232454638825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.483,
      "step": 8296
    },
    {
      "epoch": 0.28404655939746665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4819,
      "step": 8297
    },
    {
      "epoch": 0.28408079424854504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4542,
      "step": 8298
    },
    {
      "epoch": 0.28411502909962344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4913,
      "step": 8299
    },
    {
      "epoch": 0.28414926395070184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.477,
      "step": 8300
    },
    {
      "epoch": 0.28418349880178023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4794,
      "step": 8301
    },
    {
      "epoch": 0.28421773365285863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4735,
      "step": 8302
    },
    {
      "epoch": 0.284251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4732,
      "step": 8303
    },
    {
      "epoch": 0.2842862033550154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4932,
      "step": 8304
    },
    {
      "epoch": 0.2843204382060938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4639,
      "step": 8305
    },
    {
      "epoch": 0.2843546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4791,
      "step": 8306
    },
    {
      "epoch": 0.2843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 8307
    },
    {
      "epoch": 0.284423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4871,
      "step": 8308
    },
    {
      "epoch": 0.2844573776104074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4774,
      "step": 8309
    },
    {
      "epoch": 0.2844916124614858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4746,
      "step": 8310
    },
    {
      "epoch": 0.2845258473125642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4788,
      "step": 8311
    },
    {
      "epoch": 0.2845600821636426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4781,
      "step": 8312
    },
    {
      "epoch": 0.284594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4739,
      "step": 8313
    },
    {
      "epoch": 0.2846285518657994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4838,
      "step": 8314
    },
    {
      "epoch": 0.2846627867168778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4995,
      "step": 8315
    },
    {
      "epoch": 0.2846970215679562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4918,
      "step": 8316
    },
    {
      "epoch": 0.2847312564190346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4644,
      "step": 8317
    },
    {
      "epoch": 0.28476549127011297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4844,
      "step": 8318
    },
    {
      "epoch": 0.28479972612119137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5004,
      "step": 8319
    },
    {
      "epoch": 0.28483396097226976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4831,
      "step": 8320
    },
    {
      "epoch": 0.28486819582334816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4805,
      "step": 8321
    },
    {
      "epoch": 0.28490243067442655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4793,
      "step": 8322
    },
    {
      "epoch": 0.28493666552550495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4766,
      "step": 8323
    },
    {
      "epoch": 0.28497090037658335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4688,
      "step": 8324
    },
    {
      "epoch": 0.28500513522766174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4716,
      "step": 8325
    },
    {
      "epoch": 0.28503937007874014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4986,
      "step": 8326
    },
    {
      "epoch": 0.28507360492981854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4765,
      "step": 8327
    },
    {
      "epoch": 0.28510783978089693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4576,
      "step": 8328
    },
    {
      "epoch": 0.28514207463197533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4679,
      "step": 8329
    },
    {
      "epoch": 0.2851763094830537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.473,
      "step": 8330
    },
    {
      "epoch": 0.2852105443341321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4641,
      "step": 8331
    },
    {
      "epoch": 0.2852447791852105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4825,
      "step": 8332
    },
    {
      "epoch": 0.28527901403628897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4849,
      "step": 8333
    },
    {
      "epoch": 0.28531324888736737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4773,
      "step": 8334
    },
    {
      "epoch": 0.28534748373844576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4874,
      "step": 8335
    },
    {
      "epoch": 0.28538171858952416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4696,
      "step": 8336
    },
    {
      "epoch": 0.28541595344060255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4562,
      "step": 8337
    },
    {
      "epoch": 0.28545018829168095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4934,
      "step": 8338
    },
    {
      "epoch": 0.28548442314275935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4851,
      "step": 8339
    },
    {
      "epoch": 0.28551865799383774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4653,
      "step": 8340
    },
    {
      "epoch": 0.28555289284491614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4985,
      "step": 8341
    },
    {
      "epoch": 0.28558712769599454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4645,
      "step": 8342
    },
    {
      "epoch": 0.28562136254707293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.455,
      "step": 8343
    },
    {
      "epoch": 0.28565559739815133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4813,
      "step": 8344
    },
    {
      "epoch": 0.2856898322492297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4543,
      "step": 8345
    },
    {
      "epoch": 0.2857240671003081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4744,
      "step": 8346
    },
    {
      "epoch": 0.2857583019513865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.463,
      "step": 8347
    },
    {
      "epoch": 0.2857925368024649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.48,
      "step": 8348
    },
    {
      "epoch": 0.2858267716535433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4743,
      "step": 8349
    },
    {
      "epoch": 0.2858610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4811,
      "step": 8350
    },
    {
      "epoch": 0.2858952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4927,
      "step": 8351
    },
    {
      "epoch": 0.2859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4837,
      "step": 8352
    },
    {
      "epoch": 0.2859637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4365,
      "step": 8353
    },
    {
      "epoch": 0.2859979459089353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4813,
      "step": 8354
    },
    {
      "epoch": 0.2860321807600137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4849,
      "step": 8355
    },
    {
      "epoch": 0.2860664156110921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4874,
      "step": 8356
    },
    {
      "epoch": 0.2861006504621705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4811,
      "step": 8357
    },
    {
      "epoch": 0.2861348853132489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4731,
      "step": 8358
    },
    {
      "epoch": 0.2861691201643273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4868,
      "step": 8359
    },
    {
      "epoch": 0.28620335501540567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4741,
      "step": 8360
    },
    {
      "epoch": 0.28623758986648407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4631,
      "step": 8361
    },
    {
      "epoch": 0.28627182471756246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4912,
      "step": 8362
    },
    {
      "epoch": 0.28630605956864086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4639,
      "step": 8363
    },
    {
      "epoch": 0.28634029441971925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4741,
      "step": 8364
    },
    {
      "epoch": 0.28637452927079765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4586,
      "step": 8365
    },
    {
      "epoch": 0.28640876412187605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4586,
      "step": 8366
    },
    {
      "epoch": 0.28644299897295444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4703,
      "step": 8367
    },
    {
      "epoch": 0.28647723382403284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4986,
      "step": 8368
    },
    {
      "epoch": 0.28651146867511124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.463,
      "step": 8369
    },
    {
      "epoch": 0.2865457035261897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.465,
      "step": 8370
    },
    {
      "epoch": 0.2865799383772681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4885,
      "step": 8371
    },
    {
      "epoch": 0.2866141732283465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4643,
      "step": 8372
    },
    {
      "epoch": 0.2866484080794249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4574,
      "step": 8373
    },
    {
      "epoch": 0.2866826429305033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4903,
      "step": 8374
    },
    {
      "epoch": 0.28671687778158167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4679,
      "step": 8375
    },
    {
      "epoch": 0.28675111263266007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4765,
      "step": 8376
    },
    {
      "epoch": 0.28678534748373846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4878,
      "step": 8377
    },
    {
      "epoch": 0.28681958233481686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4848,
      "step": 8378
    },
    {
      "epoch": 0.28685381718589525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4741,
      "step": 8379
    },
    {
      "epoch": 0.28688805203697365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.476,
      "step": 8380
    },
    {
      "epoch": 0.28692228688805205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4943,
      "step": 8381
    },
    {
      "epoch": 0.28695652173913044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4735,
      "step": 8382
    },
    {
      "epoch": 0.28699075659020884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4909,
      "step": 8383
    },
    {
      "epoch": 0.28702499144128724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4753,
      "step": 8384
    },
    {
      "epoch": 0.28705922629236563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4891,
      "step": 8385
    },
    {
      "epoch": 0.28709346114344403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4824,
      "step": 8386
    },
    {
      "epoch": 0.2871276959945224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.478,
      "step": 8387
    },
    {
      "epoch": 0.2871619308456008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4774,
      "step": 8388
    },
    {
      "epoch": 0.2871961656966792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4698,
      "step": 8389
    },
    {
      "epoch": 0.2872304005477576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4861,
      "step": 8390
    },
    {
      "epoch": 0.287264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4891,
      "step": 8391
    },
    {
      "epoch": 0.2872988702499144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4652,
      "step": 8392
    },
    {
      "epoch": 0.2873331051009928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4602,
      "step": 8393
    },
    {
      "epoch": 0.2873673399520712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4606,
      "step": 8394
    },
    {
      "epoch": 0.2874015748031496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4988,
      "step": 8395
    },
    {
      "epoch": 0.287435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4787,
      "step": 8396
    },
    {
      "epoch": 0.2874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.464,
      "step": 8397
    },
    {
      "epoch": 0.2875042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4705,
      "step": 8398
    },
    {
      "epoch": 0.2875385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4765,
      "step": 8399
    },
    {
      "epoch": 0.2875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4533,
      "step": 8400
    },
    {
      "epoch": 0.28760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4762,
      "step": 8401
    },
    {
      "epoch": 0.28764121876069837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4572,
      "step": 8402
    },
    {
      "epoch": 0.28767545361177677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4941,
      "step": 8403
    },
    {
      "epoch": 0.28770968846285516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4762,
      "step": 8404
    },
    {
      "epoch": 0.28774392331393356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4862,
      "step": 8405
    },
    {
      "epoch": 0.28777815816501195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4699,
      "step": 8406
    },
    {
      "epoch": 0.2878123930160904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4569,
      "step": 8407
    },
    {
      "epoch": 0.2878466278671688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5022,
      "step": 8408
    },
    {
      "epoch": 0.2878808627182472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 8409
    },
    {
      "epoch": 0.2879150975693256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4611,
      "step": 8410
    },
    {
      "epoch": 0.287949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4529,
      "step": 8411
    },
    {
      "epoch": 0.2879835672714824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4608,
      "step": 8412
    },
    {
      "epoch": 0.2880178021225608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4709,
      "step": 8413
    },
    {
      "epoch": 0.2880520369736392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.457,
      "step": 8414
    },
    {
      "epoch": 0.2880862718247176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4503,
      "step": 8415
    },
    {
      "epoch": 0.288120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4901,
      "step": 8416
    },
    {
      "epoch": 0.28815474152687437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.481,
      "step": 8417
    },
    {
      "epoch": 0.28818897637795277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4466,
      "step": 8418
    },
    {
      "epoch": 0.28822321122903116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4845,
      "step": 8419
    },
    {
      "epoch": 0.28825744608010956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4741,
      "step": 8420
    },
    {
      "epoch": 0.28829168093118795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4706,
      "step": 8421
    },
    {
      "epoch": 0.28832591578226635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4617,
      "step": 8422
    },
    {
      "epoch": 0.28836015063334475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4692,
      "step": 8423
    },
    {
      "epoch": 0.28839438548442314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4789,
      "step": 8424
    },
    {
      "epoch": 0.28842862033550154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4818,
      "step": 8425
    },
    {
      "epoch": 0.28846285518657994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4786,
      "step": 8426
    },
    {
      "epoch": 0.28849709003765833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4509,
      "step": 8427
    },
    {
      "epoch": 0.28853132488873673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4529,
      "step": 8428
    },
    {
      "epoch": 0.2885655597398151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.469,
      "step": 8429
    },
    {
      "epoch": 0.2885997945908935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.462,
      "step": 8430
    },
    {
      "epoch": 0.2886340294419719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4678,
      "step": 8431
    },
    {
      "epoch": 0.2886682642930503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.473,
      "step": 8432
    },
    {
      "epoch": 0.2887024991441287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 8433
    },
    {
      "epoch": 0.2887367339952071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4671,
      "step": 8434
    },
    {
      "epoch": 0.2887709688462855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4617,
      "step": 8435
    },
    {
      "epoch": 0.2888052036973639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4696,
      "step": 8436
    },
    {
      "epoch": 0.2888394385484423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4927,
      "step": 8437
    },
    {
      "epoch": 0.2888736733995207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4829,
      "step": 8438
    },
    {
      "epoch": 0.2889079082505991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4767,
      "step": 8439
    },
    {
      "epoch": 0.2889421431016775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4658,
      "step": 8440
    },
    {
      "epoch": 0.2889763779527559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4833,
      "step": 8441
    },
    {
      "epoch": 0.2890106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4697,
      "step": 8442
    },
    {
      "epoch": 0.28904484765491273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4858,
      "step": 8443
    },
    {
      "epoch": 0.2890790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4845,
      "step": 8444
    },
    {
      "epoch": 0.2891133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4806,
      "step": 8445
    },
    {
      "epoch": 0.2891475522081479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4669,
      "step": 8446
    },
    {
      "epoch": 0.2891817870592263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.463,
      "step": 8447
    },
    {
      "epoch": 0.2892160219103047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4804,
      "step": 8448
    },
    {
      "epoch": 0.2892502567613831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4729,
      "step": 8449
    },
    {
      "epoch": 0.2892844916124615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4732,
      "step": 8450
    },
    {
      "epoch": 0.2893187264635399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4549,
      "step": 8451
    },
    {
      "epoch": 0.2893529613146183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4641,
      "step": 8452
    },
    {
      "epoch": 0.2893871961656967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.48,
      "step": 8453
    },
    {
      "epoch": 0.2894214310167751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4695,
      "step": 8454
    },
    {
      "epoch": 0.2894556658678535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4553,
      "step": 8455
    },
    {
      "epoch": 0.2894899007189319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4804,
      "step": 8456
    },
    {
      "epoch": 0.2895241355700103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4506,
      "step": 8457
    },
    {
      "epoch": 0.2895583704210887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4743,
      "step": 8458
    },
    {
      "epoch": 0.28959260527216707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4692,
      "step": 8459
    },
    {
      "epoch": 0.28962684012324547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.471,
      "step": 8460
    },
    {
      "epoch": 0.28966107497432386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4797,
      "step": 8461
    },
    {
      "epoch": 0.28969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4722,
      "step": 8462
    },
    {
      "epoch": 0.28972954467648065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4556,
      "step": 8463
    },
    {
      "epoch": 0.28976377952755905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4733,
      "step": 8464
    },
    {
      "epoch": 0.28979801437863745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.489,
      "step": 8465
    },
    {
      "epoch": 0.28983224922971584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4782,
      "step": 8466
    },
    {
      "epoch": 0.28986648408079424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4566,
      "step": 8467
    },
    {
      "epoch": 0.28990071893187264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4887,
      "step": 8468
    },
    {
      "epoch": 0.28993495378295103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4752,
      "step": 8469
    },
    {
      "epoch": 0.28996918863402943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4762,
      "step": 8470
    },
    {
      "epoch": 0.2900034234851078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4833,
      "step": 8471
    },
    {
      "epoch": 0.2900376583361862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4823,
      "step": 8472
    },
    {
      "epoch": 0.2900718931872646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.478,
      "step": 8473
    },
    {
      "epoch": 0.290106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.46,
      "step": 8474
    },
    {
      "epoch": 0.2901403628894214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4676,
      "step": 8475
    },
    {
      "epoch": 0.2901745977404998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4653,
      "step": 8476
    },
    {
      "epoch": 0.2902088325915782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4785,
      "step": 8477
    },
    {
      "epoch": 0.2902430674426566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4645,
      "step": 8478
    },
    {
      "epoch": 0.290277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4563,
      "step": 8479
    },
    {
      "epoch": 0.29031153714481345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4615,
      "step": 8480
    },
    {
      "epoch": 0.29034577199589184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4643,
      "step": 8481
    },
    {
      "epoch": 0.29038000684697024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4653,
      "step": 8482
    },
    {
      "epoch": 0.29041424169804864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4685,
      "step": 8483
    },
    {
      "epoch": 0.29044847654912703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 8484
    },
    {
      "epoch": 0.29048271140020543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4768,
      "step": 8485
    },
    {
      "epoch": 0.2905169462512838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4779,
      "step": 8486
    },
    {
      "epoch": 0.2905511811023622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4678,
      "step": 8487
    },
    {
      "epoch": 0.2905854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4786,
      "step": 8488
    },
    {
      "epoch": 0.290619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.468,
      "step": 8489
    },
    {
      "epoch": 0.2906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4688,
      "step": 8490
    },
    {
      "epoch": 0.2906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4568,
      "step": 8491
    },
    {
      "epoch": 0.2907223553577542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4752,
      "step": 8492
    },
    {
      "epoch": 0.2907565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4599,
      "step": 8493
    },
    {
      "epoch": 0.290790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4669,
      "step": 8494
    },
    {
      "epoch": 0.2908250599109894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4704,
      "step": 8495
    },
    {
      "epoch": 0.2908592947620678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.466,
      "step": 8496
    },
    {
      "epoch": 0.2908935296131462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4565,
      "step": 8497
    },
    {
      "epoch": 0.2909277644642246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4721,
      "step": 8498
    },
    {
      "epoch": 0.290961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4557,
      "step": 8499
    },
    {
      "epoch": 0.2909962341663814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4765,
      "step": 8500
    },
    {
      "epoch": 0.29103046901745977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4777,
      "step": 8501
    },
    {
      "epoch": 0.29106470386853817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4999,
      "step": 8502
    },
    {
      "epoch": 0.29109893871961656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4615,
      "step": 8503
    },
    {
      "epoch": 0.29113317357069496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4646,
      "step": 8504
    },
    {
      "epoch": 0.29116740842177335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4595,
      "step": 8505
    },
    {
      "epoch": 0.29120164327285175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4652,
      "step": 8506
    },
    {
      "epoch": 0.29123587812393015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4476,
      "step": 8507
    },
    {
      "epoch": 0.29127011297500854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4586,
      "step": 8508
    },
    {
      "epoch": 0.29130434782608694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4618,
      "step": 8509
    },
    {
      "epoch": 0.29133858267716534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4692,
      "step": 8510
    },
    {
      "epoch": 0.29137281752824373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4631,
      "step": 8511
    },
    {
      "epoch": 0.29140705237932213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4793,
      "step": 8512
    },
    {
      "epoch": 0.2914412872304005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4642,
      "step": 8513
    },
    {
      "epoch": 0.2914755220814789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.466,
      "step": 8514
    },
    {
      "epoch": 0.2915097569325573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4696,
      "step": 8515
    },
    {
      "epoch": 0.2915439917836357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4646,
      "step": 8516
    },
    {
      "epoch": 0.29157822663471417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4492,
      "step": 8517
    },
    {
      "epoch": 0.29161246148579256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4465,
      "step": 8518
    },
    {
      "epoch": 0.29164669633687096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4622,
      "step": 8519
    },
    {
      "epoch": 0.29168093118794935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4661,
      "step": 8520
    },
    {
      "epoch": 0.29171516603902775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4497,
      "step": 8521
    },
    {
      "epoch": 0.29174940089010615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.458,
      "step": 8522
    },
    {
      "epoch": 0.29178363574118454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.442,
      "step": 8523
    },
    {
      "epoch": 0.29181787059226294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4596,
      "step": 8524
    },
    {
      "epoch": 0.29185210544334134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4676,
      "step": 8525
    },
    {
      "epoch": 0.29188634029441973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4583,
      "step": 8526
    },
    {
      "epoch": 0.29192057514549813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4625,
      "step": 8527
    },
    {
      "epoch": 0.2919548099965765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4374,
      "step": 8528
    },
    {
      "epoch": 0.2919890448476549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4773,
      "step": 8529
    },
    {
      "epoch": 0.2920232796987333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4718,
      "step": 8530
    },
    {
      "epoch": 0.2920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4564,
      "step": 8531
    },
    {
      "epoch": 0.2920917494008901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4613,
      "step": 8532
    },
    {
      "epoch": 0.2921259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4565,
      "step": 8533
    },
    {
      "epoch": 0.2921602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4597,
      "step": 8534
    },
    {
      "epoch": 0.2921944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4918,
      "step": 8535
    },
    {
      "epoch": 0.2922286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4609,
      "step": 8536
    },
    {
      "epoch": 0.2922629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.462,
      "step": 8537
    },
    {
      "epoch": 0.2922971585073605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4587,
      "step": 8538
    },
    {
      "epoch": 0.2923313933584389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4636,
      "step": 8539
    },
    {
      "epoch": 0.2923656282095173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4639,
      "step": 8540
    },
    {
      "epoch": 0.2923998630605957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4666,
      "step": 8541
    },
    {
      "epoch": 0.2924340979116741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4508,
      "step": 8542
    },
    {
      "epoch": 0.29246833276275247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4616,
      "step": 8543
    },
    {
      "epoch": 0.29250256761383087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4601,
      "step": 8544
    },
    {
      "epoch": 0.29253680246490926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.475,
      "step": 8545
    },
    {
      "epoch": 0.29257103731598766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4461,
      "step": 8546
    },
    {
      "epoch": 0.29260527216706606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.457,
      "step": 8547
    },
    {
      "epoch": 0.29263950701814445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4731,
      "step": 8548
    },
    {
      "epoch": 0.29267374186922285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4678,
      "step": 8549
    },
    {
      "epoch": 0.29270797672030124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4409,
      "step": 8550
    },
    {
      "epoch": 0.29274221157137964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4757,
      "step": 8551
    },
    {
      "epoch": 0.29277644642245804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4577,
      "step": 8552
    },
    {
      "epoch": 0.29281068127353643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4472,
      "step": 8553
    },
    {
      "epoch": 0.2928449161246149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.442,
      "step": 8554
    },
    {
      "epoch": 0.2928791509756933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4382,
      "step": 8555
    },
    {
      "epoch": 0.2929133858267717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4537,
      "step": 8556
    },
    {
      "epoch": 0.2929476206778501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 8557
    },
    {
      "epoch": 0.29298185552892847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4597,
      "step": 8558
    },
    {
      "epoch": 0.29301609038000687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4735,
      "step": 8559
    },
    {
      "epoch": 0.29305032523108526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4668,
      "step": 8560
    },
    {
      "epoch": 0.29308456008216366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4658,
      "step": 8561
    },
    {
      "epoch": 0.29311879493324206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4712,
      "step": 8562
    },
    {
      "epoch": 0.29315302978432045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4837,
      "step": 8563
    },
    {
      "epoch": 0.29318726463539885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4498,
      "step": 8564
    },
    {
      "epoch": 0.29322149948647724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4703,
      "step": 8565
    },
    {
      "epoch": 0.29325573433755564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4868,
      "step": 8566
    },
    {
      "epoch": 0.29328996918863404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4558,
      "step": 8567
    },
    {
      "epoch": 0.29332420403971243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4713,
      "step": 8568
    },
    {
      "epoch": 0.29335843889079083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4625,
      "step": 8569
    },
    {
      "epoch": 0.2933926737418692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4544,
      "step": 8570
    },
    {
      "epoch": 0.2934269085929476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4651,
      "step": 8571
    },
    {
      "epoch": 0.293461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4558,
      "step": 8572
    },
    {
      "epoch": 0.2934953782951044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4577,
      "step": 8573
    },
    {
      "epoch": 0.2935296131461828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4634,
      "step": 8574
    },
    {
      "epoch": 0.2935638479972612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.457,
      "step": 8575
    },
    {
      "epoch": 0.2935980828483396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4794,
      "step": 8576
    },
    {
      "epoch": 0.293632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4496,
      "step": 8577
    },
    {
      "epoch": 0.2936665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4364,
      "step": 8578
    },
    {
      "epoch": 0.2937007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4769,
      "step": 8579
    },
    {
      "epoch": 0.2937350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4663,
      "step": 8580
    },
    {
      "epoch": 0.2937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.461,
      "step": 8581
    },
    {
      "epoch": 0.29380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4516,
      "step": 8582
    },
    {
      "epoch": 0.2938377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4661,
      "step": 8583
    },
    {
      "epoch": 0.2938719616569668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.458,
      "step": 8584
    },
    {
      "epoch": 0.29390619650804517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4583,
      "step": 8585
    },
    {
      "epoch": 0.29394043135912357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4585,
      "step": 8586
    },
    {
      "epoch": 0.29397466621020196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4576,
      "step": 8587
    },
    {
      "epoch": 0.29400890106128036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4404,
      "step": 8588
    },
    {
      "epoch": 0.29404313591235876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4487,
      "step": 8589
    },
    {
      "epoch": 0.29407737076343715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4649,
      "step": 8590
    },
    {
      "epoch": 0.2941116056145156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 8591
    },
    {
      "epoch": 0.294145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4741,
      "step": 8592
    },
    {
      "epoch": 0.2941800753166724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4408,
      "step": 8593
    },
    {
      "epoch": 0.2942143101677508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4556,
      "step": 8594
    },
    {
      "epoch": 0.2942485450188292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4476,
      "step": 8595
    },
    {
      "epoch": 0.2942827798699076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4605,
      "step": 8596
    },
    {
      "epoch": 0.294317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.477,
      "step": 8597
    },
    {
      "epoch": 0.2943512495720644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4486,
      "step": 8598
    },
    {
      "epoch": 0.2943854844231428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4761,
      "step": 8599
    },
    {
      "epoch": 0.29441971927422117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4667,
      "step": 8600
    },
    {
      "epoch": 0.29445395412529957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4523,
      "step": 8601
    },
    {
      "epoch": 0.29448818897637796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4631,
      "step": 8602
    },
    {
      "epoch": 0.29452242382745636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4603,
      "step": 8603
    },
    {
      "epoch": 0.29455665867853476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4711,
      "step": 8604
    },
    {
      "epoch": 0.29459089352961315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4627,
      "step": 8605
    },
    {
      "epoch": 0.29462512838069155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4983,
      "step": 8606
    },
    {
      "epoch": 0.29465936323176994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4634,
      "step": 8607
    },
    {
      "epoch": 0.29469359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4448,
      "step": 8608
    },
    {
      "epoch": 0.29472783293392674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4697,
      "step": 8609
    },
    {
      "epoch": 0.29476206778500513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4561,
      "step": 8610
    },
    {
      "epoch": 0.29479630263608353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4473,
      "step": 8611
    },
    {
      "epoch": 0.2948305374871619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4634,
      "step": 8612
    },
    {
      "epoch": 0.2948647723382403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4447,
      "step": 8613
    },
    {
      "epoch": 0.2948990071893187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4493,
      "step": 8614
    },
    {
      "epoch": 0.2949332420403971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4851,
      "step": 8615
    },
    {
      "epoch": 0.2949674768914755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4448,
      "step": 8616
    },
    {
      "epoch": 0.2950017117425539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4706,
      "step": 8617
    },
    {
      "epoch": 0.2950359465936323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4417,
      "step": 8618
    },
    {
      "epoch": 0.2950701814447107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4459,
      "step": 8619
    },
    {
      "epoch": 0.2951044162957891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4628,
      "step": 8620
    },
    {
      "epoch": 0.2951386511468675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4793,
      "step": 8621
    },
    {
      "epoch": 0.2951728859979459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4681,
      "step": 8622
    },
    {
      "epoch": 0.2952071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4693,
      "step": 8623
    },
    {
      "epoch": 0.2952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.457,
      "step": 8624
    },
    {
      "epoch": 0.2952755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4525,
      "step": 8625
    },
    {
      "epoch": 0.2953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4835,
      "step": 8626
    },
    {
      "epoch": 0.2953440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4649,
      "step": 8627
    },
    {
      "epoch": 0.2953782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4678,
      "step": 8628
    },
    {
      "epoch": 0.2954125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.443,
      "step": 8629
    },
    {
      "epoch": 0.2954467648065731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4447,
      "step": 8630
    },
    {
      "epoch": 0.2954809996576515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4531,
      "step": 8631
    },
    {
      "epoch": 0.2955152345087299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4519,
      "step": 8632
    },
    {
      "epoch": 0.2955494693598083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4668,
      "step": 8633
    },
    {
      "epoch": 0.2955837042108867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4614,
      "step": 8634
    },
    {
      "epoch": 0.2956179390619651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4486,
      "step": 8635
    },
    {
      "epoch": 0.2956521739130435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4428,
      "step": 8636
    },
    {
      "epoch": 0.2956864087641219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.453,
      "step": 8637
    },
    {
      "epoch": 0.2957206436152003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4439,
      "step": 8638
    },
    {
      "epoch": 0.2957548784662787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4645,
      "step": 8639
    },
    {
      "epoch": 0.2957891133173571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4558,
      "step": 8640
    },
    {
      "epoch": 0.2958233481684355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4829,
      "step": 8641
    },
    {
      "epoch": 0.29585758301951387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4523,
      "step": 8642
    },
    {
      "epoch": 0.29589181787059227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4739,
      "step": 8643
    },
    {
      "epoch": 0.29592605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4569,
      "step": 8644
    },
    {
      "epoch": 0.29596028757274906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4635,
      "step": 8645
    },
    {
      "epoch": 0.29599452242382746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 8646
    },
    {
      "epoch": 0.29602875727490585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4719,
      "step": 8647
    },
    {
      "epoch": 0.29606299212598425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4623,
      "step": 8648
    },
    {
      "epoch": 0.29609722697706264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4539,
      "step": 8649
    },
    {
      "epoch": 0.29613146182814104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4643,
      "step": 8650
    },
    {
      "epoch": 0.29616569667921944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4641,
      "step": 8651
    },
    {
      "epoch": 0.29619993153029783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4521,
      "step": 8652
    },
    {
      "epoch": 0.29623416638137623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4696,
      "step": 8653
    },
    {
      "epoch": 0.2962684012324546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4786,
      "step": 8654
    },
    {
      "epoch": 0.296302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4517,
      "step": 8655
    },
    {
      "epoch": 0.2963368709346114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4574,
      "step": 8656
    },
    {
      "epoch": 0.2963711057856898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4423,
      "step": 8657
    },
    {
      "epoch": 0.2964053406367682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4561,
      "step": 8658
    },
    {
      "epoch": 0.2964395754878466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.454,
      "step": 8659
    },
    {
      "epoch": 0.296473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4669,
      "step": 8660
    },
    {
      "epoch": 0.2965080451900034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4658,
      "step": 8661
    },
    {
      "epoch": 0.2965422800410818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4622,
      "step": 8662
    },
    {
      "epoch": 0.2965765148921602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.462,
      "step": 8663
    },
    {
      "epoch": 0.29661074974323864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4388,
      "step": 8664
    },
    {
      "epoch": 0.29664498459431704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4699,
      "step": 8665
    },
    {
      "epoch": 0.29667921944539544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4299,
      "step": 8666
    },
    {
      "epoch": 0.29671345429647383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4362,
      "step": 8667
    },
    {
      "epoch": 0.29674768914755223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.462,
      "step": 8668
    },
    {
      "epoch": 0.2967819239986306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4442,
      "step": 8669
    },
    {
      "epoch": 0.296816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4394,
      "step": 8670
    },
    {
      "epoch": 0.2968503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4619,
      "step": 8671
    },
    {
      "epoch": 0.2968846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4597,
      "step": 8672
    },
    {
      "epoch": 0.2969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4498,
      "step": 8673
    },
    {
      "epoch": 0.2969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4528,
      "step": 8674
    },
    {
      "epoch": 0.296987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4529,
      "step": 8675
    },
    {
      "epoch": 0.2970215679561794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4341,
      "step": 8676
    },
    {
      "epoch": 0.2970558028072578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4512,
      "step": 8677
    },
    {
      "epoch": 0.2970900376583362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4473,
      "step": 8678
    },
    {
      "epoch": 0.2971242725094146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4565,
      "step": 8679
    },
    {
      "epoch": 0.297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.461,
      "step": 8680
    },
    {
      "epoch": 0.2971927422115714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4566,
      "step": 8681
    },
    {
      "epoch": 0.2972269770626498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4725,
      "step": 8682
    },
    {
      "epoch": 0.2972612119137282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.471,
      "step": 8683
    },
    {
      "epoch": 0.29729544676480657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4543,
      "step": 8684
    },
    {
      "epoch": 0.29732968161588497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4838,
      "step": 8685
    },
    {
      "epoch": 0.29736391646696336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4555,
      "step": 8686
    },
    {
      "epoch": 0.29739815131804176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.455,
      "step": 8687
    },
    {
      "epoch": 0.29743238616912016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4584,
      "step": 8688
    },
    {
      "epoch": 0.29746662102019855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4324,
      "step": 8689
    },
    {
      "epoch": 0.29750085587127695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4752,
      "step": 8690
    },
    {
      "epoch": 0.29753509072235534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4535,
      "step": 8691
    },
    {
      "epoch": 0.29756932557343374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 8692
    },
    {
      "epoch": 0.29760356042451214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4577,
      "step": 8693
    },
    {
      "epoch": 0.29763779527559053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4762,
      "step": 8694
    },
    {
      "epoch": 0.29767203012666893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.44,
      "step": 8695
    },
    {
      "epoch": 0.2977062649777473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4493,
      "step": 8696
    },
    {
      "epoch": 0.2977404998288257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4469,
      "step": 8697
    },
    {
      "epoch": 0.2977747346799041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4569,
      "step": 8698
    },
    {
      "epoch": 0.2978089695309825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4657,
      "step": 8699
    },
    {
      "epoch": 0.2978432043820609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4802,
      "step": 8700
    },
    {
      "epoch": 0.29787743923313936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.434,
      "step": 8701
    },
    {
      "epoch": 0.29791167408421776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.454,
      "step": 8702
    },
    {
      "epoch": 0.29794590893529616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4537,
      "step": 8703
    },
    {
      "epoch": 0.29798014378637455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4421,
      "step": 8704
    },
    {
      "epoch": 0.29801437863745295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.449,
      "step": 8705
    },
    {
      "epoch": 0.29804861348853134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4645,
      "step": 8706
    },
    {
      "epoch": 0.29808284833960974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4464,
      "step": 8707
    },
    {
      "epoch": 0.29811708319068814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4645,
      "step": 8708
    },
    {
      "epoch": 0.29815131804176653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4327,
      "step": 8709
    },
    {
      "epoch": 0.29818555289284493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4459,
      "step": 8710
    },
    {
      "epoch": 0.2982197877439233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4542,
      "step": 8711
    },
    {
      "epoch": 0.2982540225950017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4668,
      "step": 8712
    },
    {
      "epoch": 0.2982882574460801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4642,
      "step": 8713
    },
    {
      "epoch": 0.2983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4651,
      "step": 8714
    },
    {
      "epoch": 0.2983567271482369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4302,
      "step": 8715
    },
    {
      "epoch": 0.2983909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4601,
      "step": 8716
    },
    {
      "epoch": 0.2984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4616,
      "step": 8717
    },
    {
      "epoch": 0.2984594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4566,
      "step": 8718
    },
    {
      "epoch": 0.2984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4516,
      "step": 8719
    },
    {
      "epoch": 0.2985279014036289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4407,
      "step": 8720
    },
    {
      "epoch": 0.2985621362547073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4452,
      "step": 8721
    },
    {
      "epoch": 0.2985963711057857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4495,
      "step": 8722
    },
    {
      "epoch": 0.2986306059568641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.441,
      "step": 8723
    },
    {
      "epoch": 0.2986648408079425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4486,
      "step": 8724
    },
    {
      "epoch": 0.2986990756590209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4381,
      "step": 8725
    },
    {
      "epoch": 0.29873331051009927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4787,
      "step": 8726
    },
    {
      "epoch": 0.29876754536117767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4564,
      "step": 8727
    },
    {
      "epoch": 0.29880178021225606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4509,
      "step": 8728
    },
    {
      "epoch": 0.29883601506333446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4697,
      "step": 8729
    },
    {
      "epoch": 0.29887024991441286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4538,
      "step": 8730
    },
    {
      "epoch": 0.29890448476549125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4359,
      "step": 8731
    },
    {
      "epoch": 0.29893871961656965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4754,
      "step": 8732
    },
    {
      "epoch": 0.29897295446764804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4455,
      "step": 8733
    },
    {
      "epoch": 0.29900718931872644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4501,
      "step": 8734
    },
    {
      "epoch": 0.29904142416980484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4465,
      "step": 8735
    },
    {
      "epoch": 0.29907565902088323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4733,
      "step": 8736
    },
    {
      "epoch": 0.29910989387196163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4606,
      "step": 8737
    },
    {
      "epoch": 0.2991441287230401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4481,
      "step": 8738
    },
    {
      "epoch": 0.2991783635741185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4538,
      "step": 8739
    },
    {
      "epoch": 0.2992125984251969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4578,
      "step": 8740
    },
    {
      "epoch": 0.29924683327627527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4383,
      "step": 8741
    },
    {
      "epoch": 0.29928106812735367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4421,
      "step": 8742
    },
    {
      "epoch": 0.29931530297843206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4405,
      "step": 8743
    },
    {
      "epoch": 0.29934953782951046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4417,
      "step": 8744
    },
    {
      "epoch": 0.29938377268058886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4249,
      "step": 8745
    },
    {
      "epoch": 0.29941800753166725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4512,
      "step": 8746
    },
    {
      "epoch": 0.29945224238274565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4495,
      "step": 8747
    },
    {
      "epoch": 0.29948647723382404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4581,
      "step": 8748
    },
    {
      "epoch": 0.29952071208490244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4818,
      "step": 8749
    },
    {
      "epoch": 0.29955494693598084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4486,
      "step": 8750
    },
    {
      "epoch": 0.29958918178705923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4666,
      "step": 8751
    },
    {
      "epoch": 0.29962341663813763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4627,
      "step": 8752
    },
    {
      "epoch": 0.299657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4467,
      "step": 8753
    },
    {
      "epoch": 0.2996918863402944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.432,
      "step": 8754
    },
    {
      "epoch": 0.2997261211913728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4695,
      "step": 8755
    },
    {
      "epoch": 0.2997603560424512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4533,
      "step": 8756
    },
    {
      "epoch": 0.2997945908935296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4804,
      "step": 8757
    },
    {
      "epoch": 0.299828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4609,
      "step": 8758
    },
    {
      "epoch": 0.2998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4612,
      "step": 8759
    },
    {
      "epoch": 0.2998972954467648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4331,
      "step": 8760
    },
    {
      "epoch": 0.2999315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4579,
      "step": 8761
    },
    {
      "epoch": 0.2999657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.468,
      "step": 8762
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4651,
      "step": 8763
    },
    {
      "epoch": 0.3000342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4647,
      "step": 8764
    },
    {
      "epoch": 0.3000684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4364,
      "step": 8765
    },
    {
      "epoch": 0.3001027045532352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4593,
      "step": 8766
    },
    {
      "epoch": 0.3001369394043136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.468,
      "step": 8767
    },
    {
      "epoch": 0.30017117425539197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4649,
      "step": 8768
    },
    {
      "epoch": 0.30020540910647037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4562,
      "step": 8769
    },
    {
      "epoch": 0.30023964395754876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4462,
      "step": 8770
    },
    {
      "epoch": 0.30027387880862716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4457,
      "step": 8771
    },
    {
      "epoch": 0.30030811365970556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4348,
      "step": 8772
    },
    {
      "epoch": 0.30034234851078395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4447,
      "step": 8773
    },
    {
      "epoch": 0.30037658336186235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4334,
      "step": 8774
    },
    {
      "epoch": 0.3004108182129408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4661,
      "step": 8775
    },
    {
      "epoch": 0.3004450530640192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4558,
      "step": 8776
    },
    {
      "epoch": 0.3004792879150976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4421,
      "step": 8777
    },
    {
      "epoch": 0.300513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4421,
      "step": 8778
    },
    {
      "epoch": 0.3005477576172544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4487,
      "step": 8779
    },
    {
      "epoch": 0.3005819924683328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4331,
      "step": 8780
    },
    {
      "epoch": 0.3006162273194112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4268,
      "step": 8781
    },
    {
      "epoch": 0.3006504621704896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4518,
      "step": 8782
    },
    {
      "epoch": 0.30068469702156797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4487,
      "step": 8783
    },
    {
      "epoch": 0.30071893187264637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4412,
      "step": 8784
    },
    {
      "epoch": 0.30075316672372476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4413,
      "step": 8785
    },
    {
      "epoch": 0.30078740157480316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4551,
      "step": 8786
    },
    {
      "epoch": 0.30082163642588156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4368,
      "step": 8787
    },
    {
      "epoch": 0.30085587127695995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.459,
      "step": 8788
    },
    {
      "epoch": 0.30089010612803835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4609,
      "step": 8789
    },
    {
      "epoch": 0.30092434097911674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4704,
      "step": 8790
    },
    {
      "epoch": 0.30095857583019514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4453,
      "step": 8791
    },
    {
      "epoch": 0.30099281068127354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4481,
      "step": 8792
    },
    {
      "epoch": 0.30102704553235193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4491,
      "step": 8793
    },
    {
      "epoch": 0.30106128038343033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4469,
      "step": 8794
    },
    {
      "epoch": 0.3010955152345087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4603,
      "step": 8795
    },
    {
      "epoch": 0.3011297500855871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.444,
      "step": 8796
    },
    {
      "epoch": 0.3011639849366655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4509,
      "step": 8797
    },
    {
      "epoch": 0.3011982197877439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4586,
      "step": 8798
    },
    {
      "epoch": 0.3012324546388223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4492,
      "step": 8799
    },
    {
      "epoch": 0.3012666894899007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4414,
      "step": 8800
    },
    {
      "epoch": 0.3013009243409791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4588,
      "step": 8801
    },
    {
      "epoch": 0.3013351591920575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4461,
      "step": 8802
    },
    {
      "epoch": 0.3013693940431359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.462,
      "step": 8803
    },
    {
      "epoch": 0.3014036288942143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4497,
      "step": 8804
    },
    {
      "epoch": 0.3014378637452927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4561,
      "step": 8805
    },
    {
      "epoch": 0.3014720985963711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4513,
      "step": 8806
    },
    {
      "epoch": 0.3015063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4699,
      "step": 8807
    },
    {
      "epoch": 0.3015405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4348,
      "step": 8808
    },
    {
      "epoch": 0.3015748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.437,
      "step": 8809
    },
    {
      "epoch": 0.30160903800068467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.441,
      "step": 8810
    },
    {
      "epoch": 0.3016432728517631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4727,
      "step": 8811
    },
    {
      "epoch": 0.3016775077028415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4431,
      "step": 8812
    },
    {
      "epoch": 0.3017117425539199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4678,
      "step": 8813
    },
    {
      "epoch": 0.3017459774049983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.446,
      "step": 8814
    },
    {
      "epoch": 0.3017802122560767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.428,
      "step": 8815
    },
    {
      "epoch": 0.3018144471071551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4575,
      "step": 8816
    },
    {
      "epoch": 0.3018486819582335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4211,
      "step": 8817
    },
    {
      "epoch": 0.3018829168093119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4409,
      "step": 8818
    },
    {
      "epoch": 0.3019171516603903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4351,
      "step": 8819
    },
    {
      "epoch": 0.3019513865114687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4547,
      "step": 8820
    },
    {
      "epoch": 0.3019856213625471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4589,
      "step": 8821
    },
    {
      "epoch": 0.3020198562136255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4456,
      "step": 8822
    },
    {
      "epoch": 0.3020540910647039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4414,
      "step": 8823
    },
    {
      "epoch": 0.3020883259157823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4567,
      "step": 8824
    },
    {
      "epoch": 0.30212256076686067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.453,
      "step": 8825
    },
    {
      "epoch": 0.30215679561793907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4748,
      "step": 8826
    },
    {
      "epoch": 0.30219103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4387,
      "step": 8827
    },
    {
      "epoch": 0.30222526532009586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.458,
      "step": 8828
    },
    {
      "epoch": 0.30225950017117426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4568,
      "step": 8829
    },
    {
      "epoch": 0.30229373502225265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4607,
      "step": 8830
    },
    {
      "epoch": 0.30232796987333105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4484,
      "step": 8831
    },
    {
      "epoch": 0.30236220472440944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4353,
      "step": 8832
    },
    {
      "epoch": 0.30239643957548784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4512,
      "step": 8833
    },
    {
      "epoch": 0.30243067442656624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4608,
      "step": 8834
    },
    {
      "epoch": 0.30246490927764463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4587,
      "step": 8835
    },
    {
      "epoch": 0.30249914412872303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.449,
      "step": 8836
    },
    {
      "epoch": 0.3025333789798014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 8837
    },
    {
      "epoch": 0.3025676138308798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4194,
      "step": 8838
    },
    {
      "epoch": 0.3026018486819582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.46,
      "step": 8839
    },
    {
      "epoch": 0.3026360835330366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4439,
      "step": 8840
    },
    {
      "epoch": 0.302670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4575,
      "step": 8841
    },
    {
      "epoch": 0.3027045532351934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4514,
      "step": 8842
    },
    {
      "epoch": 0.3027387880862718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4473,
      "step": 8843
    },
    {
      "epoch": 0.3027730229373502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4526,
      "step": 8844
    },
    {
      "epoch": 0.3028072577884286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4473,
      "step": 8845
    },
    {
      "epoch": 0.302841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4228,
      "step": 8846
    },
    {
      "epoch": 0.3028757274905854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4334,
      "step": 8847
    },
    {
      "epoch": 0.30290996234166384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4479,
      "step": 8848
    },
    {
      "epoch": 0.30294419719274224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4443,
      "step": 8849
    },
    {
      "epoch": 0.30297843204382063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4295,
      "step": 8850
    },
    {
      "epoch": 0.30301266689489903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4386,
      "step": 8851
    },
    {
      "epoch": 0.3030469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4459,
      "step": 8852
    },
    {
      "epoch": 0.3030811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4431,
      "step": 8853
    },
    {
      "epoch": 0.3031153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4441,
      "step": 8854
    },
    {
      "epoch": 0.3031496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 8855
    },
    {
      "epoch": 0.303183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4487,
      "step": 8856
    },
    {
      "epoch": 0.3032180760013694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4601,
      "step": 8857
    },
    {
      "epoch": 0.3032523108524478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4313,
      "step": 8858
    },
    {
      "epoch": 0.3032865457035262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4249,
      "step": 8859
    },
    {
      "epoch": 0.3033207805546046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4643,
      "step": 8860
    },
    {
      "epoch": 0.303355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4448,
      "step": 8861
    },
    {
      "epoch": 0.3033892502567614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4388,
      "step": 8862
    },
    {
      "epoch": 0.3034234851078398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4581,
      "step": 8863
    },
    {
      "epoch": 0.3034577199589182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4404,
      "step": 8864
    },
    {
      "epoch": 0.3034919548099966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4523,
      "step": 8865
    },
    {
      "epoch": 0.303526189661075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4592,
      "step": 8866
    },
    {
      "epoch": 0.30356042451215337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4602,
      "step": 8867
    },
    {
      "epoch": 0.30359465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4462,
      "step": 8868
    },
    {
      "epoch": 0.30362889421431016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4652,
      "step": 8869
    },
    {
      "epoch": 0.30366312906538856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4519,
      "step": 8870
    },
    {
      "epoch": 0.30369736391646696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4352,
      "step": 8871
    },
    {
      "epoch": 0.30373159876754535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4298,
      "step": 8872
    },
    {
      "epoch": 0.30376583361862375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4407,
      "step": 8873
    },
    {
      "epoch": 0.30380006846970214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4387,
      "step": 8874
    },
    {
      "epoch": 0.30383430332078054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4411,
      "step": 8875
    },
    {
      "epoch": 0.30386853817185894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4331,
      "step": 8876
    },
    {
      "epoch": 0.30390277302293733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4303,
      "step": 8877
    },
    {
      "epoch": 0.30393700787401573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4374,
      "step": 8878
    },
    {
      "epoch": 0.3039712427250941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 8879
    },
    {
      "epoch": 0.3040054775761725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4521,
      "step": 8880
    },
    {
      "epoch": 0.3040397124272509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4335,
      "step": 8881
    },
    {
      "epoch": 0.3040739472783293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4526,
      "step": 8882
    },
    {
      "epoch": 0.3041081821294077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4542,
      "step": 8883
    },
    {
      "epoch": 0.3041424169804861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4528,
      "step": 8884
    },
    {
      "epoch": 0.30417665183156456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4472,
      "step": 8885
    },
    {
      "epoch": 0.30421088668264296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4294,
      "step": 8886
    },
    {
      "epoch": 0.30424512153372135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4649,
      "step": 8887
    },
    {
      "epoch": 0.30427935638479975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4567,
      "step": 8888
    },
    {
      "epoch": 0.30431359123587814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4664,
      "step": 8889
    },
    {
      "epoch": 0.30434782608695654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4626,
      "step": 8890
    },
    {
      "epoch": 0.30438206093803494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4454,
      "step": 8891
    },
    {
      "epoch": 0.30441629578911333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4602,
      "step": 8892
    },
    {
      "epoch": 0.30445053064019173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4358,
      "step": 8893
    },
    {
      "epoch": 0.3044847654912701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4435,
      "step": 8894
    },
    {
      "epoch": 0.3045190003423485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4437,
      "step": 8895
    },
    {
      "epoch": 0.3045532351934269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4493,
      "step": 8896
    },
    {
      "epoch": 0.3045874700445053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4525,
      "step": 8897
    },
    {
      "epoch": 0.3046217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 8898
    },
    {
      "epoch": 0.3046559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4376,
      "step": 8899
    },
    {
      "epoch": 0.3046901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4576,
      "step": 8900
    },
    {
      "epoch": 0.3047244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4437,
      "step": 8901
    },
    {
      "epoch": 0.3047586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4471,
      "step": 8902
    },
    {
      "epoch": 0.3047928791509757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4516,
      "step": 8903
    },
    {
      "epoch": 0.3048271140020541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4594,
      "step": 8904
    },
    {
      "epoch": 0.3048613488531325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.442,
      "step": 8905
    },
    {
      "epoch": 0.3048955837042109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4371,
      "step": 8906
    },
    {
      "epoch": 0.3049298185552893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.447,
      "step": 8907
    },
    {
      "epoch": 0.3049640534063677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4379,
      "step": 8908
    },
    {
      "epoch": 0.30499828825744607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.467,
      "step": 8909
    },
    {
      "epoch": 0.30503252310852447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4486,
      "step": 8910
    },
    {
      "epoch": 0.30506675795960286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4557,
      "step": 8911
    },
    {
      "epoch": 0.30510099281068126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4498,
      "step": 8912
    },
    {
      "epoch": 0.30513522766175966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4523,
      "step": 8913
    },
    {
      "epoch": 0.30516946251283805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4412,
      "step": 8914
    },
    {
      "epoch": 0.30520369736391645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4518,
      "step": 8915
    },
    {
      "epoch": 0.30523793221499484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4695,
      "step": 8916
    },
    {
      "epoch": 0.30527216706607324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4592,
      "step": 8917
    },
    {
      "epoch": 0.30530640191715164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4608,
      "step": 8918
    },
    {
      "epoch": 0.30534063676823003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4307,
      "step": 8919
    },
    {
      "epoch": 0.30537487161930843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4625,
      "step": 8920
    },
    {
      "epoch": 0.3054091064703868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4445,
      "step": 8921
    },
    {
      "epoch": 0.3054433413214653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4348,
      "step": 8922
    },
    {
      "epoch": 0.3054775761725437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4273,
      "step": 8923
    },
    {
      "epoch": 0.30551181102362207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.415,
      "step": 8924
    },
    {
      "epoch": 0.30554604587470047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4508,
      "step": 8925
    },
    {
      "epoch": 0.30558028072577886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4303,
      "step": 8926
    },
    {
      "epoch": 0.30561451557685726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4559,
      "step": 8927
    },
    {
      "epoch": 0.30564875042793566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4354,
      "step": 8928
    },
    {
      "epoch": 0.30568298527901405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4614,
      "step": 8929
    },
    {
      "epoch": 0.30571722013009245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4277,
      "step": 8930
    },
    {
      "epoch": 0.30575145498117084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 7.0434,
      "step": 8931
    },
    {
      "epoch": 0.30578568983224924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.6426,
      "step": 8932
    },
    {
      "epoch": 0.30581992468332764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 5.1058,
      "step": 8933
    },
    {
      "epoch": 0.30585415953440603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.6238,
      "step": 8934
    },
    {
      "epoch": 0.30588839438548443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5592,
      "step": 8935
    },
    {
      "epoch": 0.3059226292365628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.7523,
      "step": 8936
    },
    {
      "epoch": 0.3059568640876412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.6597,
      "step": 8937
    },
    {
      "epoch": 0.3059910989387196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.9457,
      "step": 8938
    },
    {
      "epoch": 0.306025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 10.9551,
      "step": 8939
    },
    {
      "epoch": 0.3060595686408764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 8.9601,
      "step": 8940
    },
    {
      "epoch": 0.3060938034919548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 5.1361,
      "step": 8941
    },
    {
      "epoch": 0.3061280383430332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 9.2583,
      "step": 8942
    },
    {
      "epoch": 0.3061622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 8.4725,
      "step": 8943
    },
    {
      "epoch": 0.30619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 11.9059,
      "step": 8944
    },
    {
      "epoch": 0.3062307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 8.9147,
      "step": 8945
    },
    {
      "epoch": 0.3062649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 8.327,
      "step": 8946
    },
    {
      "epoch": 0.3062992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 13.6663,
      "step": 8947
    },
    {
      "epoch": 0.3063334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 17.3183,
      "step": 8948
    },
    {
      "epoch": 0.306367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 6.5117,
      "step": 8949
    },
    {
      "epoch": 0.3064019171516604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.9096,
      "step": 8950
    },
    {
      "epoch": 0.30643615200273877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.8293,
      "step": 8951
    },
    {
      "epoch": 0.30647038685381717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 5.1092,
      "step": 8952
    },
    {
      "epoch": 0.30650462170489556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.7768,
      "step": 8953
    },
    {
      "epoch": 0.30653885655597396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.717,
      "step": 8954
    },
    {
      "epoch": 0.30657309140705236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.7122,
      "step": 8955
    },
    {
      "epoch": 0.30660732625813075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.7122,
      "step": 8956
    },
    {
      "epoch": 0.30664156110920915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.705,
      "step": 8957
    },
    {
      "epoch": 0.3066757959602876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.7245,
      "step": 8958
    },
    {
      "epoch": 0.306710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.6931,
      "step": 8959
    },
    {
      "epoch": 0.3067442656624444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.6908,
      "step": 8960
    },
    {
      "epoch": 0.3067785005135228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.6879,
      "step": 8961
    },
    {
      "epoch": 0.3068127353646012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.6809,
      "step": 8962
    },
    {
      "epoch": 0.3068469702156796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.6153,
      "step": 8963
    },
    {
      "epoch": 0.306881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.6323,
      "step": 8964
    },
    {
      "epoch": 0.3069154399178364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.6106,
      "step": 8965
    },
    {
      "epoch": 0.30694967476891477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5981,
      "step": 8966
    },
    {
      "epoch": 0.30698390961999317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.6026,
      "step": 8967
    },
    {
      "epoch": 0.30701814447107156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5897,
      "step": 8968
    },
    {
      "epoch": 0.30705237932214996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5587,
      "step": 8969
    },
    {
      "epoch": 0.30708661417322836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5844,
      "step": 8970
    },
    {
      "epoch": 0.30712084902430675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5709,
      "step": 8971
    },
    {
      "epoch": 0.30715508387538515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5939,
      "step": 8972
    },
    {
      "epoch": 0.30718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.6045,
      "step": 8973
    },
    {
      "epoch": 0.30722355357754194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5769,
      "step": 8974
    },
    {
      "epoch": 0.30725778842862034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5798,
      "step": 8975
    },
    {
      "epoch": 0.30729202327969873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5514,
      "step": 8976
    },
    {
      "epoch": 0.30732625813077713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5851,
      "step": 8977
    },
    {
      "epoch": 0.3073604929818555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5887,
      "step": 8978
    },
    {
      "epoch": 0.3073947278329339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5519,
      "step": 8979
    },
    {
      "epoch": 0.3074289626840123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5531,
      "step": 8980
    },
    {
      "epoch": 0.3074631975350907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5549,
      "step": 8981
    },
    {
      "epoch": 0.3074974323861691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5414,
      "step": 8982
    },
    {
      "epoch": 0.3075316672372475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5734,
      "step": 8983
    },
    {
      "epoch": 0.3075659020883259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5416,
      "step": 8984
    },
    {
      "epoch": 0.3076001369394043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5292,
      "step": 8985
    },
    {
      "epoch": 0.3076343717904827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5474,
      "step": 8986
    },
    {
      "epoch": 0.3076686066415611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5457,
      "step": 8987
    },
    {
      "epoch": 0.3077028414926395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5461,
      "step": 8988
    },
    {
      "epoch": 0.3077370763437179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5189,
      "step": 8989
    },
    {
      "epoch": 0.3077713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5072,
      "step": 8990
    },
    {
      "epoch": 0.3078055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5562,
      "step": 8991
    },
    {
      "epoch": 0.3078397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.498,
      "step": 8992
    },
    {
      "epoch": 0.30787401574803147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5169,
      "step": 8993
    },
    {
      "epoch": 0.30790825059910987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5306,
      "step": 8994
    },
    {
      "epoch": 0.3079424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5118,
      "step": 8995
    },
    {
      "epoch": 0.3079767203012667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5024,
      "step": 8996
    },
    {
      "epoch": 0.3080109551523451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5206,
      "step": 8997
    },
    {
      "epoch": 0.3080451900034235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.508,
      "step": 8998
    },
    {
      "epoch": 0.3080794248545019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5081,
      "step": 8999
    },
    {
      "epoch": 0.3081136597055803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5562,
      "step": 9000
    },
    {
      "epoch": 0.3081478945566587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4833,
      "step": 9001
    },
    {
      "epoch": 0.3081821294077371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4774,
      "step": 9002
    },
    {
      "epoch": 0.3082163642588155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5049,
      "step": 9003
    },
    {
      "epoch": 0.3082505991098939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4389,
      "step": 9004
    },
    {
      "epoch": 0.3082848339609723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5031,
      "step": 9005
    },
    {
      "epoch": 0.3083190688120507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4959,
      "step": 9006
    },
    {
      "epoch": 0.3083533036631291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4306,
      "step": 9007
    },
    {
      "epoch": 0.30838753851420747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4356,
      "step": 9008
    },
    {
      "epoch": 0.30842177336528587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4853,
      "step": 9009
    },
    {
      "epoch": 0.30845600821636426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4451,
      "step": 9010
    },
    {
      "epoch": 0.30849024306744266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4579,
      "step": 9011
    },
    {
      "epoch": 0.30852447791852106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4889,
      "step": 9012
    },
    {
      "epoch": 0.30855871276959945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4947,
      "step": 9013
    },
    {
      "epoch": 0.30859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4128,
      "step": 9014
    },
    {
      "epoch": 0.30862718247175625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4103,
      "step": 9015
    },
    {
      "epoch": 0.30866141732283464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4203,
      "step": 9016
    },
    {
      "epoch": 0.30869565217391304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3765,
      "step": 9017
    },
    {
      "epoch": 0.30872988702499143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4073,
      "step": 9018
    },
    {
      "epoch": 0.30876412187606983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3807,
      "step": 9019
    },
    {
      "epoch": 0.3087983567271482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3808,
      "step": 9020
    },
    {
      "epoch": 0.3088325915782266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3634,
      "step": 9021
    },
    {
      "epoch": 0.308866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4009,
      "step": 9022
    },
    {
      "epoch": 0.3089010612803834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4561,
      "step": 9023
    },
    {
      "epoch": 0.3089352961314618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3747,
      "step": 9024
    },
    {
      "epoch": 0.3089695309825402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.7352,
      "step": 9025
    },
    {
      "epoch": 0.3090037658336186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.572,
      "step": 9026
    },
    {
      "epoch": 0.309038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5264,
      "step": 9027
    },
    {
      "epoch": 0.3090722355357754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5388,
      "step": 9028
    },
    {
      "epoch": 0.3091064703868538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4928,
      "step": 9029
    },
    {
      "epoch": 0.3091407052379322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5057,
      "step": 9030
    },
    {
      "epoch": 0.3091749400890106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5876,
      "step": 9031
    },
    {
      "epoch": 0.30920917494008904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4699,
      "step": 9032
    },
    {
      "epoch": 0.30924340979116743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4809,
      "step": 9033
    },
    {
      "epoch": 0.30927764464224583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4941,
      "step": 9034
    },
    {
      "epoch": 0.3093118794933242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5059,
      "step": 9035
    },
    {
      "epoch": 0.3093461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.5158,
      "step": 9036
    },
    {
      "epoch": 0.309380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4928,
      "step": 9037
    },
    {
      "epoch": 0.3094145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4815,
      "step": 9038
    },
    {
      "epoch": 0.3094488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4606,
      "step": 9039
    },
    {
      "epoch": 0.3094830537487162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4923,
      "step": 9040
    },
    {
      "epoch": 0.3095172885997946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4571,
      "step": 9041
    },
    {
      "epoch": 0.309551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4157,
      "step": 9042
    },
    {
      "epoch": 0.3095857583019514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4862,
      "step": 9043
    },
    {
      "epoch": 0.3096199931530298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4731,
      "step": 9044
    },
    {
      "epoch": 0.3096542280041082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4053,
      "step": 9045
    },
    {
      "epoch": 0.3096884628551866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4302,
      "step": 9046
    },
    {
      "epoch": 0.309722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4689,
      "step": 9047
    },
    {
      "epoch": 0.3097569325573434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3684,
      "step": 9048
    },
    {
      "epoch": 0.3097911674084218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4377,
      "step": 9049
    },
    {
      "epoch": 0.30982540225950017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4275,
      "step": 9050
    },
    {
      "epoch": 0.30985963711057857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4224,
      "step": 9051
    },
    {
      "epoch": 0.30989387196165696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4057,
      "step": 9052
    },
    {
      "epoch": 0.30992810681273536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3935,
      "step": 9053
    },
    {
      "epoch": 0.30996234166381376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4451,
      "step": 9054
    },
    {
      "epoch": 0.30999657651489215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4198,
      "step": 9055
    },
    {
      "epoch": 0.31003081136597055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4292,
      "step": 9056
    },
    {
      "epoch": 0.31006504621704895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4242,
      "step": 9057
    },
    {
      "epoch": 0.31009928106812734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3965,
      "step": 9058
    },
    {
      "epoch": 0.31013351591920574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.391,
      "step": 9059
    },
    {
      "epoch": 0.31016775077028413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4136,
      "step": 9060
    },
    {
      "epoch": 0.31020198562136253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3844,
      "step": 9061
    },
    {
      "epoch": 0.3102362204724409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.4168,
      "step": 9062
    },
    {
      "epoch": 0.3102704553235193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3586,
      "step": 9063
    },
    {
      "epoch": 0.3103046901745977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.319,
      "step": 9064
    },
    {
      "epoch": 0.3103389250256761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.396,
      "step": 9065
    },
    {
      "epoch": 0.3103731598767545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3105,
      "step": 9066
    },
    {
      "epoch": 0.3104073947278329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3827,
      "step": 9067
    },
    {
      "epoch": 0.3104416295789113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3155,
      "step": 9068
    },
    {
      "epoch": 0.31047586442998976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3669,
      "step": 9069
    },
    {
      "epoch": 0.31051009928106815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.2986,
      "step": 9070
    },
    {
      "epoch": 0.31054433413214655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.2576,
      "step": 9071
    },
    {
      "epoch": 0.31057856898322495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.2916,
      "step": 9072
    },
    {
      "epoch": 0.31061280383430334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.2972,
      "step": 9073
    },
    {
      "epoch": 0.31064703868538174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.31,
      "step": 9074
    },
    {
      "epoch": 0.31068127353646013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.3179,
      "step": 9075
    },
    {
      "epoch": 0.31071550838753853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.1992,
      "step": 9076
    },
    {
      "epoch": 0.3107497432386169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.317,
      "step": 9077
    },
    {
      "epoch": 0.3107839780896953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.2247,
      "step": 9078
    },
    {
      "epoch": 0.3108182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.1986,
      "step": 9079
    },
    {
      "epoch": 0.3108524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.221,
      "step": 9080
    },
    {
      "epoch": 0.3108866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.2375,
      "step": 9081
    },
    {
      "epoch": 0.3109209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.1401,
      "step": 9082
    },
    {
      "epoch": 0.3109551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.1118,
      "step": 9083
    },
    {
      "epoch": 0.3109893871961657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.1092,
      "step": 9084
    },
    {
      "epoch": 0.3110236220472441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.1266,
      "step": 9085
    },
    {
      "epoch": 0.3110578568983225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0949,
      "step": 9086
    },
    {
      "epoch": 0.3110920917494009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.06,
      "step": 9087
    },
    {
      "epoch": 0.3111263266004793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0659,
      "step": 9088
    },
    {
      "epoch": 0.3111605614515577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0537,
      "step": 9089
    },
    {
      "epoch": 0.3111947963026361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9929,
      "step": 9090
    },
    {
      "epoch": 0.3112290311537145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0608,
      "step": 9091
    },
    {
      "epoch": 0.31126326600479287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0227,
      "step": 9092
    },
    {
      "epoch": 0.31129750085587127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9809,
      "step": 9093
    },
    {
      "epoch": 0.31133173570694966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0553,
      "step": 9094
    },
    {
      "epoch": 0.31136597055802806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.005,
      "step": 9095
    },
    {
      "epoch": 0.31140020540910646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0229,
      "step": 9096
    },
    {
      "epoch": 0.31143444026018485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9847,
      "step": 9097
    },
    {
      "epoch": 0.31146867511126325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0182,
      "step": 9098
    },
    {
      "epoch": 0.31150290996234165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0046,
      "step": 9099
    },
    {
      "epoch": 0.31153714481342004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.017,
      "step": 9100
    },
    {
      "epoch": 0.31157137966449844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0961,
      "step": 9101
    },
    {
      "epoch": 0.31160561451557683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.1599,
      "step": 9102
    },
    {
      "epoch": 0.31163984936665523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.1051,
      "step": 9103
    },
    {
      "epoch": 0.3116740842177336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.991,
      "step": 9104
    },
    {
      "epoch": 0.311708319068812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0373,
      "step": 9105
    },
    {
      "epoch": 0.3117425539198905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9676,
      "step": 9106
    },
    {
      "epoch": 0.31177678877096887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0057,
      "step": 9107
    },
    {
      "epoch": 0.31181102362204727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9473,
      "step": 9108
    },
    {
      "epoch": 0.31184525847312566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9684,
      "step": 9109
    },
    {
      "epoch": 0.31187949332420406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9933,
      "step": 9110
    },
    {
      "epoch": 0.31191372817528246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9929,
      "step": 9111
    },
    {
      "epoch": 0.31194796302636085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9369,
      "step": 9112
    },
    {
      "epoch": 0.31198219787743925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9579,
      "step": 9113
    },
    {
      "epoch": 0.31201643272851765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0082,
      "step": 9114
    },
    {
      "epoch": 0.31205066757959604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.977,
      "step": 9115
    },
    {
      "epoch": 0.31208490243067444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 4.0014,
      "step": 9116
    },
    {
      "epoch": 0.31211913728175283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9407,
      "step": 9117
    },
    {
      "epoch": 0.31215337213283123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.932,
      "step": 9118
    },
    {
      "epoch": 0.3121876069839096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8994,
      "step": 9119
    },
    {
      "epoch": 0.312221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8774,
      "step": 9120
    },
    {
      "epoch": 0.3122560766860664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9223,
      "step": 9121
    },
    {
      "epoch": 0.3122903115371448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8948,
      "step": 9122
    },
    {
      "epoch": 0.3123245463882232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9562,
      "step": 9123
    },
    {
      "epoch": 0.3123587812393016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9283,
      "step": 9124
    },
    {
      "epoch": 0.31239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9002,
      "step": 9125
    },
    {
      "epoch": 0.3124272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.91,
      "step": 9126
    },
    {
      "epoch": 0.3124614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9096,
      "step": 9127
    },
    {
      "epoch": 0.3124957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8838,
      "step": 9128
    },
    {
      "epoch": 0.3125299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9111,
      "step": 9129
    },
    {
      "epoch": 0.312564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.895,
      "step": 9130
    },
    {
      "epoch": 0.3125984251968504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9048,
      "step": 9131
    },
    {
      "epoch": 0.3126326600479288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9126,
      "step": 9132
    },
    {
      "epoch": 0.3126668948990072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9083,
      "step": 9133
    },
    {
      "epoch": 0.31270112975008557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9008,
      "step": 9134
    },
    {
      "epoch": 0.31273536460116397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8578,
      "step": 9135
    },
    {
      "epoch": 0.31276959945224236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8788,
      "step": 9136
    },
    {
      "epoch": 0.31280383430332076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8713,
      "step": 9137
    },
    {
      "epoch": 0.31283806915439916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8548,
      "step": 9138
    },
    {
      "epoch": 0.31287230400547755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8612,
      "step": 9139
    },
    {
      "epoch": 0.31290653885655595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9107,
      "step": 9140
    },
    {
      "epoch": 0.31294077370763435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8641,
      "step": 9141
    },
    {
      "epoch": 0.3129750085587128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8739,
      "step": 9142
    },
    {
      "epoch": 0.3130092434097912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8866,
      "step": 9143
    },
    {
      "epoch": 0.3130434782608696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8396,
      "step": 9144
    },
    {
      "epoch": 0.313077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8459,
      "step": 9145
    },
    {
      "epoch": 0.3131119479630264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.9143,
      "step": 9146
    },
    {
      "epoch": 0.3131461828141048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8783,
      "step": 9147
    },
    {
      "epoch": 0.3131804176651832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8563,
      "step": 9148
    },
    {
      "epoch": 0.31321465251626157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8266,
      "step": 9149
    },
    {
      "epoch": 0.31324888736733997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8794,
      "step": 9150
    },
    {
      "epoch": 0.31328312221841836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8248,
      "step": 9151
    },
    {
      "epoch": 0.31331735706949676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8128,
      "step": 9152
    },
    {
      "epoch": 0.31335159192057516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7913,
      "step": 9153
    },
    {
      "epoch": 0.31338582677165355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7934,
      "step": 9154
    },
    {
      "epoch": 0.31342006162273195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8048,
      "step": 9155
    },
    {
      "epoch": 0.31345429647381035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8676,
      "step": 9156
    },
    {
      "epoch": 0.31348853132488874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7597,
      "step": 9157
    },
    {
      "epoch": 0.31352276617596714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8131,
      "step": 9158
    },
    {
      "epoch": 0.31355700102704553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7888,
      "step": 9159
    },
    {
      "epoch": 0.31359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7998,
      "step": 9160
    },
    {
      "epoch": 0.3136254707292023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7541,
      "step": 9161
    },
    {
      "epoch": 0.3136597055802807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.744,
      "step": 9162
    },
    {
      "epoch": 0.3136939404313591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7937,
      "step": 9163
    },
    {
      "epoch": 0.3137281752824375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7442,
      "step": 9164
    },
    {
      "epoch": 0.3137624101335159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7753,
      "step": 9165
    },
    {
      "epoch": 0.3137966449845943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8532,
      "step": 9166
    },
    {
      "epoch": 0.3138308798356727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8377,
      "step": 9167
    },
    {
      "epoch": 0.3138651146867511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.832,
      "step": 9168
    },
    {
      "epoch": 0.3138993495378295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8332,
      "step": 9169
    },
    {
      "epoch": 0.3139335843889079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8214,
      "step": 9170
    },
    {
      "epoch": 0.3139678192399863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.821,
      "step": 9171
    },
    {
      "epoch": 0.3140020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8763,
      "step": 9172
    },
    {
      "epoch": 0.3140362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8408,
      "step": 9173
    },
    {
      "epoch": 0.3140705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8618,
      "step": 9174
    },
    {
      "epoch": 0.3141047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8203,
      "step": 9175
    },
    {
      "epoch": 0.31413899349537827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8425,
      "step": 9176
    },
    {
      "epoch": 0.31417322834645667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8156,
      "step": 9177
    },
    {
      "epoch": 0.31420746319753506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.8288,
      "step": 9178
    },
    {
      "epoch": 0.3142416980486135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7542,
      "step": 9179
    },
    {
      "epoch": 0.3142759328996919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7979,
      "step": 9180
    },
    {
      "epoch": 0.3143101677507703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7984,
      "step": 9181
    },
    {
      "epoch": 0.3143444026018487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.748,
      "step": 9182
    },
    {
      "epoch": 0.3143786374529271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7736,
      "step": 9183
    },
    {
      "epoch": 0.3144128723040055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7254,
      "step": 9184
    },
    {
      "epoch": 0.3144471071550839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7332,
      "step": 9185
    },
    {
      "epoch": 0.3144813420061623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7586,
      "step": 9186
    },
    {
      "epoch": 0.3145155768572407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7248,
      "step": 9187
    },
    {
      "epoch": 0.3145498117083191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7209,
      "step": 9188
    },
    {
      "epoch": 0.3145840465593975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7369,
      "step": 9189
    },
    {
      "epoch": 0.3146182814104759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.695,
      "step": 9190
    },
    {
      "epoch": 0.31465251626155427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7568,
      "step": 9191
    },
    {
      "epoch": 0.31468675111263267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6895,
      "step": 9192
    },
    {
      "epoch": 0.31472098596371106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6989,
      "step": 9193
    },
    {
      "epoch": 0.31475522081478946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7267,
      "step": 9194
    },
    {
      "epoch": 0.31478945566586786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.674,
      "step": 9195
    },
    {
      "epoch": 0.31482369051694625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6808,
      "step": 9196
    },
    {
      "epoch": 0.31485792536802465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.668,
      "step": 9197
    },
    {
      "epoch": 0.31489216021910305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6613,
      "step": 9198
    },
    {
      "epoch": 0.31492639507018144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6813,
      "step": 9199
    },
    {
      "epoch": 0.31496062992125984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6855,
      "step": 9200
    },
    {
      "epoch": 0.31499486477233823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6504,
      "step": 9201
    },
    {
      "epoch": 0.31502909962341663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6664,
      "step": 9202
    },
    {
      "epoch": 0.315063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6623,
      "step": 9203
    },
    {
      "epoch": 0.3150975693255734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6414,
      "step": 9204
    },
    {
      "epoch": 0.3151318041766518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6506,
      "step": 9205
    },
    {
      "epoch": 0.3151660390277302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6489,
      "step": 9206
    },
    {
      "epoch": 0.3152002738788086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6597,
      "step": 9207
    },
    {
      "epoch": 0.315234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.673,
      "step": 9208
    },
    {
      "epoch": 0.3152687435809654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6114,
      "step": 9209
    },
    {
      "epoch": 0.3153029784320438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.639,
      "step": 9210
    },
    {
      "epoch": 0.3153372132831222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6509,
      "step": 9211
    },
    {
      "epoch": 0.3153714481342006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.7302,
      "step": 9212
    },
    {
      "epoch": 0.315405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6523,
      "step": 9213
    },
    {
      "epoch": 0.3154399178363574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6109,
      "step": 9214
    },
    {
      "epoch": 0.3154741526874358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6519,
      "step": 9215
    },
    {
      "epoch": 0.31550838753851423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6563,
      "step": 9216
    },
    {
      "epoch": 0.31554262238959263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6657,
      "step": 9217
    },
    {
      "epoch": 0.315576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.664,
      "step": 9218
    },
    {
      "epoch": 0.3156110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6633,
      "step": 9219
    },
    {
      "epoch": 0.3156453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.667,
      "step": 9220
    },
    {
      "epoch": 0.3156795617939062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6172,
      "step": 9221
    },
    {
      "epoch": 0.3157137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6447,
      "step": 9222
    },
    {
      "epoch": 0.315748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6935,
      "step": 9223
    },
    {
      "epoch": 0.3157822663471414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6854,
      "step": 9224
    },
    {
      "epoch": 0.3158165011982198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6324,
      "step": 9225
    },
    {
      "epoch": 0.3158507360492982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6642,
      "step": 9226
    },
    {
      "epoch": 0.3158849709003766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6499,
      "step": 9227
    },
    {
      "epoch": 0.315919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6043,
      "step": 9228
    },
    {
      "epoch": 0.3159534406025334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5911,
      "step": 9229
    },
    {
      "epoch": 0.3159876754536118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5743,
      "step": 9230
    },
    {
      "epoch": 0.3160219103046902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6036,
      "step": 9231
    },
    {
      "epoch": 0.3160561451557686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6175,
      "step": 9232
    },
    {
      "epoch": 0.31609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5859,
      "step": 9233
    },
    {
      "epoch": 0.31612461485792537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5807,
      "step": 9234
    },
    {
      "epoch": 0.31615884970900376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5808,
      "step": 9235
    },
    {
      "epoch": 0.31619308456008216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6114,
      "step": 9236
    },
    {
      "epoch": 0.31622731941116056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5889,
      "step": 9237
    },
    {
      "epoch": 0.31626155426223895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5447,
      "step": 9238
    },
    {
      "epoch": 0.31629578911331735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5611,
      "step": 9239
    },
    {
      "epoch": 0.31633002396439575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5369,
      "step": 9240
    },
    {
      "epoch": 0.31636425881547414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5137,
      "step": 9241
    },
    {
      "epoch": 0.31639849366655254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4962,
      "step": 9242
    },
    {
      "epoch": 0.31643272851763093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5603,
      "step": 9243
    },
    {
      "epoch": 0.31646696336870933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5219,
      "step": 9244
    },
    {
      "epoch": 0.3165011982197877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5468,
      "step": 9245
    },
    {
      "epoch": 0.3165354330708661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5407,
      "step": 9246
    },
    {
      "epoch": 0.3165696679219445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5354,
      "step": 9247
    },
    {
      "epoch": 0.3166039027730229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5464,
      "step": 9248
    },
    {
      "epoch": 0.3166381376241013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5898,
      "step": 9249
    },
    {
      "epoch": 0.3166723724751797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5395,
      "step": 9250
    },
    {
      "epoch": 0.3167066073262581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5601,
      "step": 9251
    },
    {
      "epoch": 0.3167408421773365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.6172,
      "step": 9252
    },
    {
      "epoch": 0.31677507702841495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5939,
      "step": 9253
    },
    {
      "epoch": 0.31680931187949335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5891,
      "step": 9254
    },
    {
      "epoch": 0.31684354673057175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5359,
      "step": 9255
    },
    {
      "epoch": 0.31687778158165014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.545,
      "step": 9256
    },
    {
      "epoch": 0.31691201643272854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5674,
      "step": 9257
    },
    {
      "epoch": 0.31694625128380693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.567,
      "step": 9258
    },
    {
      "epoch": 0.31698048613488533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5486,
      "step": 9259
    },
    {
      "epoch": 0.3170147209859637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5175,
      "step": 9260
    },
    {
      "epoch": 0.3170489558370421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5056,
      "step": 9261
    },
    {
      "epoch": 0.3170831906881205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5181,
      "step": 9262
    },
    {
      "epoch": 0.3171174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4751,
      "step": 9263
    },
    {
      "epoch": 0.3171516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5111,
      "step": 9264
    },
    {
      "epoch": 0.3171858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.482,
      "step": 9265
    },
    {
      "epoch": 0.3172201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5061,
      "step": 9266
    },
    {
      "epoch": 0.3172543649435125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4751,
      "step": 9267
    },
    {
      "epoch": 0.3172885997945909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5076,
      "step": 9268
    },
    {
      "epoch": 0.3173228346456693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4258,
      "step": 9269
    },
    {
      "epoch": 0.3173570694967477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.46,
      "step": 9270
    },
    {
      "epoch": 0.3173913043478261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4698,
      "step": 9271
    },
    {
      "epoch": 0.3174255391989045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4413,
      "step": 9272
    },
    {
      "epoch": 0.3174597740499829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4414,
      "step": 9273
    },
    {
      "epoch": 0.3174940089010613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4568,
      "step": 9274
    },
    {
      "epoch": 0.31752824375213967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4599,
      "step": 9275
    },
    {
      "epoch": 0.31756247860321807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4213,
      "step": 9276
    },
    {
      "epoch": 0.31759671345429646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4636,
      "step": 9277
    },
    {
      "epoch": 0.31763094830537486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4474,
      "step": 9278
    },
    {
      "epoch": 0.31766518315645326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4492,
      "step": 9279
    },
    {
      "epoch": 0.31769941800753165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4483,
      "step": 9280
    },
    {
      "epoch": 0.31773365285861005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4738,
      "step": 9281
    },
    {
      "epoch": 0.31776788770968845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4678,
      "step": 9282
    },
    {
      "epoch": 0.31780212256076684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5647,
      "step": 9283
    },
    {
      "epoch": 0.31783635741184524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5753,
      "step": 9284
    },
    {
      "epoch": 0.31787059226292363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5431,
      "step": 9285
    },
    {
      "epoch": 0.31790482711400203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5051,
      "step": 9286
    },
    {
      "epoch": 0.3179390619650804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.5177,
      "step": 9287
    },
    {
      "epoch": 0.3179732968161588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.504,
      "step": 9288
    },
    {
      "epoch": 0.3180075316672372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4602,
      "step": 9289
    },
    {
      "epoch": 0.31804176651831567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4619,
      "step": 9290
    },
    {
      "epoch": 0.31807600136939407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.452,
      "step": 9291
    },
    {
      "epoch": 0.31811023622047246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4256,
      "step": 9292
    },
    {
      "epoch": 0.31814447107155086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4676,
      "step": 9293
    },
    {
      "epoch": 0.31817870592262926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4218,
      "step": 9294
    },
    {
      "epoch": 0.31821294077370765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.418,
      "step": 9295
    },
    {
      "epoch": 0.31824717562478605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.402,
      "step": 9296
    },
    {
      "epoch": 0.31828141047586445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4113,
      "step": 9297
    },
    {
      "epoch": 0.31831564532694284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4009,
      "step": 9298
    },
    {
      "epoch": 0.31834988017802124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3789,
      "step": 9299
    },
    {
      "epoch": 0.31838411502909963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3783,
      "step": 9300
    },
    {
      "epoch": 0.31841834988017803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.388,
      "step": 9301
    },
    {
      "epoch": 0.3184525847312564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.332,
      "step": 9302
    },
    {
      "epoch": 0.3184868195823348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3759,
      "step": 9303
    },
    {
      "epoch": 0.3185210544334132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3479,
      "step": 9304
    },
    {
      "epoch": 0.3185552892844916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3699,
      "step": 9305
    },
    {
      "epoch": 0.31858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3384,
      "step": 9306
    },
    {
      "epoch": 0.3186237589866484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3359,
      "step": 9307
    },
    {
      "epoch": 0.3186579938377268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3128,
      "step": 9308
    },
    {
      "epoch": 0.3186922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3707,
      "step": 9309
    },
    {
      "epoch": 0.3187264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3527,
      "step": 9310
    },
    {
      "epoch": 0.318760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3045,
      "step": 9311
    },
    {
      "epoch": 0.3187949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.33,
      "step": 9312
    },
    {
      "epoch": 0.3188291680931188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3201,
      "step": 9313
    },
    {
      "epoch": 0.3188634029441972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3837,
      "step": 9314
    },
    {
      "epoch": 0.3188976377952756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3741,
      "step": 9315
    },
    {
      "epoch": 0.318931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3543,
      "step": 9316
    },
    {
      "epoch": 0.31896610749743237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2981,
      "step": 9317
    },
    {
      "epoch": 0.31900034234851077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2984,
      "step": 9318
    },
    {
      "epoch": 0.31903457719958916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3099,
      "step": 9319
    },
    {
      "epoch": 0.31906881205066756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2959,
      "step": 9320
    },
    {
      "epoch": 0.31910304690174596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2601,
      "step": 9321
    },
    {
      "epoch": 0.31913728175282435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2949,
      "step": 9322
    },
    {
      "epoch": 0.31917151660390275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2785,
      "step": 9323
    },
    {
      "epoch": 0.31920575145498115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2289,
      "step": 9324
    },
    {
      "epoch": 0.31923998630605954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2898,
      "step": 9325
    },
    {
      "epoch": 0.319274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2721,
      "step": 9326
    },
    {
      "epoch": 0.3193084560082164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3117,
      "step": 9327
    },
    {
      "epoch": 0.3193426908592948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3571,
      "step": 9328
    },
    {
      "epoch": 0.3193769257103732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.319,
      "step": 9329
    },
    {
      "epoch": 0.3194111605614516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3508,
      "step": 9330
    },
    {
      "epoch": 0.31944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.391,
      "step": 9331
    },
    {
      "epoch": 0.31947963026360837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4223,
      "step": 9332
    },
    {
      "epoch": 0.31951386511468677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3754,
      "step": 9333
    },
    {
      "epoch": 0.31954809996576516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3914,
      "step": 9334
    },
    {
      "epoch": 0.31958233481684356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3601,
      "step": 9335
    },
    {
      "epoch": 0.31961656966792196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3882,
      "step": 9336
    },
    {
      "epoch": 0.31965080451900035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.365,
      "step": 9337
    },
    {
      "epoch": 0.31968503937007875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3143,
      "step": 9338
    },
    {
      "epoch": 0.31971927422115715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3236,
      "step": 9339
    },
    {
      "epoch": 0.31975350907223554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2817,
      "step": 9340
    },
    {
      "epoch": 0.31978774392331394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.285,
      "step": 9341
    },
    {
      "epoch": 0.31982197877439233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.265,
      "step": 9342
    },
    {
      "epoch": 0.31985621362547073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2325,
      "step": 9343
    },
    {
      "epoch": 0.3198904484765491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2496,
      "step": 9344
    },
    {
      "epoch": 0.3199246833276275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2332,
      "step": 9345
    },
    {
      "epoch": 0.3199589181787059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2398,
      "step": 9346
    },
    {
      "epoch": 0.3199931530297843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2174,
      "step": 9347
    },
    {
      "epoch": 0.3200273878808627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1879,
      "step": 9348
    },
    {
      "epoch": 0.3200616227319411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1913,
      "step": 9349
    },
    {
      "epoch": 0.3200958575830195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1798,
      "step": 9350
    },
    {
      "epoch": 0.3201300924340979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1626,
      "step": 9351
    },
    {
      "epoch": 0.3201643272851763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.16,
      "step": 9352
    },
    {
      "epoch": 0.3201985621362547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1555,
      "step": 9353
    },
    {
      "epoch": 0.3202327969873331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1596,
      "step": 9354
    },
    {
      "epoch": 0.3202670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1498,
      "step": 9355
    },
    {
      "epoch": 0.3203012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1425,
      "step": 9356
    },
    {
      "epoch": 0.3203355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1222,
      "step": 9357
    },
    {
      "epoch": 0.3203697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.152,
      "step": 9358
    },
    {
      "epoch": 0.32040397124272507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1403,
      "step": 9359
    },
    {
      "epoch": 0.32043820609380347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0852,
      "step": 9360
    },
    {
      "epoch": 0.32047244094488186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1305,
      "step": 9361
    },
    {
      "epoch": 0.32050667579596026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0995,
      "step": 9362
    },
    {
      "epoch": 0.3205409106470387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0761,
      "step": 9363
    },
    {
      "epoch": 0.3205751454981171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0783,
      "step": 9364
    },
    {
      "epoch": 0.3206093803491955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0845,
      "step": 9365
    },
    {
      "epoch": 0.3206436152002739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0785,
      "step": 9366
    },
    {
      "epoch": 0.3206778500513523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0541,
      "step": 9367
    },
    {
      "epoch": 0.3207120849024307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0382,
      "step": 9368
    },
    {
      "epoch": 0.3207463197535091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0383,
      "step": 9369
    },
    {
      "epoch": 0.3207805546045875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0568,
      "step": 9370
    },
    {
      "epoch": 0.3208147894556659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0391,
      "step": 9371
    },
    {
      "epoch": 0.3208490243067443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0222,
      "step": 9372
    },
    {
      "epoch": 0.3208832591578227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0318,
      "step": 9373
    },
    {
      "epoch": 0.32091749400890107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0379,
      "step": 9374
    },
    {
      "epoch": 0.32095172885997947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0076,
      "step": 9375
    },
    {
      "epoch": 0.32098596371105786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0348,
      "step": 9376
    },
    {
      "epoch": 0.32102019856213626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9881,
      "step": 9377
    },
    {
      "epoch": 0.32105443341321466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0102,
      "step": 9378
    },
    {
      "epoch": 0.32108866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9827,
      "step": 9379
    },
    {
      "epoch": 0.32112290311537145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9858,
      "step": 9380
    },
    {
      "epoch": 0.32115713796644985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9629,
      "step": 9381
    },
    {
      "epoch": 0.32119137281752824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9889,
      "step": 9382
    },
    {
      "epoch": 0.32122560766860664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9605,
      "step": 9383
    },
    {
      "epoch": 0.32125984251968503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9591,
      "step": 9384
    },
    {
      "epoch": 0.32129407737076343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9601,
      "step": 9385
    },
    {
      "epoch": 0.3213283122218418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9213,
      "step": 9386
    },
    {
      "epoch": 0.3213625470729202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9294,
      "step": 9387
    },
    {
      "epoch": 0.3213967819239986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8707,
      "step": 9388
    },
    {
      "epoch": 0.321431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9257,
      "step": 9389
    },
    {
      "epoch": 0.3214652516261554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9164,
      "step": 9390
    },
    {
      "epoch": 0.3214994864772338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9175,
      "step": 9391
    },
    {
      "epoch": 0.3215337213283122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9215,
      "step": 9392
    },
    {
      "epoch": 0.3215679561793906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9303,
      "step": 9393
    },
    {
      "epoch": 0.321602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1056,
      "step": 9394
    },
    {
      "epoch": 0.3216364258815474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0031,
      "step": 9395
    },
    {
      "epoch": 0.3216706607326258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1769,
      "step": 9396
    },
    {
      "epoch": 0.3217048955837042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2585,
      "step": 9397
    },
    {
      "epoch": 0.3217391304347826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4497,
      "step": 9398
    },
    {
      "epoch": 0.321773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.4192,
      "step": 9399
    },
    {
      "epoch": 0.32180760013693943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.3696,
      "step": 9400
    },
    {
      "epoch": 0.3218418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.311,
      "step": 9401
    },
    {
      "epoch": 0.3218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2959,
      "step": 9402
    },
    {
      "epoch": 0.3219103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2774,
      "step": 9403
    },
    {
      "epoch": 0.321944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2402,
      "step": 9404
    },
    {
      "epoch": 0.3219787743923314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.2128,
      "step": 9405
    },
    {
      "epoch": 0.3220130092434098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1536,
      "step": 9406
    },
    {
      "epoch": 0.3220472440944882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1376,
      "step": 9407
    },
    {
      "epoch": 0.3220814789455666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1024,
      "step": 9408
    },
    {
      "epoch": 0.322115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0805,
      "step": 9409
    },
    {
      "epoch": 0.3221499486477234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.1022,
      "step": 9410
    },
    {
      "epoch": 0.3221841834988018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0667,
      "step": 9411
    },
    {
      "epoch": 0.3222184183498802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 3.0007,
      "step": 9412
    },
    {
      "epoch": 0.3222526532009586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9969,
      "step": 9413
    },
    {
      "epoch": 0.322286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9684,
      "step": 9414
    },
    {
      "epoch": 0.3223211229031154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9337,
      "step": 9415
    },
    {
      "epoch": 0.32235535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9118,
      "step": 9416
    },
    {
      "epoch": 0.32238959260527217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9135,
      "step": 9417
    },
    {
      "epoch": 0.32242382745635056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9502,
      "step": 9418
    },
    {
      "epoch": 0.32245806230742896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.9144,
      "step": 9419
    },
    {
      "epoch": 0.32249229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.918,
      "step": 9420
    },
    {
      "epoch": 0.32252653200958575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8869,
      "step": 9421
    },
    {
      "epoch": 0.32256076686066415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8802,
      "step": 9422
    },
    {
      "epoch": 0.32259500171174255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8844,
      "step": 9423
    },
    {
      "epoch": 0.32262923656282094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8771,
      "step": 9424
    },
    {
      "epoch": 0.32266347141389934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8289,
      "step": 9425
    },
    {
      "epoch": 0.32269770626497774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8218,
      "step": 9426
    },
    {
      "epoch": 0.32273194111605613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8336,
      "step": 9427
    },
    {
      "epoch": 0.3227661759671345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8149,
      "step": 9428
    },
    {
      "epoch": 0.3228004108182129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8484,
      "step": 9429
    },
    {
      "epoch": 0.3228346456692913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8489,
      "step": 9430
    },
    {
      "epoch": 0.3228688805203697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8154,
      "step": 9431
    },
    {
      "epoch": 0.3229031153714481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7946,
      "step": 9432
    },
    {
      "epoch": 0.3229373502225265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.8025,
      "step": 9433
    },
    {
      "epoch": 0.3229715850736049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7529,
      "step": 9434
    },
    {
      "epoch": 0.3230058199246833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7828,
      "step": 9435
    },
    {
      "epoch": 0.3230400547757617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7654,
      "step": 9436
    },
    {
      "epoch": 0.32307428962684015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7693,
      "step": 9437
    },
    {
      "epoch": 0.32310852447791855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7335,
      "step": 9438
    },
    {
      "epoch": 0.32314275932899694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7238,
      "step": 9439
    },
    {
      "epoch": 0.32317699418007534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7607,
      "step": 9440
    },
    {
      "epoch": 0.32321122903115374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7185,
      "step": 9441
    },
    {
      "epoch": 0.32324546388223213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7403,
      "step": 9442
    },
    {
      "epoch": 0.3232796987333105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7218,
      "step": 9443
    },
    {
      "epoch": 0.3233139335843889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7303,
      "step": 9444
    },
    {
      "epoch": 0.3233481684354673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7192,
      "step": 9445
    },
    {
      "epoch": 0.3233824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7156,
      "step": 9446
    },
    {
      "epoch": 0.3234166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7045,
      "step": 9447
    },
    {
      "epoch": 0.3234508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7073,
      "step": 9448
    },
    {
      "epoch": 0.3234851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6974,
      "step": 9449
    },
    {
      "epoch": 0.3235193426908593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.689,
      "step": 9450
    },
    {
      "epoch": 0.3235535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6595,
      "step": 9451
    },
    {
      "epoch": 0.3235878123930161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6628,
      "step": 9452
    },
    {
      "epoch": 0.3236220472440945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.7074,
      "step": 9453
    },
    {
      "epoch": 0.3236562820951729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6609,
      "step": 9454
    },
    {
      "epoch": 0.3236905169462513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.707,
      "step": 9455
    },
    {
      "epoch": 0.3237247517973297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6656,
      "step": 9456
    },
    {
      "epoch": 0.3237589866484081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6383,
      "step": 9457
    },
    {
      "epoch": 0.32379322149948647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6519,
      "step": 9458
    },
    {
      "epoch": 0.32382745635056487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6645,
      "step": 9459
    },
    {
      "epoch": 0.32386169120164326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6151,
      "step": 9460
    },
    {
      "epoch": 0.32389592605272166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6638,
      "step": 9461
    },
    {
      "epoch": 0.32393016090380006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6547,
      "step": 9462
    },
    {
      "epoch": 0.32396439575487845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6564,
      "step": 9463
    },
    {
      "epoch": 0.32399863060595685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6506,
      "step": 9464
    },
    {
      "epoch": 0.32403286545703525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5993,
      "step": 9465
    },
    {
      "epoch": 0.32406710030811364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.63,
      "step": 9466
    },
    {
      "epoch": 0.32410133515919204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6535,
      "step": 9467
    },
    {
      "epoch": 0.32413557001027044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6233,
      "step": 9468
    },
    {
      "epoch": 0.32416980486134883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6089,
      "step": 9469
    },
    {
      "epoch": 0.3242040397124272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5836,
      "step": 9470
    },
    {
      "epoch": 0.3242382745635056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5887,
      "step": 9471
    },
    {
      "epoch": 0.324272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6042,
      "step": 9472
    },
    {
      "epoch": 0.3243067442656624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6127,
      "step": 9473
    },
    {
      "epoch": 0.32434097911674087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6584,
      "step": 9474
    },
    {
      "epoch": 0.32437521396781926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.679,
      "step": 9475
    },
    {
      "epoch": 0.32440944881889766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6982,
      "step": 9476
    },
    {
      "epoch": 0.32444368366997606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6776,
      "step": 9477
    },
    {
      "epoch": 0.32447791852105445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6708,
      "step": 9478
    },
    {
      "epoch": 0.32451215337213285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6625,
      "step": 9479
    },
    {
      "epoch": 0.32454638822321125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6138,
      "step": 9480
    },
    {
      "epoch": 0.32458062307428964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6532,
      "step": 9481
    },
    {
      "epoch": 0.32461485792536804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6482,
      "step": 9482
    },
    {
      "epoch": 0.32464909277644644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.607,
      "step": 9483
    },
    {
      "epoch": 0.32468332762752483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.6269,
      "step": 9484
    },
    {
      "epoch": 0.3247175624786032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.592,
      "step": 9485
    },
    {
      "epoch": 0.3247517973296816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5621,
      "step": 9486
    },
    {
      "epoch": 0.32478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.633,
      "step": 9487
    },
    {
      "epoch": 0.3248202670318384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5747,
      "step": 9488
    },
    {
      "epoch": 0.3248545018829168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5832,
      "step": 9489
    },
    {
      "epoch": 0.3248887367339952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5334,
      "step": 9490
    },
    {
      "epoch": 0.3249229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5564,
      "step": 9491
    },
    {
      "epoch": 0.324957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5479,
      "step": 9492
    },
    {
      "epoch": 0.3249914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.54,
      "step": 9493
    },
    {
      "epoch": 0.3250256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5372,
      "step": 9494
    },
    {
      "epoch": 0.3250599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5354,
      "step": 9495
    },
    {
      "epoch": 0.3250941458404656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4851,
      "step": 9496
    },
    {
      "epoch": 0.325128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5339,
      "step": 9497
    },
    {
      "epoch": 0.3251626155426224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5044,
      "step": 9498
    },
    {
      "epoch": 0.3251968503937008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4926,
      "step": 9499
    },
    {
      "epoch": 0.3252310852447792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4963,
      "step": 9500
    },
    {
      "epoch": 0.32526532009585757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4906,
      "step": 9501
    },
    {
      "epoch": 0.32529955494693596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.5026,
      "step": 9502
    },
    {
      "epoch": 0.32533378979801436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4719,
      "step": 9503
    },
    {
      "epoch": 0.32536802464909276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4984,
      "step": 9504
    },
    {
      "epoch": 0.32540225950017115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4782,
      "step": 9505
    },
    {
      "epoch": 0.32543649435124955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4582,
      "step": 9506
    },
    {
      "epoch": 0.32547072920232795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4765,
      "step": 9507
    },
    {
      "epoch": 0.32550496405340634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4929,
      "step": 9508
    },
    {
      "epoch": 0.32553919890448474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4464,
      "step": 9509
    },
    {
      "epoch": 0.3255734337555632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4207,
      "step": 9510
    },
    {
      "epoch": 0.3256076686066416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4414,
      "step": 9511
    },
    {
      "epoch": 0.32564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4607,
      "step": 9512
    },
    {
      "epoch": 0.3256761383087984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4726,
      "step": 9513
    },
    {
      "epoch": 0.3257103731598768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4435,
      "step": 9514
    },
    {
      "epoch": 0.3257446080109552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4472,
      "step": 9515
    },
    {
      "epoch": 0.32577884286203357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4529,
      "step": 9516
    },
    {
      "epoch": 0.32581307771311196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4252,
      "step": 9517
    },
    {
      "epoch": 0.32584731256419036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3928,
      "step": 9518
    },
    {
      "epoch": 0.32588154741526876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4043,
      "step": 9519
    },
    {
      "epoch": 0.32591578226634715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.41,
      "step": 9520
    },
    {
      "epoch": 0.32595001711742555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4173,
      "step": 9521
    },
    {
      "epoch": 0.32598425196850395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4031,
      "step": 9522
    },
    {
      "epoch": 0.32601848681958234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4012,
      "step": 9523
    },
    {
      "epoch": 0.32605272167066074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4035,
      "step": 9524
    },
    {
      "epoch": 0.32608695652173914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4165,
      "step": 9525
    },
    {
      "epoch": 0.32612119137281753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4211,
      "step": 9526
    },
    {
      "epoch": 0.32615542622389593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3946,
      "step": 9527
    },
    {
      "epoch": 0.3261896610749743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3644,
      "step": 9528
    },
    {
      "epoch": 0.3262238959260527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3772,
      "step": 9529
    },
    {
      "epoch": 0.3262581307771311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4087,
      "step": 9530
    },
    {
      "epoch": 0.3262923656282095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4154,
      "step": 9531
    },
    {
      "epoch": 0.3263266004792879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3586,
      "step": 9532
    },
    {
      "epoch": 0.3263608353303663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.4007,
      "step": 9533
    },
    {
      "epoch": 0.3263950701814447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3517,
      "step": 9534
    },
    {
      "epoch": 0.3264293050325231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3678,
      "step": 9535
    },
    {
      "epoch": 0.3264635398836015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3581,
      "step": 9536
    },
    {
      "epoch": 0.3264977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.389,
      "step": 9537
    },
    {
      "epoch": 0.3265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3824,
      "step": 9538
    },
    {
      "epoch": 0.3265662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3954,
      "step": 9539
    },
    {
      "epoch": 0.3266004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3877,
      "step": 9540
    },
    {
      "epoch": 0.3266347141389935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3492,
      "step": 9541
    },
    {
      "epoch": 0.3266689489900719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3601,
      "step": 9542
    },
    {
      "epoch": 0.32670318384115027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3528,
      "step": 9543
    },
    {
      "epoch": 0.32673741869222866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3535,
      "step": 9544
    },
    {
      "epoch": 0.32677165354330706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3555,
      "step": 9545
    },
    {
      "epoch": 0.32680588839438546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3468,
      "step": 9546
    },
    {
      "epoch": 0.3268401232454639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2938,
      "step": 9547
    },
    {
      "epoch": 0.3268743580965423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3322,
      "step": 9548
    },
    {
      "epoch": 0.3269085929476207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3482,
      "step": 9549
    },
    {
      "epoch": 0.3269428277986991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3509,
      "step": 9550
    },
    {
      "epoch": 0.3269770626497775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3382,
      "step": 9551
    },
    {
      "epoch": 0.3270112975008559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3433,
      "step": 9552
    },
    {
      "epoch": 0.3270455323519343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3461,
      "step": 9553
    },
    {
      "epoch": 0.3270797672030127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3346,
      "step": 9554
    },
    {
      "epoch": 0.3271140020540911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3233,
      "step": 9555
    },
    {
      "epoch": 0.3271482369051695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3498,
      "step": 9556
    },
    {
      "epoch": 0.3271824717562479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3336,
      "step": 9557
    },
    {
      "epoch": 0.32721670660732627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3015,
      "step": 9558
    },
    {
      "epoch": 0.32725094145840466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3074,
      "step": 9559
    },
    {
      "epoch": 0.32728517630948306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.277,
      "step": 9560
    },
    {
      "epoch": 0.32731941116056146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.302,
      "step": 9561
    },
    {
      "epoch": 0.32735364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3058,
      "step": 9562
    },
    {
      "epoch": 0.32738788086271825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2793,
      "step": 9563
    },
    {
      "epoch": 0.32742211571379665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3432,
      "step": 9564
    },
    {
      "epoch": 0.32745635056487504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.3164,
      "step": 9565
    },
    {
      "epoch": 0.32749058541595344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2738,
      "step": 9566
    },
    {
      "epoch": 0.32752482026703184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2861,
      "step": 9567
    },
    {
      "epoch": 0.32755905511811023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2713,
      "step": 9568
    },
    {
      "epoch": 0.32759328996918863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2804,
      "step": 9569
    },
    {
      "epoch": 0.327627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2549,
      "step": 9570
    },
    {
      "epoch": 0.3276617596713454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2216,
      "step": 9571
    },
    {
      "epoch": 0.3276959945224238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2368,
      "step": 9572
    },
    {
      "epoch": 0.3277302293735022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2497,
      "step": 9573
    },
    {
      "epoch": 0.3277644642245806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2758,
      "step": 9574
    },
    {
      "epoch": 0.327798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2663,
      "step": 9575
    },
    {
      "epoch": 0.3278329339267374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2691,
      "step": 9576
    },
    {
      "epoch": 0.3278671687778158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2268,
      "step": 9577
    },
    {
      "epoch": 0.3279014036288942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2307,
      "step": 9578
    },
    {
      "epoch": 0.3279356384799726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2448,
      "step": 9579
    },
    {
      "epoch": 0.327969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2523,
      "step": 9580
    },
    {
      "epoch": 0.3280041081821294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2238,
      "step": 9581
    },
    {
      "epoch": 0.3280383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2103,
      "step": 9582
    },
    {
      "epoch": 0.3280725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2185,
      "step": 9583
    },
    {
      "epoch": 0.32810681273536463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2115,
      "step": 9584
    },
    {
      "epoch": 0.328141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2029,
      "step": 9585
    },
    {
      "epoch": 0.3281752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2117,
      "step": 9586
    },
    {
      "epoch": 0.3282095172885998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2321,
      "step": 9587
    },
    {
      "epoch": 0.3282437521396782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.182,
      "step": 9588
    },
    {
      "epoch": 0.3282779869907566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1782,
      "step": 9589
    },
    {
      "epoch": 0.328312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2027,
      "step": 9590
    },
    {
      "epoch": 0.3283464566929134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2169,
      "step": 9591
    },
    {
      "epoch": 0.3283806915439918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1949,
      "step": 9592
    },
    {
      "epoch": 0.3284149263950702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1977,
      "step": 9593
    },
    {
      "epoch": 0.3284491612461486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1633,
      "step": 9594
    },
    {
      "epoch": 0.328483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1591,
      "step": 9595
    },
    {
      "epoch": 0.3285176309483054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1707,
      "step": 9596
    },
    {
      "epoch": 0.3285518657993838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1392,
      "step": 9597
    },
    {
      "epoch": 0.3285861006504622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1381,
      "step": 9598
    },
    {
      "epoch": 0.3286203355015406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.166,
      "step": 9599
    },
    {
      "epoch": 0.32865457035261897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1721,
      "step": 9600
    },
    {
      "epoch": 0.32868880520369737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1195,
      "step": 9601
    },
    {
      "epoch": 0.32872304005477576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1434,
      "step": 9602
    },
    {
      "epoch": 0.32875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1251,
      "step": 9603
    },
    {
      "epoch": 0.32879150975693255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1455,
      "step": 9604
    },
    {
      "epoch": 0.32882574460801095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1558,
      "step": 9605
    },
    {
      "epoch": 0.32885997945908935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1187,
      "step": 9606
    },
    {
      "epoch": 0.32889421431016774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1333,
      "step": 9607
    },
    {
      "epoch": 0.32892844916124614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.085,
      "step": 9608
    },
    {
      "epoch": 0.32896268401232454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1114,
      "step": 9609
    },
    {
      "epoch": 0.32899691886340293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0965,
      "step": 9610
    },
    {
      "epoch": 0.32903115371448133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1238,
      "step": 9611
    },
    {
      "epoch": 0.3290653885655597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2002,
      "step": 9612
    },
    {
      "epoch": 0.3290996234166381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.2659,
      "step": 9613
    },
    {
      "epoch": 0.3291338582677165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1554,
      "step": 9614
    },
    {
      "epoch": 0.3291680931187949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.161,
      "step": 9615
    },
    {
      "epoch": 0.3292023279698733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1503,
      "step": 9616
    },
    {
      "epoch": 0.3292365628209517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1387,
      "step": 9617
    },
    {
      "epoch": 0.3292707976720301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1067,
      "step": 9618
    },
    {
      "epoch": 0.3293050325231085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.1035,
      "step": 9619
    },
    {
      "epoch": 0.3293392673741869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0852,
      "step": 9620
    },
    {
      "epoch": 0.32937350222526535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0796,
      "step": 9621
    },
    {
      "epoch": 0.32940773707634374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0539,
      "step": 9622
    },
    {
      "epoch": 0.32944197192742214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0782,
      "step": 9623
    },
    {
      "epoch": 0.32947620677850054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.06,
      "step": 9624
    },
    {
      "epoch": 0.32951044162957893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0352,
      "step": 9625
    },
    {
      "epoch": 0.32954467648065733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0329,
      "step": 9626
    },
    {
      "epoch": 0.3295789113317357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0211,
      "step": 9627
    },
    {
      "epoch": 0.3296131461828141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0057,
      "step": 9628
    },
    {
      "epoch": 0.3296473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0193,
      "step": 9629
    },
    {
      "epoch": 0.3296816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9737,
      "step": 9630
    },
    {
      "epoch": 0.3297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 2.0271,
      "step": 9631
    },
    {
      "epoch": 0.3297500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9708,
      "step": 9632
    },
    {
      "epoch": 0.3297843204382061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9705,
      "step": 9633
    },
    {
      "epoch": 0.3298185552892845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9304,
      "step": 9634
    },
    {
      "epoch": 0.3298527901403629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9474,
      "step": 9635
    },
    {
      "epoch": 0.3298870249914413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9506,
      "step": 9636
    },
    {
      "epoch": 0.3299212598425197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9333,
      "step": 9637
    },
    {
      "epoch": 0.3299554946935981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9333,
      "step": 9638
    },
    {
      "epoch": 0.3299897295446765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9303,
      "step": 9639
    },
    {
      "epoch": 0.3300239643957549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9185,
      "step": 9640
    },
    {
      "epoch": 0.3300581992468333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9195,
      "step": 9641
    },
    {
      "epoch": 0.33009243409791167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9194,
      "step": 9642
    },
    {
      "epoch": 0.33012666894899007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9179,
      "step": 9643
    },
    {
      "epoch": 0.33016090380006846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8933,
      "step": 9644
    },
    {
      "epoch": 0.33019513865114686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9113,
      "step": 9645
    },
    {
      "epoch": 0.33022937350222525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8821,
      "step": 9646
    },
    {
      "epoch": 0.33026360835330365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.9077,
      "step": 9647
    },
    {
      "epoch": 0.33029784320438205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8925,
      "step": 9648
    },
    {
      "epoch": 0.33033207805546044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8886,
      "step": 9649
    },
    {
      "epoch": 0.33036631290653884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8667,
      "step": 9650
    },
    {
      "epoch": 0.33040054775761724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8509,
      "step": 9651
    },
    {
      "epoch": 0.33043478260869563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8627,
      "step": 9652
    },
    {
      "epoch": 0.33046901745977403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8473,
      "step": 9653
    },
    {
      "epoch": 0.3305032523108524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8456,
      "step": 9654
    },
    {
      "epoch": 0.3305374871619308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8251,
      "step": 9655
    },
    {
      "epoch": 0.3305717220130092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8485,
      "step": 9656
    },
    {
      "epoch": 0.33060595686408767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8642,
      "step": 9657
    },
    {
      "epoch": 0.33064019171516607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8302,
      "step": 9658
    },
    {
      "epoch": 0.33067442656624446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8025,
      "step": 9659
    },
    {
      "epoch": 0.33070866141732286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8408,
      "step": 9660
    },
    {
      "epoch": 0.33074289626840125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8132,
      "step": 9661
    },
    {
      "epoch": 0.33077713111947965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8407,
      "step": 9662
    },
    {
      "epoch": 0.33081136597055805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8309,
      "step": 9663
    },
    {
      "epoch": 0.33084560082163644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8055,
      "step": 9664
    },
    {
      "epoch": 0.33087983567271484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8221,
      "step": 9665
    },
    {
      "epoch": 0.33091407052379324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8307,
      "step": 9666
    },
    {
      "epoch": 0.33094830537487163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7798,
      "step": 9667
    },
    {
      "epoch": 0.33098254022595003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8165,
      "step": 9668
    },
    {
      "epoch": 0.3310167750770284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8021,
      "step": 9669
    },
    {
      "epoch": 0.3310510099281068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7914,
      "step": 9670
    },
    {
      "epoch": 0.3310852447791852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8107,
      "step": 9671
    },
    {
      "epoch": 0.3311194796302636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7929,
      "step": 9672
    },
    {
      "epoch": 0.331153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8001,
      "step": 9673
    },
    {
      "epoch": 0.3311879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.784,
      "step": 9674
    },
    {
      "epoch": 0.3312221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7816,
      "step": 9675
    },
    {
      "epoch": 0.3312564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.764,
      "step": 9676
    },
    {
      "epoch": 0.3312906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7732,
      "step": 9677
    },
    {
      "epoch": 0.331324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7636,
      "step": 9678
    },
    {
      "epoch": 0.3313591235878124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7556,
      "step": 9679
    },
    {
      "epoch": 0.3313933584388908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7604,
      "step": 9680
    },
    {
      "epoch": 0.3314275932899692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7525,
      "step": 9681
    },
    {
      "epoch": 0.3314618281410476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7467,
      "step": 9682
    },
    {
      "epoch": 0.331496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7495,
      "step": 9683
    },
    {
      "epoch": 0.33153029784320437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7424,
      "step": 9684
    },
    {
      "epoch": 0.33156453269428277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7381,
      "step": 9685
    },
    {
      "epoch": 0.33159876754536116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7201,
      "step": 9686
    },
    {
      "epoch": 0.33163300239643956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7669,
      "step": 9687
    },
    {
      "epoch": 0.33166723724751795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.765,
      "step": 9688
    },
    {
      "epoch": 0.33170147209859635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7378,
      "step": 9689
    },
    {
      "epoch": 0.33173570694967475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7555,
      "step": 9690
    },
    {
      "epoch": 0.33176994180075314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7618,
      "step": 9691
    },
    {
      "epoch": 0.33180417665183154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7432,
      "step": 9692
    },
    {
      "epoch": 0.33183841150290994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7283,
      "step": 9693
    },
    {
      "epoch": 0.3318726463539884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7384,
      "step": 9694
    },
    {
      "epoch": 0.3319068812050668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7406,
      "step": 9695
    },
    {
      "epoch": 0.3319411160561452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7384,
      "step": 9696
    },
    {
      "epoch": 0.3319753509072236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7336,
      "step": 9697
    },
    {
      "epoch": 0.332009585758302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7439,
      "step": 9698
    },
    {
      "epoch": 0.33204382060938037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7167,
      "step": 9699
    },
    {
      "epoch": 0.33207805546045877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7154,
      "step": 9700
    },
    {
      "epoch": 0.33211229031153716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7001,
      "step": 9701
    },
    {
      "epoch": 0.33214652516261556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7022,
      "step": 9702
    },
    {
      "epoch": 0.33218076001369395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7017,
      "step": 9703
    },
    {
      "epoch": 0.33221499486477235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6878,
      "step": 9704
    },
    {
      "epoch": 0.33224922971585075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6871,
      "step": 9705
    },
    {
      "epoch": 0.33228346456692914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6766,
      "step": 9706
    },
    {
      "epoch": 0.33231769941800754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7052,
      "step": 9707
    },
    {
      "epoch": 0.33235193426908594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6849,
      "step": 9708
    },
    {
      "epoch": 0.33238616912016433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7063,
      "step": 9709
    },
    {
      "epoch": 0.33242040397124273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6813,
      "step": 9710
    },
    {
      "epoch": 0.3324546388223211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7267,
      "step": 9711
    },
    {
      "epoch": 0.3324888736733995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7017,
      "step": 9712
    },
    {
      "epoch": 0.3325231085244779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7117,
      "step": 9713
    },
    {
      "epoch": 0.3325573433755563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7134,
      "step": 9714
    },
    {
      "epoch": 0.3325915782266347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7019,
      "step": 9715
    },
    {
      "epoch": 0.3326258130777131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6968,
      "step": 9716
    },
    {
      "epoch": 0.3326600479287915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7011,
      "step": 9717
    },
    {
      "epoch": 0.3326942827798699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6834,
      "step": 9718
    },
    {
      "epoch": 0.3327285176309483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.7052,
      "step": 9719
    },
    {
      "epoch": 0.3327627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6998,
      "step": 9720
    },
    {
      "epoch": 0.3327969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6783,
      "step": 9721
    },
    {
      "epoch": 0.3328312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6915,
      "step": 9722
    },
    {
      "epoch": 0.3328654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6921,
      "step": 9723
    },
    {
      "epoch": 0.3328996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6962,
      "step": 9724
    },
    {
      "epoch": 0.3329339267374187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6876,
      "step": 9725
    },
    {
      "epoch": 0.33296816158849707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.662,
      "step": 9726
    },
    {
      "epoch": 0.33300239643957547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6858,
      "step": 9727
    },
    {
      "epoch": 0.33303663129065386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6494,
      "step": 9728
    },
    {
      "epoch": 0.33307086614173226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6552,
      "step": 9729
    },
    {
      "epoch": 0.33310510099281065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6662,
      "step": 9730
    },
    {
      "epoch": 0.3331393358438891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6764,
      "step": 9731
    },
    {
      "epoch": 0.3331735706949675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.649,
      "step": 9732
    },
    {
      "epoch": 0.3332078055460459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6836,
      "step": 9733
    },
    {
      "epoch": 0.3332420403971243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6573,
      "step": 9734
    },
    {
      "epoch": 0.3332762752482027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6678,
      "step": 9735
    },
    {
      "epoch": 0.3333105100992811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.707,
      "step": 9736
    },
    {
      "epoch": 0.3333447449503595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6645,
      "step": 9737
    },
    {
      "epoch": 0.3333789798014379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6682,
      "step": 9738
    },
    {
      "epoch": 0.3334132146525163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6577,
      "step": 9739
    },
    {
      "epoch": 0.3334474495035947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6744,
      "step": 9740
    },
    {
      "epoch": 0.33348168435467307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6684,
      "step": 9741
    },
    {
      "epoch": 0.33351591920575147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6783,
      "step": 9742
    },
    {
      "epoch": 0.33355015405682986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6724,
      "step": 9743
    },
    {
      "epoch": 0.33358438890790826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6282,
      "step": 9744
    },
    {
      "epoch": 0.33361862375898665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6564,
      "step": 9745
    },
    {
      "epoch": 0.33365285861006505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.652,
      "step": 9746
    },
    {
      "epoch": 0.33368709346114345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6464,
      "step": 9747
    },
    {
      "epoch": 0.33372132831222184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6824,
      "step": 9748
    },
    {
      "epoch": 0.33375556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6501,
      "step": 9749
    },
    {
      "epoch": 0.33378979801437864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6414,
      "step": 9750
    },
    {
      "epoch": 0.33382403286545703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6652,
      "step": 9751
    },
    {
      "epoch": 0.33385826771653543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6679,
      "step": 9752
    },
    {
      "epoch": 0.3338925025676138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6594,
      "step": 9753
    },
    {
      "epoch": 0.3339267374186922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.657,
      "step": 9754
    },
    {
      "epoch": 0.3339609722697706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6714,
      "step": 9755
    },
    {
      "epoch": 0.333995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.662,
      "step": 9756
    },
    {
      "epoch": 0.3340294419719274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6493,
      "step": 9757
    },
    {
      "epoch": 0.3340636768230058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6523,
      "step": 9758
    },
    {
      "epoch": 0.3340979116740842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6232,
      "step": 9759
    },
    {
      "epoch": 0.3341321465251626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6528,
      "step": 9760
    },
    {
      "epoch": 0.334166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6336,
      "step": 9761
    },
    {
      "epoch": 0.3342006162273194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6335,
      "step": 9762
    },
    {
      "epoch": 0.3342348510783978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6294,
      "step": 9763
    },
    {
      "epoch": 0.3342690859294762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6436,
      "step": 9764
    },
    {
      "epoch": 0.3343033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6415,
      "step": 9765
    },
    {
      "epoch": 0.334337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6454,
      "step": 9766
    },
    {
      "epoch": 0.3343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6307,
      "step": 9767
    },
    {
      "epoch": 0.3344060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6567,
      "step": 9768
    },
    {
      "epoch": 0.3344402601848682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6503,
      "step": 9769
    },
    {
      "epoch": 0.3344744950359466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.625,
      "step": 9770
    },
    {
      "epoch": 0.334508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6225,
      "step": 9771
    },
    {
      "epoch": 0.3345429647381034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6136,
      "step": 9772
    },
    {
      "epoch": 0.3345771995891818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6155,
      "step": 9773
    },
    {
      "epoch": 0.3346114344402602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6401,
      "step": 9774
    },
    {
      "epoch": 0.3346456692913386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6101,
      "step": 9775
    },
    {
      "epoch": 0.334679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6454,
      "step": 9776
    },
    {
      "epoch": 0.3347141389934954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6297,
      "step": 9777
    },
    {
      "epoch": 0.3347483738445738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6414,
      "step": 9778
    },
    {
      "epoch": 0.3347826086956522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6084,
      "step": 9779
    },
    {
      "epoch": 0.3348168435467306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6202,
      "step": 9780
    },
    {
      "epoch": 0.334851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6028,
      "step": 9781
    },
    {
      "epoch": 0.3348853132488874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6333,
      "step": 9782
    },
    {
      "epoch": 0.33491954809996577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6166,
      "step": 9783
    },
    {
      "epoch": 0.33495378295104417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6138,
      "step": 9784
    },
    {
      "epoch": 0.33498801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6415,
      "step": 9785
    },
    {
      "epoch": 0.33502225265320096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6279,
      "step": 9786
    },
    {
      "epoch": 0.33505648750427935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6222,
      "step": 9787
    },
    {
      "epoch": 0.33509072235535775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6189,
      "step": 9788
    },
    {
      "epoch": 0.33512495720643615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6281,
      "step": 9789
    },
    {
      "epoch": 0.33515919205751454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6287,
      "step": 9790
    },
    {
      "epoch": 0.33519342690859294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6279,
      "step": 9791
    },
    {
      "epoch": 0.33522766175967134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.62,
      "step": 9792
    },
    {
      "epoch": 0.33526189661074973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6299,
      "step": 9793
    },
    {
      "epoch": 0.33529613146182813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5948,
      "step": 9794
    },
    {
      "epoch": 0.3353303663129065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5882,
      "step": 9795
    },
    {
      "epoch": 0.3353646011639849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.625,
      "step": 9796
    },
    {
      "epoch": 0.3353988360150633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6204,
      "step": 9797
    },
    {
      "epoch": 0.3354330708661417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6305,
      "step": 9798
    },
    {
      "epoch": 0.3354673057172201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6023,
      "step": 9799
    },
    {
      "epoch": 0.3355015405682985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6175,
      "step": 9800
    },
    {
      "epoch": 0.3355357754193769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6228,
      "step": 9801
    },
    {
      "epoch": 0.3355700102704553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6119,
      "step": 9802
    },
    {
      "epoch": 0.3356042451215337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6233,
      "step": 9803
    },
    {
      "epoch": 0.3356384799726121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5951,
      "step": 9804
    },
    {
      "epoch": 0.33567271482369054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6417,
      "step": 9805
    },
    {
      "epoch": 0.33570694967476894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5995,
      "step": 9806
    },
    {
      "epoch": 0.33574118452584734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6222,
      "step": 9807
    },
    {
      "epoch": 0.33577541937692573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6083,
      "step": 9808
    },
    {
      "epoch": 0.33580965422800413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6022,
      "step": 9809
    },
    {
      "epoch": 0.3358438890790825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6044,
      "step": 9810
    },
    {
      "epoch": 0.3358781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5955,
      "step": 9811
    },
    {
      "epoch": 0.3359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6023,
      "step": 9812
    },
    {
      "epoch": 0.3359465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6167,
      "step": 9813
    },
    {
      "epoch": 0.3359808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6064,
      "step": 9814
    },
    {
      "epoch": 0.3360150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6044,
      "step": 9815
    },
    {
      "epoch": 0.3360492981855529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.629,
      "step": 9816
    },
    {
      "epoch": 0.3360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5964,
      "step": 9817
    },
    {
      "epoch": 0.3361177678877097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6278,
      "step": 9818
    },
    {
      "epoch": 0.3361520027387881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6095,
      "step": 9819
    },
    {
      "epoch": 0.3361862375898665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5896,
      "step": 9820
    },
    {
      "epoch": 0.3362204724409449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.604,
      "step": 9821
    },
    {
      "epoch": 0.3362547072920233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5957,
      "step": 9822
    },
    {
      "epoch": 0.3362889421431017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6142,
      "step": 9823
    },
    {
      "epoch": 0.3363231769941801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5959,
      "step": 9824
    },
    {
      "epoch": 0.33635741184525847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6255,
      "step": 9825
    },
    {
      "epoch": 0.33639164669633687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6055,
      "step": 9826
    },
    {
      "epoch": 0.33642588154741526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5779,
      "step": 9827
    },
    {
      "epoch": 0.33646011639849366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6038,
      "step": 9828
    },
    {
      "epoch": 0.33649435124957205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5931,
      "step": 9829
    },
    {
      "epoch": 0.33652858610065045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6048,
      "step": 9830
    },
    {
      "epoch": 0.33656282095172885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6111,
      "step": 9831
    },
    {
      "epoch": 0.33659705580280724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6376,
      "step": 9832
    },
    {
      "epoch": 0.33663129065388564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5932,
      "step": 9833
    },
    {
      "epoch": 0.33666552550496404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6091,
      "step": 9834
    },
    {
      "epoch": 0.33669976035604243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5888,
      "step": 9835
    },
    {
      "epoch": 0.33673399520712083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6053,
      "step": 9836
    },
    {
      "epoch": 0.3367682300581992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.609,
      "step": 9837
    },
    {
      "epoch": 0.3368024649092776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5954,
      "step": 9838
    },
    {
      "epoch": 0.336836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.595,
      "step": 9839
    },
    {
      "epoch": 0.3368709346114344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5967,
      "step": 9840
    },
    {
      "epoch": 0.33690516946251287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.58,
      "step": 9841
    },
    {
      "epoch": 0.33693940431359126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6159,
      "step": 9842
    },
    {
      "epoch": 0.33697363916466966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6012,
      "step": 9843
    },
    {
      "epoch": 0.33700787401574805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5816,
      "step": 9844
    },
    {
      "epoch": 0.33704210886682645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5931,
      "step": 9845
    },
    {
      "epoch": 0.33707634371790485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6033,
      "step": 9846
    },
    {
      "epoch": 0.33711057856898324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5807,
      "step": 9847
    },
    {
      "epoch": 0.33714481342006164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5819,
      "step": 9848
    },
    {
      "epoch": 0.33717904827114004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5633,
      "step": 9849
    },
    {
      "epoch": 0.33721328312221843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5838,
      "step": 9850
    },
    {
      "epoch": 0.33724751797329683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6147,
      "step": 9851
    },
    {
      "epoch": 0.3372817528243752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6081,
      "step": 9852
    },
    {
      "epoch": 0.3373159876754536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5903,
      "step": 9853
    },
    {
      "epoch": 0.337350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5826,
      "step": 9854
    },
    {
      "epoch": 0.3373844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.589,
      "step": 9855
    },
    {
      "epoch": 0.3374186922286888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5997,
      "step": 9856
    },
    {
      "epoch": 0.3374529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5953,
      "step": 9857
    },
    {
      "epoch": 0.3374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.578,
      "step": 9858
    },
    {
      "epoch": 0.337521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5978,
      "step": 9859
    },
    {
      "epoch": 0.3375556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6065,
      "step": 9860
    },
    {
      "epoch": 0.3375898664840808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.59,
      "step": 9861
    },
    {
      "epoch": 0.3376241013351592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6003,
      "step": 9862
    },
    {
      "epoch": 0.3376583361862376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6006,
      "step": 9863
    },
    {
      "epoch": 0.337692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5823,
      "step": 9864
    },
    {
      "epoch": 0.3377268058883944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5985,
      "step": 9865
    },
    {
      "epoch": 0.3377610407394728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5944,
      "step": 9866
    },
    {
      "epoch": 0.33779527559055117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6001,
      "step": 9867
    },
    {
      "epoch": 0.33782951044162957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5818,
      "step": 9868
    },
    {
      "epoch": 0.33786374529270796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6177,
      "step": 9869
    },
    {
      "epoch": 0.33789798014378636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6019,
      "step": 9870
    },
    {
      "epoch": 0.33793221499486475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5988,
      "step": 9871
    },
    {
      "epoch": 0.33796644984594315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5947,
      "step": 9872
    },
    {
      "epoch": 0.33800068469702155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5909,
      "step": 9873
    },
    {
      "epoch": 0.33803491954809994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5545,
      "step": 9874
    },
    {
      "epoch": 0.33806915439917834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.592,
      "step": 9875
    },
    {
      "epoch": 0.33810338925025674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5871,
      "step": 9876
    },
    {
      "epoch": 0.33813762410133513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5761,
      "step": 9877
    },
    {
      "epoch": 0.3381718589524136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5755,
      "step": 9878
    },
    {
      "epoch": 0.338206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5759,
      "step": 9879
    },
    {
      "epoch": 0.3382403286545704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5752,
      "step": 9880
    },
    {
      "epoch": 0.3382745635056488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5789,
      "step": 9881
    },
    {
      "epoch": 0.33830879835672717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.6032,
      "step": 9882
    },
    {
      "epoch": 0.33834303320780557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5878,
      "step": 9883
    },
    {
      "epoch": 0.33837726805888396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5854,
      "step": 9884
    },
    {
      "epoch": 0.33841150290996236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5815,
      "step": 9885
    },
    {
      "epoch": 0.33844573776104075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5685,
      "step": 9886
    },
    {
      "epoch": 0.33847997261211915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5906,
      "step": 9887
    },
    {
      "epoch": 0.33851420746319755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5924,
      "step": 9888
    },
    {
      "epoch": 0.33854844231427594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5611,
      "step": 9889
    },
    {
      "epoch": 0.33858267716535434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5577,
      "step": 9890
    },
    {
      "epoch": 0.33861691201643274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5821,
      "step": 9891
    },
    {
      "epoch": 0.33865114686751113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5866,
      "step": 9892
    },
    {
      "epoch": 0.33868538171858953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5659,
      "step": 9893
    },
    {
      "epoch": 0.3387196165696679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5843,
      "step": 9894
    },
    {
      "epoch": 0.3387538514207463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5842,
      "step": 9895
    },
    {
      "epoch": 0.3387880862718247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5725,
      "step": 9896
    },
    {
      "epoch": 0.3388223211229031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5828,
      "step": 9897
    },
    {
      "epoch": 0.3388565559739815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5729,
      "step": 9898
    },
    {
      "epoch": 0.3388907908250599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5878,
      "step": 9899
    },
    {
      "epoch": 0.3389250256761383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5888,
      "step": 9900
    },
    {
      "epoch": 0.3389592605272167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5706,
      "step": 9901
    },
    {
      "epoch": 0.3389934953782951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.596,
      "step": 9902
    },
    {
      "epoch": 0.3390277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5928,
      "step": 9903
    },
    {
      "epoch": 0.3390619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5703,
      "step": 9904
    },
    {
      "epoch": 0.3390961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5669,
      "step": 9905
    },
    {
      "epoch": 0.3391304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5541,
      "step": 9906
    },
    {
      "epoch": 0.3391646696336871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5715,
      "step": 9907
    },
    {
      "epoch": 0.3391989044847655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5614,
      "step": 9908
    },
    {
      "epoch": 0.33923313933584387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5679,
      "step": 9909
    },
    {
      "epoch": 0.33926737418692227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5637,
      "step": 9910
    },
    {
      "epoch": 0.33930160903800066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5964,
      "step": 9911
    },
    {
      "epoch": 0.33933584388907906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.557,
      "step": 9912
    },
    {
      "epoch": 0.33937007874015745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5612,
      "step": 9913
    },
    {
      "epoch": 0.33940431359123585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5732,
      "step": 9914
    },
    {
      "epoch": 0.3394385484423143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5793,
      "step": 9915
    },
    {
      "epoch": 0.3394727832933927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5467,
      "step": 9916
    },
    {
      "epoch": 0.3395070181444711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5788,
      "step": 9917
    },
    {
      "epoch": 0.3395412529955495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5674,
      "step": 9918
    },
    {
      "epoch": 0.3395754878466279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5868,
      "step": 9919
    },
    {
      "epoch": 0.3396097226977063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5818,
      "step": 9920
    },
    {
      "epoch": 0.3396439575487847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5713,
      "step": 9921
    },
    {
      "epoch": 0.3396781923998631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5829,
      "step": 9922
    },
    {
      "epoch": 0.3397124272509415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5483,
      "step": 9923
    },
    {
      "epoch": 0.33974666210201987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5789,
      "step": 9924
    },
    {
      "epoch": 0.33978089695309827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5764,
      "step": 9925
    },
    {
      "epoch": 0.33981513180417666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5951,
      "step": 9926
    },
    {
      "epoch": 0.33984936665525506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5586,
      "step": 9927
    },
    {
      "epoch": 0.33988360150633345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5636,
      "step": 9928
    },
    {
      "epoch": 0.33991783635741185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5681,
      "step": 9929
    },
    {
      "epoch": 0.33995207120849025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5804,
      "step": 9930
    },
    {
      "epoch": 0.33998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5651,
      "step": 9931
    },
    {
      "epoch": 0.34002054091064704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5722,
      "step": 9932
    },
    {
      "epoch": 0.34005477576172544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5664,
      "step": 9933
    },
    {
      "epoch": 0.34008901061280383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5781,
      "step": 9934
    },
    {
      "epoch": 0.34012324546388223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5564,
      "step": 9935
    },
    {
      "epoch": 0.3401574803149606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5573,
      "step": 9936
    },
    {
      "epoch": 0.340191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5623,
      "step": 9937
    },
    {
      "epoch": 0.3402259500171174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5661,
      "step": 9938
    },
    {
      "epoch": 0.3402601848681958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5696,
      "step": 9939
    },
    {
      "epoch": 0.3402944197192742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5576,
      "step": 9940
    },
    {
      "epoch": 0.3403286545703526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5563,
      "step": 9941
    },
    {
      "epoch": 0.340362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5407,
      "step": 9942
    },
    {
      "epoch": 0.3403971242725094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5724,
      "step": 9943
    },
    {
      "epoch": 0.3404313591235878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.555,
      "step": 9944
    },
    {
      "epoch": 0.3404655939746662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5665,
      "step": 9945
    },
    {
      "epoch": 0.3404998288257446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5581,
      "step": 9946
    },
    {
      "epoch": 0.340534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5665,
      "step": 9947
    },
    {
      "epoch": 0.3405682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5578,
      "step": 9948
    },
    {
      "epoch": 0.3406025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5728,
      "step": 9949
    },
    {
      "epoch": 0.3406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5954,
      "step": 9950
    },
    {
      "epoch": 0.34067100308113657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5707,
      "step": 9951
    },
    {
      "epoch": 0.340705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5625,
      "step": 9952
    },
    {
      "epoch": 0.3407394727832934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5707,
      "step": 9953
    },
    {
      "epoch": 0.3407737076343718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5443,
      "step": 9954
    },
    {
      "epoch": 0.3408079424854502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5608,
      "step": 9955
    },
    {
      "epoch": 0.3408421773365286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5539,
      "step": 9956
    },
    {
      "epoch": 0.340876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5674,
      "step": 9957
    },
    {
      "epoch": 0.3409106470386854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5628,
      "step": 9958
    },
    {
      "epoch": 0.3409448818897638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5649,
      "step": 9959
    },
    {
      "epoch": 0.3409791167408422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5616,
      "step": 9960
    },
    {
      "epoch": 0.3410133515919206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5562,
      "step": 9961
    },
    {
      "epoch": 0.341047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5818,
      "step": 9962
    },
    {
      "epoch": 0.3410818212940774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5508,
      "step": 9963
    },
    {
      "epoch": 0.3411160561451558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5882,
      "step": 9964
    },
    {
      "epoch": 0.3411502909962342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5915,
      "step": 9965
    },
    {
      "epoch": 0.34118452584731257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5546,
      "step": 9966
    },
    {
      "epoch": 0.34121876069839097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5869,
      "step": 9967
    },
    {
      "epoch": 0.34125299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5784,
      "step": 9968
    },
    {
      "epoch": 0.34128723040054776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5626,
      "step": 9969
    },
    {
      "epoch": 0.34132146525162615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.558,
      "step": 9970
    },
    {
      "epoch": 0.34135570010270455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5711,
      "step": 9971
    },
    {
      "epoch": 0.34138993495378295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5599,
      "step": 9972
    },
    {
      "epoch": 0.34142416980486134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5669,
      "step": 9973
    },
    {
      "epoch": 0.34145840465593974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5625,
      "step": 9974
    },
    {
      "epoch": 0.34149263950701814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5563,
      "step": 9975
    },
    {
      "epoch": 0.34152687435809653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5611,
      "step": 9976
    },
    {
      "epoch": 0.34156110920917493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.552,
      "step": 9977
    },
    {
      "epoch": 0.3415953440602533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5777,
      "step": 9978
    },
    {
      "epoch": 0.3416295789113317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5622,
      "step": 9979
    },
    {
      "epoch": 0.3416638137624101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5838,
      "step": 9980
    },
    {
      "epoch": 0.3416980486134885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5697,
      "step": 9981
    },
    {
      "epoch": 0.3417322834645669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5453,
      "step": 9982
    },
    {
      "epoch": 0.3417665183156453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5254,
      "step": 9983
    },
    {
      "epoch": 0.3418007531667237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5688,
      "step": 9984
    },
    {
      "epoch": 0.3418349880178021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5492,
      "step": 9985
    },
    {
      "epoch": 0.3418692228688805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5669,
      "step": 9986
    },
    {
      "epoch": 0.3419034577199589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.552,
      "step": 9987
    },
    {
      "epoch": 0.3419376925710373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5624,
      "step": 9988
    },
    {
      "epoch": 0.34197192742211574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5709,
      "step": 9989
    },
    {
      "epoch": 0.34200616227319414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5387,
      "step": 9990
    },
    {
      "epoch": 0.34204039712427253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5457,
      "step": 9991
    },
    {
      "epoch": 0.34207463197535093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5537,
      "step": 9992
    },
    {
      "epoch": 0.3421088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5649,
      "step": 9993
    },
    {
      "epoch": 0.3421431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5366,
      "step": 9994
    },
    {
      "epoch": 0.3421773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5744,
      "step": 9995
    },
    {
      "epoch": 0.3422115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5664,
      "step": 9996
    },
    {
      "epoch": 0.3422458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5514,
      "step": 9997
    },
    {
      "epoch": 0.3422800410818213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5612,
      "step": 9998
    },
    {
      "epoch": 0.3423142759328997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5695,
      "step": 9999
    },
    {
      "epoch": 0.3423485107839781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5553,
      "step": 10000
    },
    {
      "epoch": 0.3423827456350565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5549,
      "step": 10001
    },
    {
      "epoch": 0.3424169804861349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5506,
      "step": 10002
    },
    {
      "epoch": 0.3424512153372133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5712,
      "step": 10003
    },
    {
      "epoch": 0.3424854501882917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5675,
      "step": 10004
    },
    {
      "epoch": 0.3425196850393701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5639,
      "step": 10005
    },
    {
      "epoch": 0.3425539198904485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.545,
      "step": 10006
    },
    {
      "epoch": 0.3425881547415269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5501,
      "step": 10007
    },
    {
      "epoch": 0.34262238959260527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5517,
      "step": 10008
    },
    {
      "epoch": 0.34265662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5747,
      "step": 10009
    },
    {
      "epoch": 0.34269085929476206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5525,
      "step": 10010
    },
    {
      "epoch": 0.34272509414584046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5548,
      "step": 10011
    },
    {
      "epoch": 0.34275932899691886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5685,
      "step": 10012
    },
    {
      "epoch": 0.34279356384799725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5447,
      "step": 10013
    },
    {
      "epoch": 0.34282779869907565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5603,
      "step": 10014
    },
    {
      "epoch": 0.34286203355015404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5569,
      "step": 10015
    },
    {
      "epoch": 0.34289626840123244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5666,
      "step": 10016
    },
    {
      "epoch": 0.34293050325231084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5497,
      "step": 10017
    },
    {
      "epoch": 0.34296473810338923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5476,
      "step": 10018
    },
    {
      "epoch": 0.34299897295446763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5477,
      "step": 10019
    },
    {
      "epoch": 0.343033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5149,
      "step": 10020
    },
    {
      "epoch": 0.3430674426566244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5324,
      "step": 10021
    },
    {
      "epoch": 0.3431016775077028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5595,
      "step": 10022
    },
    {
      "epoch": 0.3431359123587812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5538,
      "step": 10023
    },
    {
      "epoch": 0.3431701472098596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5396,
      "step": 10024
    },
    {
      "epoch": 0.34320438206093806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.552,
      "step": 10025
    },
    {
      "epoch": 0.34323861691201646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5586,
      "step": 10026
    },
    {
      "epoch": 0.34327285176309486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5347,
      "step": 10027
    },
    {
      "epoch": 0.34330708661417325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5324,
      "step": 10028
    },
    {
      "epoch": 0.34334132146525165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5622,
      "step": 10029
    },
    {
      "epoch": 0.34337555631633004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5367,
      "step": 10030
    },
    {
      "epoch": 0.34340979116740844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5421,
      "step": 10031
    },
    {
      "epoch": 0.34344402601848684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5327,
      "step": 10032
    },
    {
      "epoch": 0.34347826086956523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.545,
      "step": 10033
    },
    {
      "epoch": 0.34351249572064363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5392,
      "step": 10034
    },
    {
      "epoch": 0.343546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5678,
      "step": 10035
    },
    {
      "epoch": 0.3435809654228004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5459,
      "step": 10036
    },
    {
      "epoch": 0.3436152002738788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5607,
      "step": 10037
    },
    {
      "epoch": 0.3436494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5451,
      "step": 10038
    },
    {
      "epoch": 0.3436836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5699,
      "step": 10039
    },
    {
      "epoch": 0.343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5334,
      "step": 10040
    },
    {
      "epoch": 0.3437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5298,
      "step": 10041
    },
    {
      "epoch": 0.3437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5401,
      "step": 10042
    },
    {
      "epoch": 0.3438206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5401,
      "step": 10043
    },
    {
      "epoch": 0.3438548442314276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5406,
      "step": 10044
    },
    {
      "epoch": 0.343889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5662,
      "step": 10045
    },
    {
      "epoch": 0.3439233139335844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5407,
      "step": 10046
    },
    {
      "epoch": 0.3439575487846628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5448,
      "step": 10047
    },
    {
      "epoch": 0.3439917836357412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5393,
      "step": 10048
    },
    {
      "epoch": 0.3440260184868196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5578,
      "step": 10049
    },
    {
      "epoch": 0.34406025333789797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5276,
      "step": 10050
    },
    {
      "epoch": 0.34409448818897637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5519,
      "step": 10051
    },
    {
      "epoch": 0.34412872304005476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5432,
      "step": 10052
    },
    {
      "epoch": 0.34416295789113316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.557,
      "step": 10053
    },
    {
      "epoch": 0.34419719274221156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5512,
      "step": 10054
    },
    {
      "epoch": 0.34423142759328995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5307,
      "step": 10055
    },
    {
      "epoch": 0.34426566244436835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5415,
      "step": 10056
    },
    {
      "epoch": 0.34429989729544674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.537,
      "step": 10057
    },
    {
      "epoch": 0.34433413214652514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5444,
      "step": 10058
    },
    {
      "epoch": 0.34436836699760354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5432,
      "step": 10059
    },
    {
      "epoch": 0.34440260184868193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5478,
      "step": 10060
    },
    {
      "epoch": 0.34443683669976033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5445,
      "step": 10061
    },
    {
      "epoch": 0.3444710715508388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5667,
      "step": 10062
    },
    {
      "epoch": 0.3445053064019172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5642,
      "step": 10063
    },
    {
      "epoch": 0.3445395412529956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5571,
      "step": 10064
    },
    {
      "epoch": 0.34457377610407397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5338,
      "step": 10065
    },
    {
      "epoch": 0.34460801095515237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5468,
      "step": 10066
    },
    {
      "epoch": 0.34464224580623076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5425,
      "step": 10067
    },
    {
      "epoch": 0.34467648065730916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5381,
      "step": 10068
    },
    {
      "epoch": 0.34471071550838756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5396,
      "step": 10069
    },
    {
      "epoch": 0.34474495035946595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5268,
      "step": 10070
    },
    {
      "epoch": 0.34477918521054435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5513,
      "step": 10071
    },
    {
      "epoch": 0.34481342006162274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5607,
      "step": 10072
    },
    {
      "epoch": 0.34484765491270114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5343,
      "step": 10073
    },
    {
      "epoch": 0.34488188976377954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5526,
      "step": 10074
    },
    {
      "epoch": 0.34491612461485793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5347,
      "step": 10075
    },
    {
      "epoch": 0.34495035946593633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5558,
      "step": 10076
    },
    {
      "epoch": 0.3449845943170147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5276,
      "step": 10077
    },
    {
      "epoch": 0.3450188291680931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5245,
      "step": 10078
    },
    {
      "epoch": 0.3450530640191715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5488,
      "step": 10079
    },
    {
      "epoch": 0.3450872988702499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5426,
      "step": 10080
    },
    {
      "epoch": 0.3451215337213283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5402,
      "step": 10081
    },
    {
      "epoch": 0.3451557685724067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5211,
      "step": 10082
    },
    {
      "epoch": 0.3451900034234851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5583,
      "step": 10083
    },
    {
      "epoch": 0.3452242382745635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5483,
      "step": 10084
    },
    {
      "epoch": 0.3452584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5327,
      "step": 10085
    },
    {
      "epoch": 0.3452927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5373,
      "step": 10086
    },
    {
      "epoch": 0.3453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5569,
      "step": 10087
    },
    {
      "epoch": 0.3453611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5387,
      "step": 10088
    },
    {
      "epoch": 0.3453954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5747,
      "step": 10089
    },
    {
      "epoch": 0.3454296473810339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5418,
      "step": 10090
    },
    {
      "epoch": 0.3454638822321123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5345,
      "step": 10091
    },
    {
      "epoch": 0.34549811708319067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5677,
      "step": 10092
    },
    {
      "epoch": 0.34553235193426907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5367,
      "step": 10093
    },
    {
      "epoch": 0.34556658678534746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5514,
      "step": 10094
    },
    {
      "epoch": 0.34560082163642586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5449,
      "step": 10095
    },
    {
      "epoch": 0.34563505648750426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5141,
      "step": 10096
    },
    {
      "epoch": 0.34566929133858265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5368,
      "step": 10097
    },
    {
      "epoch": 0.34570352618966105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5453,
      "step": 10098
    },
    {
      "epoch": 0.3457377610407395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.546,
      "step": 10099
    },
    {
      "epoch": 0.3457719958918179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5423,
      "step": 10100
    },
    {
      "epoch": 0.3458062307428963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5357,
      "step": 10101
    },
    {
      "epoch": 0.3458404655939747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5346,
      "step": 10102
    },
    {
      "epoch": 0.3458747004450531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.556,
      "step": 10103
    },
    {
      "epoch": 0.3459089352961315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5529,
      "step": 10104
    },
    {
      "epoch": 0.3459431701472099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5501,
      "step": 10105
    },
    {
      "epoch": 0.3459774049982883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5548,
      "step": 10106
    },
    {
      "epoch": 0.34601163984936667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5291,
      "step": 10107
    },
    {
      "epoch": 0.34604587470044507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5392,
      "step": 10108
    },
    {
      "epoch": 0.34608010955152346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5091,
      "step": 10109
    },
    {
      "epoch": 0.34611434440260186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5407,
      "step": 10110
    },
    {
      "epoch": 0.34614857925368026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5418,
      "step": 10111
    },
    {
      "epoch": 0.34618281410475865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5272,
      "step": 10112
    },
    {
      "epoch": 0.34621704895583705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5338,
      "step": 10113
    },
    {
      "epoch": 0.34625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5205,
      "step": 10114
    },
    {
      "epoch": 0.34628551865799384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5327,
      "step": 10115
    },
    {
      "epoch": 0.34631975350907224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.528,
      "step": 10116
    },
    {
      "epoch": 0.34635398836015063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5565,
      "step": 10117
    },
    {
      "epoch": 0.34638822321122903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5266,
      "step": 10118
    },
    {
      "epoch": 0.3464224580623074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5348,
      "step": 10119
    },
    {
      "epoch": 0.3464566929133858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5446,
      "step": 10120
    },
    {
      "epoch": 0.3464909277644642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5466,
      "step": 10121
    },
    {
      "epoch": 0.3465251626155426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.539,
      "step": 10122
    },
    {
      "epoch": 0.346559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5289,
      "step": 10123
    },
    {
      "epoch": 0.3465936323176994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5337,
      "step": 10124
    },
    {
      "epoch": 0.3466278671687778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.539,
      "step": 10125
    },
    {
      "epoch": 0.3466621020198562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5401,
      "step": 10126
    },
    {
      "epoch": 0.3466963368709346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5495,
      "step": 10127
    },
    {
      "epoch": 0.346730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5329,
      "step": 10128
    },
    {
      "epoch": 0.3467648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5177,
      "step": 10129
    },
    {
      "epoch": 0.3467990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5539,
      "step": 10130
    },
    {
      "epoch": 0.3468332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.536,
      "step": 10131
    },
    {
      "epoch": 0.3468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5283,
      "step": 10132
    },
    {
      "epoch": 0.346901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.531,
      "step": 10133
    },
    {
      "epoch": 0.34693598082848337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5641,
      "step": 10134
    },
    {
      "epoch": 0.34697021567956177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5349,
      "step": 10135
    },
    {
      "epoch": 0.3470044505306402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5421,
      "step": 10136
    },
    {
      "epoch": 0.3470386853817186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5423,
      "step": 10137
    },
    {
      "epoch": 0.347072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5412,
      "step": 10138
    },
    {
      "epoch": 0.3471071550838754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5323,
      "step": 10139
    },
    {
      "epoch": 0.3471413899349538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5571,
      "step": 10140
    },
    {
      "epoch": 0.3471756247860322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5548,
      "step": 10141
    },
    {
      "epoch": 0.3472098596371106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5348,
      "step": 10142
    },
    {
      "epoch": 0.347244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5342,
      "step": 10143
    },
    {
      "epoch": 0.3472783293392674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5448,
      "step": 10144
    },
    {
      "epoch": 0.3473125641903458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5502,
      "step": 10145
    },
    {
      "epoch": 0.3473467990414242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5448,
      "step": 10146
    },
    {
      "epoch": 0.3473810338925026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5291,
      "step": 10147
    },
    {
      "epoch": 0.347415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5415,
      "step": 10148
    },
    {
      "epoch": 0.34744950359465937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5395,
      "step": 10149
    },
    {
      "epoch": 0.34748373844573777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5313,
      "step": 10150
    },
    {
      "epoch": 0.34751797329681616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.534,
      "step": 10151
    },
    {
      "epoch": 0.34755220814789456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5295,
      "step": 10152
    },
    {
      "epoch": 0.34758644299897296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5323,
      "step": 10153
    },
    {
      "epoch": 0.34762067785005135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5257,
      "step": 10154
    },
    {
      "epoch": 0.34765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5121,
      "step": 10155
    },
    {
      "epoch": 0.34768914755220814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5335,
      "step": 10156
    },
    {
      "epoch": 0.34772338240328654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5358,
      "step": 10157
    },
    {
      "epoch": 0.34775761725436494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5211,
      "step": 10158
    },
    {
      "epoch": 0.34779185210544333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5525,
      "step": 10159
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5305,
      "step": 10160
    },
    {
      "epoch": 0.3478603218076001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5222,
      "step": 10161
    },
    {
      "epoch": 0.3478945566586785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5319,
      "step": 10162
    },
    {
      "epoch": 0.3479287915097569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5156,
      "step": 10163
    },
    {
      "epoch": 0.3479630263608353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5433,
      "step": 10164
    },
    {
      "epoch": 0.3479972612119137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5179,
      "step": 10165
    },
    {
      "epoch": 0.3480314960629921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5234,
      "step": 10166
    },
    {
      "epoch": 0.3480657309140705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5513,
      "step": 10167
    },
    {
      "epoch": 0.3480999657651489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5186,
      "step": 10168
    },
    {
      "epoch": 0.3481342006162273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5274,
      "step": 10169
    },
    {
      "epoch": 0.3481684354673057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5234,
      "step": 10170
    },
    {
      "epoch": 0.3482026703183841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5233,
      "step": 10171
    },
    {
      "epoch": 0.34823690516946254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5425,
      "step": 10172
    },
    {
      "epoch": 0.34827114002054094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5195,
      "step": 10173
    },
    {
      "epoch": 0.34830537487161933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5308,
      "step": 10174
    },
    {
      "epoch": 0.34833960972269773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5319,
      "step": 10175
    },
    {
      "epoch": 0.3483738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5363,
      "step": 10176
    },
    {
      "epoch": 0.3484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.536,
      "step": 10177
    },
    {
      "epoch": 0.3484423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5108,
      "step": 10178
    },
    {
      "epoch": 0.3484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5246,
      "step": 10179
    },
    {
      "epoch": 0.3485107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5064,
      "step": 10180
    },
    {
      "epoch": 0.3485450188291681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.516,
      "step": 10181
    },
    {
      "epoch": 0.3485792536802465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5225,
      "step": 10182
    },
    {
      "epoch": 0.3486134885313249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5309,
      "step": 10183
    },
    {
      "epoch": 0.3486477233824033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5418,
      "step": 10184
    },
    {
      "epoch": 0.3486819582334817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5269,
      "step": 10185
    },
    {
      "epoch": 0.3487161930845601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5395,
      "step": 10186
    },
    {
      "epoch": 0.3487504279356385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5503,
      "step": 10187
    },
    {
      "epoch": 0.3487846627867169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5475,
      "step": 10188
    },
    {
      "epoch": 0.3488188976377953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5451,
      "step": 10189
    },
    {
      "epoch": 0.3488531324888737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5315,
      "step": 10190
    },
    {
      "epoch": 0.34888736733995207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.537,
      "step": 10191
    },
    {
      "epoch": 0.34892160219103047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5385,
      "step": 10192
    },
    {
      "epoch": 0.34895583704210886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.537,
      "step": 10193
    },
    {
      "epoch": 0.34899007189318726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5183,
      "step": 10194
    },
    {
      "epoch": 0.34902430674426566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5497,
      "step": 10195
    },
    {
      "epoch": 0.34905854159534405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5181,
      "step": 10196
    },
    {
      "epoch": 0.34909277644642245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5136,
      "step": 10197
    },
    {
      "epoch": 0.34912701129750084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.517,
      "step": 10198
    },
    {
      "epoch": 0.34916124614857924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5372,
      "step": 10199
    },
    {
      "epoch": 0.34919548099965764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5072,
      "step": 10200
    },
    {
      "epoch": 0.34922971585073603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5232,
      "step": 10201
    },
    {
      "epoch": 0.34926395070181443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5401,
      "step": 10202
    },
    {
      "epoch": 0.3492981855528928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5412,
      "step": 10203
    },
    {
      "epoch": 0.3493324204039712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.531,
      "step": 10204
    },
    {
      "epoch": 0.3493666552550496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5364,
      "step": 10205
    },
    {
      "epoch": 0.349400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5348,
      "step": 10206
    },
    {
      "epoch": 0.3494351249572064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5261,
      "step": 10207
    },
    {
      "epoch": 0.3494693598082848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5369,
      "step": 10208
    },
    {
      "epoch": 0.34950359465936326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5242,
      "step": 10209
    },
    {
      "epoch": 0.34953782951044166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5282,
      "step": 10210
    },
    {
      "epoch": 0.34957206436152005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5263,
      "step": 10211
    },
    {
      "epoch": 0.34960629921259845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5227,
      "step": 10212
    },
    {
      "epoch": 0.34964053406367684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5203,
      "step": 10213
    },
    {
      "epoch": 0.34967476891475524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5342,
      "step": 10214
    },
    {
      "epoch": 0.34970900376583364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5196,
      "step": 10215
    },
    {
      "epoch": 0.34974323861691203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5296,
      "step": 10216
    },
    {
      "epoch": 0.34977747346799043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5305,
      "step": 10217
    },
    {
      "epoch": 0.3498117083190688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5323,
      "step": 10218
    },
    {
      "epoch": 0.3498459431701472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5187,
      "step": 10219
    },
    {
      "epoch": 0.3498801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5377,
      "step": 10220
    },
    {
      "epoch": 0.349914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.55,
      "step": 10221
    },
    {
      "epoch": 0.3499486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5248,
      "step": 10222
    },
    {
      "epoch": 0.3499828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5515,
      "step": 10223
    },
    {
      "epoch": 0.3500171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.526,
      "step": 10224
    },
    {
      "epoch": 0.3500513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5295,
      "step": 10225
    },
    {
      "epoch": 0.350085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5445,
      "step": 10226
    },
    {
      "epoch": 0.3501198219787744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5286,
      "step": 10227
    },
    {
      "epoch": 0.3501540568298528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5443,
      "step": 10228
    },
    {
      "epoch": 0.3501882916809312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5336,
      "step": 10229
    },
    {
      "epoch": 0.3502225265320096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5094,
      "step": 10230
    },
    {
      "epoch": 0.350256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5147,
      "step": 10231
    },
    {
      "epoch": 0.3502909962341664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.511,
      "step": 10232
    },
    {
      "epoch": 0.35032523108524477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5268,
      "step": 10233
    },
    {
      "epoch": 0.35035946593632317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5115,
      "step": 10234
    },
    {
      "epoch": 0.35039370078740156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5133,
      "step": 10235
    },
    {
      "epoch": 0.35042793563847996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5268,
      "step": 10236
    },
    {
      "epoch": 0.35046217048955836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5376,
      "step": 10237
    },
    {
      "epoch": 0.35049640534063675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5061,
      "step": 10238
    },
    {
      "epoch": 0.35053064019171515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5091,
      "step": 10239
    },
    {
      "epoch": 0.35056487504279354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5079,
      "step": 10240
    },
    {
      "epoch": 0.35059910989387194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.544,
      "step": 10241
    },
    {
      "epoch": 0.35063334474495034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5215,
      "step": 10242
    },
    {
      "epoch": 0.35066757959602873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5075,
      "step": 10243
    },
    {
      "epoch": 0.35070181444710713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.536,
      "step": 10244
    },
    {
      "epoch": 0.3507360492981855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5409,
      "step": 10245
    },
    {
      "epoch": 0.350770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5368,
      "step": 10246
    },
    {
      "epoch": 0.3508045190003424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5252,
      "step": 10247
    },
    {
      "epoch": 0.35083875385142077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5091,
      "step": 10248
    },
    {
      "epoch": 0.35087298870249917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5311,
      "step": 10249
    },
    {
      "epoch": 0.35090722355357756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5272,
      "step": 10250
    },
    {
      "epoch": 0.35094145840465596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5232,
      "step": 10251
    },
    {
      "epoch": 0.35097569325573436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5194,
      "step": 10252
    },
    {
      "epoch": 0.35100992810681275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5279,
      "step": 10253
    },
    {
      "epoch": 0.35104416295789115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5127,
      "step": 10254
    },
    {
      "epoch": 0.35107839780896954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5332,
      "step": 10255
    },
    {
      "epoch": 0.35111263266004794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5154,
      "step": 10256
    },
    {
      "epoch": 0.35114686751112634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5408,
      "step": 10257
    },
    {
      "epoch": 0.35118110236220473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5094,
      "step": 10258
    },
    {
      "epoch": 0.35121533721328313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5307,
      "step": 10259
    },
    {
      "epoch": 0.3512495720643615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5367,
      "step": 10260
    },
    {
      "epoch": 0.3512838069154399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5192,
      "step": 10261
    },
    {
      "epoch": 0.3513180417665183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5033,
      "step": 10262
    },
    {
      "epoch": 0.3513522766175967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.505,
      "step": 10263
    },
    {
      "epoch": 0.3513865114686751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5113,
      "step": 10264
    },
    {
      "epoch": 0.3514207463197535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5143,
      "step": 10265
    },
    {
      "epoch": 0.3514549811708319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5084,
      "step": 10266
    },
    {
      "epoch": 0.3514892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5122,
      "step": 10267
    },
    {
      "epoch": 0.3515234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.513,
      "step": 10268
    },
    {
      "epoch": 0.3515576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5134,
      "step": 10269
    },
    {
      "epoch": 0.3515919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5359,
      "step": 10270
    },
    {
      "epoch": 0.3516261554262239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5159,
      "step": 10271
    },
    {
      "epoch": 0.3516603902773023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5322,
      "step": 10272
    },
    {
      "epoch": 0.3516946251283807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5417,
      "step": 10273
    },
    {
      "epoch": 0.3517288599794591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5353,
      "step": 10274
    },
    {
      "epoch": 0.35176309483053747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5069,
      "step": 10275
    },
    {
      "epoch": 0.35179732968161587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5268,
      "step": 10276
    },
    {
      "epoch": 0.35183156453269426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5223,
      "step": 10277
    },
    {
      "epoch": 0.35186579938377266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5119,
      "step": 10278
    },
    {
      "epoch": 0.35190003423485106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5287,
      "step": 10279
    },
    {
      "epoch": 0.35193426908592945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5336,
      "step": 10280
    },
    {
      "epoch": 0.35196850393700785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5194,
      "step": 10281
    },
    {
      "epoch": 0.35200273878808624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5208,
      "step": 10282
    },
    {
      "epoch": 0.3520369736391647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5319,
      "step": 10283
    },
    {
      "epoch": 0.3520712084902431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5128,
      "step": 10284
    },
    {
      "epoch": 0.3521054433413215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5471,
      "step": 10285
    },
    {
      "epoch": 0.3521396781923999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.542,
      "step": 10286
    },
    {
      "epoch": 0.3521739130434783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5047,
      "step": 10287
    },
    {
      "epoch": 0.3522081478945567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4995,
      "step": 10288
    },
    {
      "epoch": 0.3522423827456351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5043,
      "step": 10289
    },
    {
      "epoch": 0.35227661759671347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5202,
      "step": 10290
    },
    {
      "epoch": 0.35231085244779187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5167,
      "step": 10291
    },
    {
      "epoch": 0.35234508729887026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5305,
      "step": 10292
    },
    {
      "epoch": 0.35237932214994866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5307,
      "step": 10293
    },
    {
      "epoch": 0.35241355700102706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5253,
      "step": 10294
    },
    {
      "epoch": 0.35244779185210545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5387,
      "step": 10295
    },
    {
      "epoch": 0.35248202670318385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.526,
      "step": 10296
    },
    {
      "epoch": 0.35251626155426224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5142,
      "step": 10297
    },
    {
      "epoch": 0.35255049640534064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5371,
      "step": 10298
    },
    {
      "epoch": 0.35258473125641904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5264,
      "step": 10299
    },
    {
      "epoch": 0.35261896610749743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5374,
      "step": 10300
    },
    {
      "epoch": 0.35265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5089,
      "step": 10301
    },
    {
      "epoch": 0.3526874358096542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5174,
      "step": 10302
    },
    {
      "epoch": 0.3527216706607326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5175,
      "step": 10303
    },
    {
      "epoch": 0.352755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5304,
      "step": 10304
    },
    {
      "epoch": 0.3527901403628894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.517,
      "step": 10305
    },
    {
      "epoch": 0.3528243752139678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5045,
      "step": 10306
    },
    {
      "epoch": 0.3528586100650462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5265,
      "step": 10307
    },
    {
      "epoch": 0.3528928449161246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5081,
      "step": 10308
    },
    {
      "epoch": 0.352927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5126,
      "step": 10309
    },
    {
      "epoch": 0.3529613146182814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5041,
      "step": 10310
    },
    {
      "epoch": 0.3529955494693598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5076,
      "step": 10311
    },
    {
      "epoch": 0.3530297843204382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.524,
      "step": 10312
    },
    {
      "epoch": 0.3530640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5232,
      "step": 10313
    },
    {
      "epoch": 0.353098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5115,
      "step": 10314
    },
    {
      "epoch": 0.3531324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5081,
      "step": 10315
    },
    {
      "epoch": 0.3531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5238,
      "step": 10316
    },
    {
      "epoch": 0.35320095857583017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5089,
      "step": 10317
    },
    {
      "epoch": 0.35323519342690857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.506,
      "step": 10318
    },
    {
      "epoch": 0.35326942827798696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5328,
      "step": 10319
    },
    {
      "epoch": 0.3533036631290654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5282,
      "step": 10320
    },
    {
      "epoch": 0.3533378979801438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5297,
      "step": 10321
    },
    {
      "epoch": 0.3533721328312222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5205,
      "step": 10322
    },
    {
      "epoch": 0.3534063676823006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5263,
      "step": 10323
    },
    {
      "epoch": 0.353440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5093,
      "step": 10324
    },
    {
      "epoch": 0.3534748373844574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.505,
      "step": 10325
    },
    {
      "epoch": 0.3535090722355358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.508,
      "step": 10326
    },
    {
      "epoch": 0.3535433070866142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5162,
      "step": 10327
    },
    {
      "epoch": 0.3535775419376926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5208,
      "step": 10328
    },
    {
      "epoch": 0.353611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5263,
      "step": 10329
    },
    {
      "epoch": 0.3536460116398494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5064,
      "step": 10330
    },
    {
      "epoch": 0.3536802464909278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5177,
      "step": 10331
    },
    {
      "epoch": 0.35371448134200617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5328,
      "step": 10332
    },
    {
      "epoch": 0.35374871619308457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5321,
      "step": 10333
    },
    {
      "epoch": 0.35378295104416296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5009,
      "step": 10334
    },
    {
      "epoch": 0.35381718589524136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.505,
      "step": 10335
    },
    {
      "epoch": 0.35385142074631976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5257,
      "step": 10336
    },
    {
      "epoch": 0.35388565559739815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5037,
      "step": 10337
    },
    {
      "epoch": 0.35391989044847655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5267,
      "step": 10338
    },
    {
      "epoch": 0.35395412529955494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4939,
      "step": 10339
    },
    {
      "epoch": 0.35398836015063334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5277,
      "step": 10340
    },
    {
      "epoch": 0.35402259500171174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5398,
      "step": 10341
    },
    {
      "epoch": 0.35405682985279013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5133,
      "step": 10342
    },
    {
      "epoch": 0.35409106470386853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5213,
      "step": 10343
    },
    {
      "epoch": 0.3541252995549469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5031,
      "step": 10344
    },
    {
      "epoch": 0.3541595344060253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5189,
      "step": 10345
    },
    {
      "epoch": 0.3541937692571037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5239,
      "step": 10346
    },
    {
      "epoch": 0.3542280041081821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4999,
      "step": 10347
    },
    {
      "epoch": 0.3542622389592605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5225,
      "step": 10348
    },
    {
      "epoch": 0.3542964738103389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5161,
      "step": 10349
    },
    {
      "epoch": 0.3543307086614173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5424,
      "step": 10350
    },
    {
      "epoch": 0.3543649435124957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4966,
      "step": 10351
    },
    {
      "epoch": 0.3543991783635741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5365,
      "step": 10352
    },
    {
      "epoch": 0.3544334132146525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5194,
      "step": 10353
    },
    {
      "epoch": 0.3544676480657309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5235,
      "step": 10354
    },
    {
      "epoch": 0.3545018829168093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4953,
      "step": 10355
    },
    {
      "epoch": 0.35453611776788774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.497,
      "step": 10356
    },
    {
      "epoch": 0.35457035261896613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5058,
      "step": 10357
    },
    {
      "epoch": 0.35460458747004453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5124,
      "step": 10358
    },
    {
      "epoch": 0.3546388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4944,
      "step": 10359
    },
    {
      "epoch": 0.3546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5023,
      "step": 10360
    },
    {
      "epoch": 0.3547072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5354,
      "step": 10361
    },
    {
      "epoch": 0.3547415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5071,
      "step": 10362
    },
    {
      "epoch": 0.3547757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.531,
      "step": 10363
    },
    {
      "epoch": 0.3548099965765149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5269,
      "step": 10364
    },
    {
      "epoch": 0.3548442314275933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5241,
      "step": 10365
    },
    {
      "epoch": 0.3548784662786717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4879,
      "step": 10366
    },
    {
      "epoch": 0.3549127011297501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5211,
      "step": 10367
    },
    {
      "epoch": 0.3549469359808285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5091,
      "step": 10368
    },
    {
      "epoch": 0.3549811708319069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5144,
      "step": 10369
    },
    {
      "epoch": 0.3550154056829853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5099,
      "step": 10370
    },
    {
      "epoch": 0.3550496405340637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5132,
      "step": 10371
    },
    {
      "epoch": 0.3550838753851421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5244,
      "step": 10372
    },
    {
      "epoch": 0.3551181102362205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5171,
      "step": 10373
    },
    {
      "epoch": 0.35515234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5137,
      "step": 10374
    },
    {
      "epoch": 0.35518657993837727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5109,
      "step": 10375
    },
    {
      "epoch": 0.35522081478945566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5265,
      "step": 10376
    },
    {
      "epoch": 0.35525504964053406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5094,
      "step": 10377
    },
    {
      "epoch": 0.35528928449161246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5366,
      "step": 10378
    },
    {
      "epoch": 0.35532351934269085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5105,
      "step": 10379
    },
    {
      "epoch": 0.35535775419376925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5103,
      "step": 10380
    },
    {
      "epoch": 0.35539198904484764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5036,
      "step": 10381
    },
    {
      "epoch": 0.35542622389592604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5187,
      "step": 10382
    },
    {
      "epoch": 0.35546045874700444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.505,
      "step": 10383
    },
    {
      "epoch": 0.35549469359808283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4883,
      "step": 10384
    },
    {
      "epoch": 0.35552892844916123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5136,
      "step": 10385
    },
    {
      "epoch": 0.3555631633002396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5144,
      "step": 10386
    },
    {
      "epoch": 0.355597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5127,
      "step": 10387
    },
    {
      "epoch": 0.3556316330023964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4997,
      "step": 10388
    },
    {
      "epoch": 0.3556658678534748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5407,
      "step": 10389
    },
    {
      "epoch": 0.3557001027045532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.532,
      "step": 10390
    },
    {
      "epoch": 0.3557343375556316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5228,
      "step": 10391
    },
    {
      "epoch": 0.35576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5224,
      "step": 10392
    },
    {
      "epoch": 0.35580280725778846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5155,
      "step": 10393
    },
    {
      "epoch": 0.35583704210886685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5018,
      "step": 10394
    },
    {
      "epoch": 0.35587127695994525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5111,
      "step": 10395
    },
    {
      "epoch": 0.35590551181102364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.509,
      "step": 10396
    },
    {
      "epoch": 0.35593974666210204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.527,
      "step": 10397
    },
    {
      "epoch": 0.35597398151318044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5222,
      "step": 10398
    },
    {
      "epoch": 0.35600821636425883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5117,
      "step": 10399
    },
    {
      "epoch": 0.35604245121533723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5015,
      "step": 10400
    },
    {
      "epoch": 0.3560766860664156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5235,
      "step": 10401
    },
    {
      "epoch": 0.356110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5039,
      "step": 10402
    },
    {
      "epoch": 0.3561451557685724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5023,
      "step": 10403
    },
    {
      "epoch": 0.3561793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5174,
      "step": 10404
    },
    {
      "epoch": 0.3562136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5138,
      "step": 10405
    },
    {
      "epoch": 0.3562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4939,
      "step": 10406
    },
    {
      "epoch": 0.356282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5103,
      "step": 10407
    },
    {
      "epoch": 0.3563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4906,
      "step": 10408
    },
    {
      "epoch": 0.3563505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5081,
      "step": 10409
    },
    {
      "epoch": 0.3563847997261212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4996,
      "step": 10410
    },
    {
      "epoch": 0.3564190345771996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5058,
      "step": 10411
    },
    {
      "epoch": 0.356453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5156,
      "step": 10412
    },
    {
      "epoch": 0.3564875042793564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5305,
      "step": 10413
    },
    {
      "epoch": 0.3565217391304348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5195,
      "step": 10414
    },
    {
      "epoch": 0.3565559739815132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4973,
      "step": 10415
    },
    {
      "epoch": 0.35659020883259157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5245,
      "step": 10416
    },
    {
      "epoch": 0.35662444368366997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5045,
      "step": 10417
    },
    {
      "epoch": 0.35665867853474836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5102,
      "step": 10418
    },
    {
      "epoch": 0.35669291338582676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5015,
      "step": 10419
    },
    {
      "epoch": 0.35672714823690516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5083,
      "step": 10420
    },
    {
      "epoch": 0.35676138308798355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5149,
      "step": 10421
    },
    {
      "epoch": 0.35679561793906195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5063,
      "step": 10422
    },
    {
      "epoch": 0.35682985279014034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5105,
      "step": 10423
    },
    {
      "epoch": 0.35686408764121874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5356,
      "step": 10424
    },
    {
      "epoch": 0.35689832249229714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5061,
      "step": 10425
    },
    {
      "epoch": 0.35693255734337553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5094,
      "step": 10426
    },
    {
      "epoch": 0.35696679219445393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4998,
      "step": 10427
    },
    {
      "epoch": 0.3570010270455323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5051,
      "step": 10428
    },
    {
      "epoch": 0.3570352618966107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4975,
      "step": 10429
    },
    {
      "epoch": 0.3570694967476892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4976,
      "step": 10430
    },
    {
      "epoch": 0.35710373159876757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5302,
      "step": 10431
    },
    {
      "epoch": 0.35713796644984597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5246,
      "step": 10432
    },
    {
      "epoch": 0.35717220130092436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.513,
      "step": 10433
    },
    {
      "epoch": 0.35720643615200276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4979,
      "step": 10434
    },
    {
      "epoch": 0.35724067100308116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5079,
      "step": 10435
    },
    {
      "epoch": 0.35727490585415955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5096,
      "step": 10436
    },
    {
      "epoch": 0.35730914070523795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4998,
      "step": 10437
    },
    {
      "epoch": 0.35734337555631634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4987,
      "step": 10438
    },
    {
      "epoch": 0.35737761040739474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4979,
      "step": 10439
    },
    {
      "epoch": 0.35741184525847314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4923,
      "step": 10440
    },
    {
      "epoch": 0.35744608010955153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.508,
      "step": 10441
    },
    {
      "epoch": 0.35748031496062993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5093,
      "step": 10442
    },
    {
      "epoch": 0.3575145498117083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5108,
      "step": 10443
    },
    {
      "epoch": 0.3575487846627867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5236,
      "step": 10444
    },
    {
      "epoch": 0.3575830195138651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5199,
      "step": 10445
    },
    {
      "epoch": 0.3576172543649435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5092,
      "step": 10446
    },
    {
      "epoch": 0.3576514892160219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5047,
      "step": 10447
    },
    {
      "epoch": 0.3576857240671003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5038,
      "step": 10448
    },
    {
      "epoch": 0.3577199589181787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5181,
      "step": 10449
    },
    {
      "epoch": 0.3577541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5104,
      "step": 10450
    },
    {
      "epoch": 0.3577884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.491,
      "step": 10451
    },
    {
      "epoch": 0.3578226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5117,
      "step": 10452
    },
    {
      "epoch": 0.3578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4912,
      "step": 10453
    },
    {
      "epoch": 0.3578911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5091,
      "step": 10454
    },
    {
      "epoch": 0.3579253680246491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5072,
      "step": 10455
    },
    {
      "epoch": 0.3579596028757275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5086,
      "step": 10456
    },
    {
      "epoch": 0.3579938377268059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5147,
      "step": 10457
    },
    {
      "epoch": 0.35802807257788427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5022,
      "step": 10458
    },
    {
      "epoch": 0.35806230742896267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5084,
      "step": 10459
    },
    {
      "epoch": 0.35809654228004106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5159,
      "step": 10460
    },
    {
      "epoch": 0.35813077713111946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4956,
      "step": 10461
    },
    {
      "epoch": 0.35816501198219786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5159,
      "step": 10462
    },
    {
      "epoch": 0.35819924683327625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5025,
      "step": 10463
    },
    {
      "epoch": 0.35823348168435465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4978,
      "step": 10464
    },
    {
      "epoch": 0.35826771653543305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5145,
      "step": 10465
    },
    {
      "epoch": 0.35830195138651144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5054,
      "step": 10466
    },
    {
      "epoch": 0.3583361862375899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4953,
      "step": 10467
    },
    {
      "epoch": 0.3583704210886683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5075,
      "step": 10468
    },
    {
      "epoch": 0.3584046559397467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5099,
      "step": 10469
    },
    {
      "epoch": 0.3584388907908251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.502,
      "step": 10470
    },
    {
      "epoch": 0.3584731256419035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5095,
      "step": 10471
    },
    {
      "epoch": 0.3585073604929819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4984,
      "step": 10472
    },
    {
      "epoch": 0.35854159534406027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5129,
      "step": 10473
    },
    {
      "epoch": 0.35857583019513867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5086,
      "step": 10474
    },
    {
      "epoch": 0.35861006504621706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5164,
      "step": 10475
    },
    {
      "epoch": 0.35864429989729546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5096,
      "step": 10476
    },
    {
      "epoch": 0.35867853474837386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5031,
      "step": 10477
    },
    {
      "epoch": 0.35871276959945225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4946,
      "step": 10478
    },
    {
      "epoch": 0.35874700445053065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.499,
      "step": 10479
    },
    {
      "epoch": 0.35878123930160905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5126,
      "step": 10480
    },
    {
      "epoch": 0.35881547415268744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5009,
      "step": 10481
    },
    {
      "epoch": 0.35884970900376584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5302,
      "step": 10482
    },
    {
      "epoch": 0.35888394385484423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.525,
      "step": 10483
    },
    {
      "epoch": 0.35891817870592263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5213,
      "step": 10484
    },
    {
      "epoch": 0.358952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5052,
      "step": 10485
    },
    {
      "epoch": 0.3589866484080794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.546,
      "step": 10486
    },
    {
      "epoch": 0.3590208832591578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4944,
      "step": 10487
    },
    {
      "epoch": 0.3590551181102362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5014,
      "step": 10488
    },
    {
      "epoch": 0.3590893529613146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5189,
      "step": 10489
    },
    {
      "epoch": 0.359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5024,
      "step": 10490
    },
    {
      "epoch": 0.3591578226634714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5031,
      "step": 10491
    },
    {
      "epoch": 0.3591920575145498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5242,
      "step": 10492
    },
    {
      "epoch": 0.3592262923656282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4974,
      "step": 10493
    },
    {
      "epoch": 0.3592605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4845,
      "step": 10494
    },
    {
      "epoch": 0.359294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5034,
      "step": 10495
    },
    {
      "epoch": 0.3593289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4843,
      "step": 10496
    },
    {
      "epoch": 0.3593632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.489,
      "step": 10497
    },
    {
      "epoch": 0.3593974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.502,
      "step": 10498
    },
    {
      "epoch": 0.3594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5222,
      "step": 10499
    },
    {
      "epoch": 0.35946593632317697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5036,
      "step": 10500
    },
    {
      "epoch": 0.35950017117425537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4964,
      "step": 10501
    },
    {
      "epoch": 0.35953440602533376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5131,
      "step": 10502
    },
    {
      "epoch": 0.35956864087641216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4924,
      "step": 10503
    },
    {
      "epoch": 0.3596028757274906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4965,
      "step": 10504
    },
    {
      "epoch": 0.359637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5217,
      "step": 10505
    },
    {
      "epoch": 0.3596713454296474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4995,
      "step": 10506
    },
    {
      "epoch": 0.3597055802807258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5017,
      "step": 10507
    },
    {
      "epoch": 0.3597398151318042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4846,
      "step": 10508
    },
    {
      "epoch": 0.3597740499828826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5171,
      "step": 10509
    },
    {
      "epoch": 0.359808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5127,
      "step": 10510
    },
    {
      "epoch": 0.3598425196850394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4817,
      "step": 10511
    },
    {
      "epoch": 0.3598767545361178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4912,
      "step": 10512
    },
    {
      "epoch": 0.3599109893871962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4936,
      "step": 10513
    },
    {
      "epoch": 0.3599452242382746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5043,
      "step": 10514
    },
    {
      "epoch": 0.35997945908935297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4982,
      "step": 10515
    },
    {
      "epoch": 0.36001369394043137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.504,
      "step": 10516
    },
    {
      "epoch": 0.36004792879150976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5095,
      "step": 10517
    },
    {
      "epoch": 0.36008216364258816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5053,
      "step": 10518
    },
    {
      "epoch": 0.36011639849366656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4961,
      "step": 10519
    },
    {
      "epoch": 0.36015063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.492,
      "step": 10520
    },
    {
      "epoch": 0.36018486819582335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.517,
      "step": 10521
    },
    {
      "epoch": 0.36021910304690175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5087,
      "step": 10522
    },
    {
      "epoch": 0.36025333789798014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4929,
      "step": 10523
    },
    {
      "epoch": 0.36028757274905854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4986,
      "step": 10524
    },
    {
      "epoch": 0.36032180760013693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5154,
      "step": 10525
    },
    {
      "epoch": 0.36035604245121533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5008,
      "step": 10526
    },
    {
      "epoch": 0.3603902773022937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4981,
      "step": 10527
    },
    {
      "epoch": 0.3604245121533721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5197,
      "step": 10528
    },
    {
      "epoch": 0.3604587470044505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5022,
      "step": 10529
    },
    {
      "epoch": 0.3604929818555289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5119,
      "step": 10530
    },
    {
      "epoch": 0.3605272167066073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5155,
      "step": 10531
    },
    {
      "epoch": 0.3605614515576857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5031,
      "step": 10532
    },
    {
      "epoch": 0.3605956864087641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5036,
      "step": 10533
    },
    {
      "epoch": 0.3606299212598425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5209,
      "step": 10534
    },
    {
      "epoch": 0.3606641561109209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5028,
      "step": 10535
    },
    {
      "epoch": 0.3606983909619993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4915,
      "step": 10536
    },
    {
      "epoch": 0.3607326258130777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5062,
      "step": 10537
    },
    {
      "epoch": 0.3607668606641561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5016,
      "step": 10538
    },
    {
      "epoch": 0.3608010955152345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4976,
      "step": 10539
    },
    {
      "epoch": 0.36083533036631293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5114,
      "step": 10540
    },
    {
      "epoch": 0.36086956521739133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5147,
      "step": 10541
    },
    {
      "epoch": 0.3609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4979,
      "step": 10542
    },
    {
      "epoch": 0.3609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4986,
      "step": 10543
    },
    {
      "epoch": 0.3609722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5105,
      "step": 10544
    },
    {
      "epoch": 0.3610065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.521,
      "step": 10545
    },
    {
      "epoch": 0.3610407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5,
      "step": 10546
    },
    {
      "epoch": 0.3610749743238617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5001,
      "step": 10547
    },
    {
      "epoch": 0.3611092091749401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5001,
      "step": 10548
    },
    {
      "epoch": 0.3611434440260185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5142,
      "step": 10549
    },
    {
      "epoch": 0.3611776788770969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.492,
      "step": 10550
    },
    {
      "epoch": 0.3612119137281753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5217,
      "step": 10551
    },
    {
      "epoch": 0.3612461485792537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5036,
      "step": 10552
    },
    {
      "epoch": 0.3612803834303321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5219,
      "step": 10553
    },
    {
      "epoch": 0.3613146182814105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4887,
      "step": 10554
    },
    {
      "epoch": 0.3613488531324889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5126,
      "step": 10555
    },
    {
      "epoch": 0.3613830879835673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4915,
      "step": 10556
    },
    {
      "epoch": 0.36141732283464567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5018,
      "step": 10557
    },
    {
      "epoch": 0.36145155768572407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5074,
      "step": 10558
    },
    {
      "epoch": 0.36148579253680246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5054,
      "step": 10559
    },
    {
      "epoch": 0.36152002738788086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5006,
      "step": 10560
    },
    {
      "epoch": 0.36155426223895926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4926,
      "step": 10561
    },
    {
      "epoch": 0.36158849709003765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5058,
      "step": 10562
    },
    {
      "epoch": 0.36162273194111605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5135,
      "step": 10563
    },
    {
      "epoch": 0.36165696679219445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4953,
      "step": 10564
    },
    {
      "epoch": 0.36169120164327284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5103,
      "step": 10565
    },
    {
      "epoch": 0.36172543649435124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4882,
      "step": 10566
    },
    {
      "epoch": 0.36175967134542963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5241,
      "step": 10567
    },
    {
      "epoch": 0.36179390619650803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5145,
      "step": 10568
    },
    {
      "epoch": 0.3618281410475864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4986,
      "step": 10569
    },
    {
      "epoch": 0.3618623758986648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4912,
      "step": 10570
    },
    {
      "epoch": 0.3618966107497432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5166,
      "step": 10571
    },
    {
      "epoch": 0.3619308456008216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5156,
      "step": 10572
    },
    {
      "epoch": 0.3619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4946,
      "step": 10573
    },
    {
      "epoch": 0.3619993153029784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5137,
      "step": 10574
    },
    {
      "epoch": 0.3620335501540568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5028,
      "step": 10575
    },
    {
      "epoch": 0.3620677850051352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5007,
      "step": 10576
    },
    {
      "epoch": 0.36210201985621365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5037,
      "step": 10577
    },
    {
      "epoch": 0.36213625470729205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.499,
      "step": 10578
    },
    {
      "epoch": 0.36217048955837045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5072,
      "step": 10579
    },
    {
      "epoch": 0.36220472440944884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4691,
      "step": 10580
    },
    {
      "epoch": 0.36223895926052724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5058,
      "step": 10581
    },
    {
      "epoch": 0.36227319411160563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4973,
      "step": 10582
    },
    {
      "epoch": 0.36230742896268403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5034,
      "step": 10583
    },
    {
      "epoch": 0.3623416638137624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.507,
      "step": 10584
    },
    {
      "epoch": 0.3623758986648408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.51,
      "step": 10585
    },
    {
      "epoch": 0.3624101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4973,
      "step": 10586
    },
    {
      "epoch": 0.3624443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5111,
      "step": 10587
    },
    {
      "epoch": 0.362478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4985,
      "step": 10588
    },
    {
      "epoch": 0.3625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5014,
      "step": 10589
    },
    {
      "epoch": 0.3625470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5056,
      "step": 10590
    },
    {
      "epoch": 0.3625813077713112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5222,
      "step": 10591
    },
    {
      "epoch": 0.3626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4834,
      "step": 10592
    },
    {
      "epoch": 0.362649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4904,
      "step": 10593
    },
    {
      "epoch": 0.3626840123245464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5145,
      "step": 10594
    },
    {
      "epoch": 0.3627182471756248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5117,
      "step": 10595
    },
    {
      "epoch": 0.3627524820267032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5076,
      "step": 10596
    },
    {
      "epoch": 0.3627867168777816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5119,
      "step": 10597
    },
    {
      "epoch": 0.36282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5013,
      "step": 10598
    },
    {
      "epoch": 0.36285518657993837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.512,
      "step": 10599
    },
    {
      "epoch": 0.36288942143101677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4987,
      "step": 10600
    },
    {
      "epoch": 0.36292365628209516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5161,
      "step": 10601
    },
    {
      "epoch": 0.36295789113317356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4952,
      "step": 10602
    },
    {
      "epoch": 0.36299212598425196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5104,
      "step": 10603
    },
    {
      "epoch": 0.36302636083533035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5241,
      "step": 10604
    },
    {
      "epoch": 0.36306059568640875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5139,
      "step": 10605
    },
    {
      "epoch": 0.36309483053748715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5047,
      "step": 10606
    },
    {
      "epoch": 0.36312906538856554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4916,
      "step": 10607
    },
    {
      "epoch": 0.36316330023964394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5028,
      "step": 10608
    },
    {
      "epoch": 0.36319753509072233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5163,
      "step": 10609
    },
    {
      "epoch": 0.36323176994180073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5133,
      "step": 10610
    },
    {
      "epoch": 0.3632660047928791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.498,
      "step": 10611
    },
    {
      "epoch": 0.3633002396439575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5148,
      "step": 10612
    },
    {
      "epoch": 0.3633344744950359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4976,
      "step": 10613
    },
    {
      "epoch": 0.36336870934611437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5017,
      "step": 10614
    },
    {
      "epoch": 0.36340294419719277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5141,
      "step": 10615
    },
    {
      "epoch": 0.36343717904827116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5023,
      "step": 10616
    },
    {
      "epoch": 0.36347141389934956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5023,
      "step": 10617
    },
    {
      "epoch": 0.36350564875042796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4807,
      "step": 10618
    },
    {
      "epoch": 0.36353988360150635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4913,
      "step": 10619
    },
    {
      "epoch": 0.36357411845258475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4852,
      "step": 10620
    },
    {
      "epoch": 0.36360835330366315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5101,
      "step": 10621
    },
    {
      "epoch": 0.36364258815474154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4942,
      "step": 10622
    },
    {
      "epoch": 0.36367682300581994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.493,
      "step": 10623
    },
    {
      "epoch": 0.36371105785689833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5007,
      "step": 10624
    },
    {
      "epoch": 0.36374529270797673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4931,
      "step": 10625
    },
    {
      "epoch": 0.3637795275590551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4801,
      "step": 10626
    },
    {
      "epoch": 0.3638137624101335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.493,
      "step": 10627
    },
    {
      "epoch": 0.3638479972612119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5108,
      "step": 10628
    },
    {
      "epoch": 0.3638822321122903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.501,
      "step": 10629
    },
    {
      "epoch": 0.3639164669633687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4982,
      "step": 10630
    },
    {
      "epoch": 0.3639507018144471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5045,
      "step": 10631
    },
    {
      "epoch": 0.3639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4962,
      "step": 10632
    },
    {
      "epoch": 0.3640191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5028,
      "step": 10633
    },
    {
      "epoch": 0.3640534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.493,
      "step": 10634
    },
    {
      "epoch": 0.3640876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5045,
      "step": 10635
    },
    {
      "epoch": 0.3641218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4927,
      "step": 10636
    },
    {
      "epoch": 0.3641561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4942,
      "step": 10637
    },
    {
      "epoch": 0.3641903457719959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.473,
      "step": 10638
    },
    {
      "epoch": 0.3642245806230743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5034,
      "step": 10639
    },
    {
      "epoch": 0.3642588154741527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5067,
      "step": 10640
    },
    {
      "epoch": 0.36429305032523107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.511,
      "step": 10641
    },
    {
      "epoch": 0.36432728517630947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4878,
      "step": 10642
    },
    {
      "epoch": 0.36436152002738786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4951,
      "step": 10643
    },
    {
      "epoch": 0.36439575487846626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4915,
      "step": 10644
    },
    {
      "epoch": 0.36442998972954466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4902,
      "step": 10645
    },
    {
      "epoch": 0.36446422458062305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.491,
      "step": 10646
    },
    {
      "epoch": 0.36449845943170145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4967,
      "step": 10647
    },
    {
      "epoch": 0.36453269428277985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5072,
      "step": 10648
    },
    {
      "epoch": 0.36456692913385824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.508,
      "step": 10649
    },
    {
      "epoch": 0.36460116398493664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4926,
      "step": 10650
    },
    {
      "epoch": 0.3646353988360151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5039,
      "step": 10651
    },
    {
      "epoch": 0.3646696336870935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5016,
      "step": 10652
    },
    {
      "epoch": 0.3647038685381719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.52,
      "step": 10653
    },
    {
      "epoch": 0.3647381033892503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5026,
      "step": 10654
    },
    {
      "epoch": 0.3647723382403287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4998,
      "step": 10655
    },
    {
      "epoch": 0.36480657309140707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4945,
      "step": 10656
    },
    {
      "epoch": 0.36484080794248547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.504,
      "step": 10657
    },
    {
      "epoch": 0.36487504279356386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.513,
      "step": 10658
    },
    {
      "epoch": 0.36490927764464226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5246,
      "step": 10659
    },
    {
      "epoch": 0.36494351249572066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4966,
      "step": 10660
    },
    {
      "epoch": 0.36497774734679905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4866,
      "step": 10661
    },
    {
      "epoch": 0.36501198219787745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4997,
      "step": 10662
    },
    {
      "epoch": 0.36504621704895585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5185,
      "step": 10663
    },
    {
      "epoch": 0.36508045190003424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4846,
      "step": 10664
    },
    {
      "epoch": 0.36511468675111264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5259,
      "step": 10665
    },
    {
      "epoch": 0.36514892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4995,
      "step": 10666
    },
    {
      "epoch": 0.36518315645326943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4984,
      "step": 10667
    },
    {
      "epoch": 0.3652173913043478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5008,
      "step": 10668
    },
    {
      "epoch": 0.3652516261554262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5078,
      "step": 10669
    },
    {
      "epoch": 0.3652858610065046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4837,
      "step": 10670
    },
    {
      "epoch": 0.365320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4798,
      "step": 10671
    },
    {
      "epoch": 0.3653543307086614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4964,
      "step": 10672
    },
    {
      "epoch": 0.3653885655597398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5163,
      "step": 10673
    },
    {
      "epoch": 0.3654228004108182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4837,
      "step": 10674
    },
    {
      "epoch": 0.3654570352618966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.505,
      "step": 10675
    },
    {
      "epoch": 0.365491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4866,
      "step": 10676
    },
    {
      "epoch": 0.3655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4747,
      "step": 10677
    },
    {
      "epoch": 0.3655597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4714,
      "step": 10678
    },
    {
      "epoch": 0.3655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.512,
      "step": 10679
    },
    {
      "epoch": 0.3656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4772,
      "step": 10680
    },
    {
      "epoch": 0.365662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5099,
      "step": 10681
    },
    {
      "epoch": 0.3656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5105,
      "step": 10682
    },
    {
      "epoch": 0.36573091407052377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5134,
      "step": 10683
    },
    {
      "epoch": 0.36576514892160217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5026,
      "step": 10684
    },
    {
      "epoch": 0.36579938377268056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4953,
      "step": 10685
    },
    {
      "epoch": 0.36583361862375896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5036,
      "step": 10686
    },
    {
      "epoch": 0.36586785347483736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.501,
      "step": 10687
    },
    {
      "epoch": 0.3659020883259158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4985,
      "step": 10688
    },
    {
      "epoch": 0.3659363231769942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5281,
      "step": 10689
    },
    {
      "epoch": 0.3659705580280726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4938,
      "step": 10690
    },
    {
      "epoch": 0.366004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4925,
      "step": 10691
    },
    {
      "epoch": 0.3660390277302294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4975,
      "step": 10692
    },
    {
      "epoch": 0.3660732625813078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4958,
      "step": 10693
    },
    {
      "epoch": 0.3661074974323862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5035,
      "step": 10694
    },
    {
      "epoch": 0.3661417322834646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5115,
      "step": 10695
    },
    {
      "epoch": 0.366175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4902,
      "step": 10696
    },
    {
      "epoch": 0.3662102019856214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4835,
      "step": 10697
    },
    {
      "epoch": 0.36624443683669977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5246,
      "step": 10698
    },
    {
      "epoch": 0.36627867168777817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4845,
      "step": 10699
    },
    {
      "epoch": 0.36631290653885656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4731,
      "step": 10700
    },
    {
      "epoch": 0.36634714138993496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5067,
      "step": 10701
    },
    {
      "epoch": 0.36638137624101336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4898,
      "step": 10702
    },
    {
      "epoch": 0.36641561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4928,
      "step": 10703
    },
    {
      "epoch": 0.36644984594317015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4799,
      "step": 10704
    },
    {
      "epoch": 0.36648408079424855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5019,
      "step": 10705
    },
    {
      "epoch": 0.36651831564532694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4891,
      "step": 10706
    },
    {
      "epoch": 0.36655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4979,
      "step": 10707
    },
    {
      "epoch": 0.36658678534748373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4861,
      "step": 10708
    },
    {
      "epoch": 0.36662102019856213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5014,
      "step": 10709
    },
    {
      "epoch": 0.3666552550496405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5066,
      "step": 10710
    },
    {
      "epoch": 0.3666894899007189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.499,
      "step": 10711
    },
    {
      "epoch": 0.3667237247517973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5054,
      "step": 10712
    },
    {
      "epoch": 0.3667579596028757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5235,
      "step": 10713
    },
    {
      "epoch": 0.3667921944539541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5015,
      "step": 10714
    },
    {
      "epoch": 0.3668264293050325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4786,
      "step": 10715
    },
    {
      "epoch": 0.3668606641561109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4822,
      "step": 10716
    },
    {
      "epoch": 0.3668948990071893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.494,
      "step": 10717
    },
    {
      "epoch": 0.3669291338582677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4942,
      "step": 10718
    },
    {
      "epoch": 0.3669633687093461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4938,
      "step": 10719
    },
    {
      "epoch": 0.3669976035604245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4774,
      "step": 10720
    },
    {
      "epoch": 0.3670318384115029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4891,
      "step": 10721
    },
    {
      "epoch": 0.3670660732625813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4943,
      "step": 10722
    },
    {
      "epoch": 0.3671003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5166,
      "step": 10723
    },
    {
      "epoch": 0.36713454296473813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4922,
      "step": 10724
    },
    {
      "epoch": 0.3671687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5126,
      "step": 10725
    },
    {
      "epoch": 0.3672030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4951,
      "step": 10726
    },
    {
      "epoch": 0.3672372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4859,
      "step": 10727
    },
    {
      "epoch": 0.3672714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4912,
      "step": 10728
    },
    {
      "epoch": 0.3673057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4934,
      "step": 10729
    },
    {
      "epoch": 0.3673399520712085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4813,
      "step": 10730
    },
    {
      "epoch": 0.3673741869222869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4902,
      "step": 10731
    },
    {
      "epoch": 0.3674084217733653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4823,
      "step": 10732
    },
    {
      "epoch": 0.3674426566244437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4961,
      "step": 10733
    },
    {
      "epoch": 0.3674768914755221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4978,
      "step": 10734
    },
    {
      "epoch": 0.3675111263266005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4824,
      "step": 10735
    },
    {
      "epoch": 0.3675453611776789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5006,
      "step": 10736
    },
    {
      "epoch": 0.3675795960287573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4987,
      "step": 10737
    },
    {
      "epoch": 0.3676138308798357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4827,
      "step": 10738
    },
    {
      "epoch": 0.3676480657309141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4827,
      "step": 10739
    },
    {
      "epoch": 0.36768230058199247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4815,
      "step": 10740
    },
    {
      "epoch": 0.36771653543307087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4901,
      "step": 10741
    },
    {
      "epoch": 0.36775077028414926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5263,
      "step": 10742
    },
    {
      "epoch": 0.36778500513522766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4886,
      "step": 10743
    },
    {
      "epoch": 0.36781923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4965,
      "step": 10744
    },
    {
      "epoch": 0.36785347483738445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5004,
      "step": 10745
    },
    {
      "epoch": 0.36788770968846285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4935,
      "step": 10746
    },
    {
      "epoch": 0.36792194453954125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4979,
      "step": 10747
    },
    {
      "epoch": 0.36795617939061964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4992,
      "step": 10748
    },
    {
      "epoch": 0.36799041424169804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4718,
      "step": 10749
    },
    {
      "epoch": 0.36802464909277643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.49,
      "step": 10750
    },
    {
      "epoch": 0.36805888394385483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4729,
      "step": 10751
    },
    {
      "epoch": 0.3680931187949332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4788,
      "step": 10752
    },
    {
      "epoch": 0.3681273536460116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4789,
      "step": 10753
    },
    {
      "epoch": 0.36816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4935,
      "step": 10754
    },
    {
      "epoch": 0.3681958233481684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4981,
      "step": 10755
    },
    {
      "epoch": 0.3682300581992468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5185,
      "step": 10756
    },
    {
      "epoch": 0.3682642930503252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5183,
      "step": 10757
    },
    {
      "epoch": 0.3682985279014036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5038,
      "step": 10758
    },
    {
      "epoch": 0.368332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4773,
      "step": 10759
    },
    {
      "epoch": 0.3683669976035604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5124,
      "step": 10760
    },
    {
      "epoch": 0.36840123245463885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.495,
      "step": 10761
    },
    {
      "epoch": 0.36843546730571725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4986,
      "step": 10762
    },
    {
      "epoch": 0.36846970215679564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4871,
      "step": 10763
    },
    {
      "epoch": 0.36850393700787404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4724,
      "step": 10764
    },
    {
      "epoch": 0.36853817185895243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4716,
      "step": 10765
    },
    {
      "epoch": 0.36857240671003083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5017,
      "step": 10766
    },
    {
      "epoch": 0.3686066415611092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.507,
      "step": 10767
    },
    {
      "epoch": 0.3686408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4907,
      "step": 10768
    },
    {
      "epoch": 0.368675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5003,
      "step": 10769
    },
    {
      "epoch": 0.3687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5183,
      "step": 10770
    },
    {
      "epoch": 0.3687435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4965,
      "step": 10771
    },
    {
      "epoch": 0.3687778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4671,
      "step": 10772
    },
    {
      "epoch": 0.3688120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4966,
      "step": 10773
    },
    {
      "epoch": 0.368846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4896,
      "step": 10774
    },
    {
      "epoch": 0.3688805203697364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4855,
      "step": 10775
    },
    {
      "epoch": 0.3689147552208148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4938,
      "step": 10776
    },
    {
      "epoch": 0.3689489900718932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5074,
      "step": 10777
    },
    {
      "epoch": 0.3689832249229716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4915,
      "step": 10778
    },
    {
      "epoch": 0.36901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4726,
      "step": 10779
    },
    {
      "epoch": 0.3690516946251284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5065,
      "step": 10780
    },
    {
      "epoch": 0.3690859294762068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.491,
      "step": 10781
    },
    {
      "epoch": 0.36912016432728517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4818,
      "step": 10782
    },
    {
      "epoch": 0.36915439917836357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.504,
      "step": 10783
    },
    {
      "epoch": 0.36918863402944196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5003,
      "step": 10784
    },
    {
      "epoch": 0.36922286888052036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4895,
      "step": 10785
    },
    {
      "epoch": 0.36925710373159876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4909,
      "step": 10786
    },
    {
      "epoch": 0.36929133858267715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.502,
      "step": 10787
    },
    {
      "epoch": 0.36932557343375555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4911,
      "step": 10788
    },
    {
      "epoch": 0.36935980828483395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4736,
      "step": 10789
    },
    {
      "epoch": 0.36939404313591234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.501,
      "step": 10790
    },
    {
      "epoch": 0.36942827798699074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4788,
      "step": 10791
    },
    {
      "epoch": 0.36946251283806913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4946,
      "step": 10792
    },
    {
      "epoch": 0.36949674768914753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4975,
      "step": 10793
    },
    {
      "epoch": 0.3695309825402259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5078,
      "step": 10794
    },
    {
      "epoch": 0.3695652173913043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4911,
      "step": 10795
    },
    {
      "epoch": 0.3695994522423827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4945,
      "step": 10796
    },
    {
      "epoch": 0.3696336870934611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4895,
      "step": 10797
    },
    {
      "epoch": 0.36966792194453957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4957,
      "step": 10798
    },
    {
      "epoch": 0.36970215679561796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4733,
      "step": 10799
    },
    {
      "epoch": 0.36973639164669636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5255,
      "step": 10800
    },
    {
      "epoch": 0.36977062649777476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5062,
      "step": 10801
    },
    {
      "epoch": 0.36980486134885315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5078,
      "step": 10802
    },
    {
      "epoch": 0.36983909619993155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4862,
      "step": 10803
    },
    {
      "epoch": 0.36987333105100995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4664,
      "step": 10804
    },
    {
      "epoch": 0.36990756590208834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4747,
      "step": 10805
    },
    {
      "epoch": 0.36994180075316674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4833,
      "step": 10806
    },
    {
      "epoch": 0.36997603560424513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4946,
      "step": 10807
    },
    {
      "epoch": 0.37001027045532353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4834,
      "step": 10808
    },
    {
      "epoch": 0.3700445053064019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4589,
      "step": 10809
    },
    {
      "epoch": 0.3700787401574803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4893,
      "step": 10810
    },
    {
      "epoch": 0.3701129750085587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5045,
      "step": 10811
    },
    {
      "epoch": 0.3701472098596371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.482,
      "step": 10812
    },
    {
      "epoch": 0.3701814447107155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.492,
      "step": 10813
    },
    {
      "epoch": 0.3702156795617939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4995,
      "step": 10814
    },
    {
      "epoch": 0.3702499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4936,
      "step": 10815
    },
    {
      "epoch": 0.3702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4849,
      "step": 10816
    },
    {
      "epoch": 0.3703183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4896,
      "step": 10817
    },
    {
      "epoch": 0.3703526189661075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4934,
      "step": 10818
    },
    {
      "epoch": 0.3703868538171859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5049,
      "step": 10819
    },
    {
      "epoch": 0.3704210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4878,
      "step": 10820
    },
    {
      "epoch": 0.3704553235193427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4643,
      "step": 10821
    },
    {
      "epoch": 0.3704895583704211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4862,
      "step": 10822
    },
    {
      "epoch": 0.3705237932214995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5056,
      "step": 10823
    },
    {
      "epoch": 0.37055802807257787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5002,
      "step": 10824
    },
    {
      "epoch": 0.37059226292365627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4974,
      "step": 10825
    },
    {
      "epoch": 0.37062649777473466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4849,
      "step": 10826
    },
    {
      "epoch": 0.37066073262581306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4917,
      "step": 10827
    },
    {
      "epoch": 0.37069496747689146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4877,
      "step": 10828
    },
    {
      "epoch": 0.37072920232796985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4704,
      "step": 10829
    },
    {
      "epoch": 0.37076343717904825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5142,
      "step": 10830
    },
    {
      "epoch": 0.37079767203012665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4824,
      "step": 10831
    },
    {
      "epoch": 0.37083190688120504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4764,
      "step": 10832
    },
    {
      "epoch": 0.37086614173228344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4947,
      "step": 10833
    },
    {
      "epoch": 0.37090037658336183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4736,
      "step": 10834
    },
    {
      "epoch": 0.3709346114344403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4828,
      "step": 10835
    },
    {
      "epoch": 0.3709688462855187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4713,
      "step": 10836
    },
    {
      "epoch": 0.3710030811365971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4639,
      "step": 10837
    },
    {
      "epoch": 0.3710373159876755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5079,
      "step": 10838
    },
    {
      "epoch": 0.37107155083875387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5012,
      "step": 10839
    },
    {
      "epoch": 0.37110578568983227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4744,
      "step": 10840
    },
    {
      "epoch": 0.37114002054091066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4919,
      "step": 10841
    },
    {
      "epoch": 0.37117425539198906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4835,
      "step": 10842
    },
    {
      "epoch": 0.37120849024306746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.481,
      "step": 10843
    },
    {
      "epoch": 0.37124272509414585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4969,
      "step": 10844
    },
    {
      "epoch": 0.37127695994522425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4913,
      "step": 10845
    },
    {
      "epoch": 0.37131119479630265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4949,
      "step": 10846
    },
    {
      "epoch": 0.37134542964738104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.486,
      "step": 10847
    },
    {
      "epoch": 0.37137966449845944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5064,
      "step": 10848
    },
    {
      "epoch": 0.37141389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4855,
      "step": 10849
    },
    {
      "epoch": 0.37144813420061623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4868,
      "step": 10850
    },
    {
      "epoch": 0.3714823690516946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4865,
      "step": 10851
    },
    {
      "epoch": 0.371516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4804,
      "step": 10852
    },
    {
      "epoch": 0.3715508387538514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5006,
      "step": 10853
    },
    {
      "epoch": 0.3715850736049298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4885,
      "step": 10854
    },
    {
      "epoch": 0.3716193084560082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4882,
      "step": 10855
    },
    {
      "epoch": 0.3716535433070866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4703,
      "step": 10856
    },
    {
      "epoch": 0.371687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4556,
      "step": 10857
    },
    {
      "epoch": 0.3717220130092434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4923,
      "step": 10858
    },
    {
      "epoch": 0.3717562478603218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4863,
      "step": 10859
    },
    {
      "epoch": 0.3717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5094,
      "step": 10860
    },
    {
      "epoch": 0.3718247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4854,
      "step": 10861
    },
    {
      "epoch": 0.371858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4806,
      "step": 10862
    },
    {
      "epoch": 0.3718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4687,
      "step": 10863
    },
    {
      "epoch": 0.3719274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4696,
      "step": 10864
    },
    {
      "epoch": 0.3719616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4693,
      "step": 10865
    },
    {
      "epoch": 0.37199589181787057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4845,
      "step": 10866
    },
    {
      "epoch": 0.37203012666894897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 10867
    },
    {
      "epoch": 0.37206436152002736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4954,
      "step": 10868
    },
    {
      "epoch": 0.37209859637110576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4934,
      "step": 10869
    },
    {
      "epoch": 0.37213283122218416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5023,
      "step": 10870
    },
    {
      "epoch": 0.3721670660732626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4777,
      "step": 10871
    },
    {
      "epoch": 0.372201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4735,
      "step": 10872
    },
    {
      "epoch": 0.3722355357754194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4789,
      "step": 10873
    },
    {
      "epoch": 0.3722697706264978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4798,
      "step": 10874
    },
    {
      "epoch": 0.3723040054775762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4818,
      "step": 10875
    },
    {
      "epoch": 0.3723382403286546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4963,
      "step": 10876
    },
    {
      "epoch": 0.372372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4818,
      "step": 10877
    },
    {
      "epoch": 0.3724067100308114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4817,
      "step": 10878
    },
    {
      "epoch": 0.3724409448818898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4763,
      "step": 10879
    },
    {
      "epoch": 0.3724751797329682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4843,
      "step": 10880
    },
    {
      "epoch": 0.37250941458404657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4851,
      "step": 10881
    },
    {
      "epoch": 0.37254364943512497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.497,
      "step": 10882
    },
    {
      "epoch": 0.37257788428620336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4917,
      "step": 10883
    },
    {
      "epoch": 0.37261211913728176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4791,
      "step": 10884
    },
    {
      "epoch": 0.37264635398836016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5028,
      "step": 10885
    },
    {
      "epoch": 0.37268058883943855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5023,
      "step": 10886
    },
    {
      "epoch": 0.37271482369051695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4787,
      "step": 10887
    },
    {
      "epoch": 0.37274905854159535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4783,
      "step": 10888
    },
    {
      "epoch": 0.37278329339267374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4741,
      "step": 10889
    },
    {
      "epoch": 0.37281752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5204,
      "step": 10890
    },
    {
      "epoch": 0.37285176309483054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4719,
      "step": 10891
    },
    {
      "epoch": 0.37288599794590893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4823,
      "step": 10892
    },
    {
      "epoch": 0.3729202327969873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4823,
      "step": 10893
    },
    {
      "epoch": 0.3729544676480657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4795,
      "step": 10894
    },
    {
      "epoch": 0.3729887024991441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4747,
      "step": 10895
    },
    {
      "epoch": 0.3730229373502225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4957,
      "step": 10896
    },
    {
      "epoch": 0.3730571722013009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 10897
    },
    {
      "epoch": 0.3730914070523793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5021,
      "step": 10898
    },
    {
      "epoch": 0.3731256419034577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4705,
      "step": 10899
    },
    {
      "epoch": 0.3731598767545361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5028,
      "step": 10900
    },
    {
      "epoch": 0.3731941116056145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5083,
      "step": 10901
    },
    {
      "epoch": 0.3732283464566929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5173,
      "step": 10902
    },
    {
      "epoch": 0.3732625813077713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4837,
      "step": 10903
    },
    {
      "epoch": 0.3732968161588497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4934,
      "step": 10904
    },
    {
      "epoch": 0.3733310510099281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4999,
      "step": 10905
    },
    {
      "epoch": 0.3733652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4772,
      "step": 10906
    },
    {
      "epoch": 0.3733995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4817,
      "step": 10907
    },
    {
      "epoch": 0.3734337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4873,
      "step": 10908
    },
    {
      "epoch": 0.3734679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4884,
      "step": 10909
    },
    {
      "epoch": 0.3735022252653201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4899,
      "step": 10910
    },
    {
      "epoch": 0.3735364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4782,
      "step": 10911
    },
    {
      "epoch": 0.3735706949674769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4785,
      "step": 10912
    },
    {
      "epoch": 0.3736049298185553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4893,
      "step": 10913
    },
    {
      "epoch": 0.3736391646696337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4661,
      "step": 10914
    },
    {
      "epoch": 0.3736733995207121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4742,
      "step": 10915
    },
    {
      "epoch": 0.3737076343717905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4784,
      "step": 10916
    },
    {
      "epoch": 0.3737418692228689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4818,
      "step": 10917
    },
    {
      "epoch": 0.3737761040739473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4729,
      "step": 10918
    },
    {
      "epoch": 0.3738103389250257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4818,
      "step": 10919
    },
    {
      "epoch": 0.3738445737761041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4792,
      "step": 10920
    },
    {
      "epoch": 0.3738788086271825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.475,
      "step": 10921
    },
    {
      "epoch": 0.3739130434782609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4684,
      "step": 10922
    },
    {
      "epoch": 0.37394727832933927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4979,
      "step": 10923
    },
    {
      "epoch": 0.37398151318041767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4962,
      "step": 10924
    },
    {
      "epoch": 0.37401574803149606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4821,
      "step": 10925
    },
    {
      "epoch": 0.37404998288257446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4818,
      "step": 10926
    },
    {
      "epoch": 0.37408421773365286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4792,
      "step": 10927
    },
    {
      "epoch": 0.37411845258473125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4894,
      "step": 10928
    },
    {
      "epoch": 0.37415268743580965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4881,
      "step": 10929
    },
    {
      "epoch": 0.37418692228688805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.501,
      "step": 10930
    },
    {
      "epoch": 0.37422115713796644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5135,
      "step": 10931
    },
    {
      "epoch": 0.37425539198904484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4797,
      "step": 10932
    },
    {
      "epoch": 0.37428962684012324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4818,
      "step": 10933
    },
    {
      "epoch": 0.37432386169120163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4804,
      "step": 10934
    },
    {
      "epoch": 0.37435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4717,
      "step": 10935
    },
    {
      "epoch": 0.3743923313933584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4807,
      "step": 10936
    },
    {
      "epoch": 0.3744265662444368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4801,
      "step": 10937
    },
    {
      "epoch": 0.3744608010955152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4823,
      "step": 10938
    },
    {
      "epoch": 0.3744950359465936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4701,
      "step": 10939
    },
    {
      "epoch": 0.374529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4632,
      "step": 10940
    },
    {
      "epoch": 0.3745635056487504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4749,
      "step": 10941
    },
    {
      "epoch": 0.3745977404998288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4882,
      "step": 10942
    },
    {
      "epoch": 0.3746319753509072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4736,
      "step": 10943
    },
    {
      "epoch": 0.3746662102019856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4678,
      "step": 10944
    },
    {
      "epoch": 0.37470044505306405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.476,
      "step": 10945
    },
    {
      "epoch": 0.37473467990414244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.475,
      "step": 10946
    },
    {
      "epoch": 0.37476891475522084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.476,
      "step": 10947
    },
    {
      "epoch": 0.37480314960629924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4709,
      "step": 10948
    },
    {
      "epoch": 0.37483738445737763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4752,
      "step": 10949
    },
    {
      "epoch": 0.374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4731,
      "step": 10950
    },
    {
      "epoch": 0.3749058541595344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4853,
      "step": 10951
    },
    {
      "epoch": 0.3749400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4901,
      "step": 10952
    },
    {
      "epoch": 0.3749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4838,
      "step": 10953
    },
    {
      "epoch": 0.3750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4778,
      "step": 10954
    },
    {
      "epoch": 0.375042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4826,
      "step": 10955
    },
    {
      "epoch": 0.3750770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4742,
      "step": 10956
    },
    {
      "epoch": 0.3751112632660048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4754,
      "step": 10957
    },
    {
      "epoch": 0.3751454981170832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4771,
      "step": 10958
    },
    {
      "epoch": 0.3751797329681616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4951,
      "step": 10959
    },
    {
      "epoch": 0.37521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4807,
      "step": 10960
    },
    {
      "epoch": 0.3752482026703184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4786,
      "step": 10961
    },
    {
      "epoch": 0.3752824375213968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4791,
      "step": 10962
    },
    {
      "epoch": 0.3753166723724752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4835,
      "step": 10963
    },
    {
      "epoch": 0.3753509072235536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5039,
      "step": 10964
    },
    {
      "epoch": 0.375385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4931,
      "step": 10965
    },
    {
      "epoch": 0.37541937692571037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4763,
      "step": 10966
    },
    {
      "epoch": 0.37545361177678876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4854,
      "step": 10967
    },
    {
      "epoch": 0.37548784662786716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4739,
      "step": 10968
    },
    {
      "epoch": 0.37552208147894556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4976,
      "step": 10969
    },
    {
      "epoch": 0.37555631633002395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4748,
      "step": 10970
    },
    {
      "epoch": 0.37559055118110235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4793,
      "step": 10971
    },
    {
      "epoch": 0.37562478603218075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.512,
      "step": 10972
    },
    {
      "epoch": 0.37565902088325914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4843,
      "step": 10973
    },
    {
      "epoch": 0.37569325573433754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4605,
      "step": 10974
    },
    {
      "epoch": 0.37572749058541594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4877,
      "step": 10975
    },
    {
      "epoch": 0.37576172543649433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4642,
      "step": 10976
    },
    {
      "epoch": 0.37579596028757273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4912,
      "step": 10977
    },
    {
      "epoch": 0.3758301951386511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4846,
      "step": 10978
    },
    {
      "epoch": 0.3758644299897295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4661,
      "step": 10979
    },
    {
      "epoch": 0.3758986648408079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.501,
      "step": 10980
    },
    {
      "epoch": 0.3759328996918863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4705,
      "step": 10981
    },
    {
      "epoch": 0.37596713454296476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4797,
      "step": 10982
    },
    {
      "epoch": 0.37600136939404316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4984,
      "step": 10983
    },
    {
      "epoch": 0.37603560424512156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4898,
      "step": 10984
    },
    {
      "epoch": 0.37606983909619995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4885,
      "step": 10985
    },
    {
      "epoch": 0.37610407394727835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4875,
      "step": 10986
    },
    {
      "epoch": 0.37613830879835675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4806,
      "step": 10987
    },
    {
      "epoch": 0.37617254364943514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4659,
      "step": 10988
    },
    {
      "epoch": 0.37620677850051354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4756,
      "step": 10989
    },
    {
      "epoch": 0.37624101335159194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4855,
      "step": 10990
    },
    {
      "epoch": 0.37627524820267033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4909,
      "step": 10991
    },
    {
      "epoch": 0.37630948305374873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4909,
      "step": 10992
    },
    {
      "epoch": 0.3763437179048271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4766,
      "step": 10993
    },
    {
      "epoch": 0.3763779527559055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4785,
      "step": 10994
    },
    {
      "epoch": 0.3764121876069839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4856,
      "step": 10995
    },
    {
      "epoch": 0.3764464224580623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.463,
      "step": 10996
    },
    {
      "epoch": 0.3764806573091407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4869,
      "step": 10997
    },
    {
      "epoch": 0.3765148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.494,
      "step": 10998
    },
    {
      "epoch": 0.3765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4694,
      "step": 10999
    },
    {
      "epoch": 0.3765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4691,
      "step": 11000
    },
    {
      "epoch": 0.3766175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4915,
      "step": 11001
    },
    {
      "epoch": 0.3766518315645327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4818,
      "step": 11002
    },
    {
      "epoch": 0.3766860664156111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4773,
      "step": 11003
    },
    {
      "epoch": 0.3767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4568,
      "step": 11004
    },
    {
      "epoch": 0.3767545361177679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4561,
      "step": 11005
    },
    {
      "epoch": 0.3767887709688463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4786,
      "step": 11006
    },
    {
      "epoch": 0.3768230058199247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4711,
      "step": 11007
    },
    {
      "epoch": 0.37685724067100307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4891,
      "step": 11008
    },
    {
      "epoch": 0.37689147552208146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4962,
      "step": 11009
    },
    {
      "epoch": 0.37692571037315986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4953,
      "step": 11010
    },
    {
      "epoch": 0.37695994522423826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4998,
      "step": 11011
    },
    {
      "epoch": 0.37699418007531665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4828,
      "step": 11012
    },
    {
      "epoch": 0.37702841492639505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4728,
      "step": 11013
    },
    {
      "epoch": 0.37706264977747345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4769,
      "step": 11014
    },
    {
      "epoch": 0.37709688462855184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4945,
      "step": 11015
    },
    {
      "epoch": 0.37713111947963024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4841,
      "step": 11016
    },
    {
      "epoch": 0.37716535433070864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4807,
      "step": 11017
    },
    {
      "epoch": 0.37719958918178703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4821,
      "step": 11018
    },
    {
      "epoch": 0.3772338240328655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.475,
      "step": 11019
    },
    {
      "epoch": 0.3772680588839439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4922,
      "step": 11020
    },
    {
      "epoch": 0.3773022937350223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4849,
      "step": 11021
    },
    {
      "epoch": 0.3773365285861007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4644,
      "step": 11022
    },
    {
      "epoch": 0.37737076343717907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4556,
      "step": 11023
    },
    {
      "epoch": 0.37740499828825746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4948,
      "step": 11024
    },
    {
      "epoch": 0.37743923313933586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4719,
      "step": 11025
    },
    {
      "epoch": 0.37747346799041426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4699,
      "step": 11026
    },
    {
      "epoch": 0.37750770284149265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4644,
      "step": 11027
    },
    {
      "epoch": 0.37754193769257105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4697,
      "step": 11028
    },
    {
      "epoch": 0.37757617254364945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4885,
      "step": 11029
    },
    {
      "epoch": 0.37761040739472784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4775,
      "step": 11030
    },
    {
      "epoch": 0.37764464224580624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4802,
      "step": 11031
    },
    {
      "epoch": 0.37767887709688464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.514,
      "step": 11032
    },
    {
      "epoch": 0.37771311194796303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4884,
      "step": 11033
    },
    {
      "epoch": 0.37774734679904143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4916,
      "step": 11034
    },
    {
      "epoch": 0.3777815816501198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4768,
      "step": 11035
    },
    {
      "epoch": 0.3778158165011982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4804,
      "step": 11036
    },
    {
      "epoch": 0.3778500513522766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4752,
      "step": 11037
    },
    {
      "epoch": 0.377884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4769,
      "step": 11038
    },
    {
      "epoch": 0.3779185210544334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4674,
      "step": 11039
    },
    {
      "epoch": 0.3779527559055118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4998,
      "step": 11040
    },
    {
      "epoch": 0.3779869907565902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4887,
      "step": 11041
    },
    {
      "epoch": 0.3780212256076686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4635,
      "step": 11042
    },
    {
      "epoch": 0.378055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4677,
      "step": 11043
    },
    {
      "epoch": 0.3780896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4851,
      "step": 11044
    },
    {
      "epoch": 0.3781239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5015,
      "step": 11045
    },
    {
      "epoch": 0.3781581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4905,
      "step": 11046
    },
    {
      "epoch": 0.3781923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.478,
      "step": 11047
    },
    {
      "epoch": 0.378226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5117,
      "step": 11048
    },
    {
      "epoch": 0.3782608695652174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4921,
      "step": 11049
    },
    {
      "epoch": 0.37829510441629577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4785,
      "step": 11050
    },
    {
      "epoch": 0.37832933926737417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4697,
      "step": 11051
    },
    {
      "epoch": 0.37836357411845256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4923,
      "step": 11052
    },
    {
      "epoch": 0.37839780896953096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4806,
      "step": 11053
    },
    {
      "epoch": 0.37843204382060935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4798,
      "step": 11054
    },
    {
      "epoch": 0.3784662786716878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4991,
      "step": 11055
    },
    {
      "epoch": 0.3785005135227662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.49,
      "step": 11056
    },
    {
      "epoch": 0.3785347483738446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4892,
      "step": 11057
    },
    {
      "epoch": 0.378568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4805,
      "step": 11058
    },
    {
      "epoch": 0.3786032180760014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4882,
      "step": 11059
    },
    {
      "epoch": 0.3786374529270798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4663,
      "step": 11060
    },
    {
      "epoch": 0.3786716877781582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4735,
      "step": 11061
    },
    {
      "epoch": 0.3787059226292366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4667,
      "step": 11062
    },
    {
      "epoch": 0.378740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4964,
      "step": 11063
    },
    {
      "epoch": 0.3787743923313934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4688,
      "step": 11064
    },
    {
      "epoch": 0.37880862718247177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4755,
      "step": 11065
    },
    {
      "epoch": 0.37884286203355017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4888,
      "step": 11066
    },
    {
      "epoch": 0.37887709688462856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4777,
      "step": 11067
    },
    {
      "epoch": 0.37891133173570696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4771,
      "step": 11068
    },
    {
      "epoch": 0.37894556658678535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4915,
      "step": 11069
    },
    {
      "epoch": 0.37897980143786375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4559,
      "step": 11070
    },
    {
      "epoch": 0.37901403628894215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.487,
      "step": 11071
    },
    {
      "epoch": 0.37904827114002054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4705,
      "step": 11072
    },
    {
      "epoch": 0.37908250599109894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4866,
      "step": 11073
    },
    {
      "epoch": 0.37911674084217734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4871,
      "step": 11074
    },
    {
      "epoch": 0.37915097569325573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4754,
      "step": 11075
    },
    {
      "epoch": 0.37918521054433413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4673,
      "step": 11076
    },
    {
      "epoch": 0.3792194453954125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4793,
      "step": 11077
    },
    {
      "epoch": 0.3792536802464909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4899,
      "step": 11078
    },
    {
      "epoch": 0.3792879150975693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4765,
      "step": 11079
    },
    {
      "epoch": 0.3793221499486477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.492,
      "step": 11080
    },
    {
      "epoch": 0.3793563847997261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4815,
      "step": 11081
    },
    {
      "epoch": 0.3793906196508045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4749,
      "step": 11082
    },
    {
      "epoch": 0.3794248545018829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4759,
      "step": 11083
    },
    {
      "epoch": 0.3794590893529613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4846,
      "step": 11084
    },
    {
      "epoch": 0.3794933242040397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4923,
      "step": 11085
    },
    {
      "epoch": 0.3795275590551181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4754,
      "step": 11086
    },
    {
      "epoch": 0.3795617939061965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4724,
      "step": 11087
    },
    {
      "epoch": 0.3795960287572749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.468,
      "step": 11088
    },
    {
      "epoch": 0.3796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4677,
      "step": 11089
    },
    {
      "epoch": 0.3796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4785,
      "step": 11090
    },
    {
      "epoch": 0.3796987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4978,
      "step": 11091
    },
    {
      "epoch": 0.3797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4846,
      "step": 11092
    },
    {
      "epoch": 0.3797672030126669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4961,
      "step": 11093
    },
    {
      "epoch": 0.3798014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4746,
      "step": 11094
    },
    {
      "epoch": 0.3798356727148237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4713,
      "step": 11095
    },
    {
      "epoch": 0.3798699075659021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.477,
      "step": 11096
    },
    {
      "epoch": 0.3799041424169805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4816,
      "step": 11097
    },
    {
      "epoch": 0.3799383772680589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.484,
      "step": 11098
    },
    {
      "epoch": 0.3799726121191373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.469,
      "step": 11099
    },
    {
      "epoch": 0.3800068469702157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4726,
      "step": 11100
    },
    {
      "epoch": 0.3800410818212941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5006,
      "step": 11101
    },
    {
      "epoch": 0.3800753166723725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4764,
      "step": 11102
    },
    {
      "epoch": 0.3801095515234509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4755,
      "step": 11103
    },
    {
      "epoch": 0.3801437863745293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4828,
      "step": 11104
    },
    {
      "epoch": 0.3801780212256077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4782,
      "step": 11105
    },
    {
      "epoch": 0.3802122560766861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4786,
      "step": 11106
    },
    {
      "epoch": 0.38024649092776447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4708,
      "step": 11107
    },
    {
      "epoch": 0.38028072577884287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.486,
      "step": 11108
    },
    {
      "epoch": 0.38031496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4573,
      "step": 11109
    },
    {
      "epoch": 0.38034919548099966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4771,
      "step": 11110
    },
    {
      "epoch": 0.38038343033207805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5119,
      "step": 11111
    },
    {
      "epoch": 0.38041766518315645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4712,
      "step": 11112
    },
    {
      "epoch": 0.38045190003423485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4676,
      "step": 11113
    },
    {
      "epoch": 0.38048613488531324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4741,
      "step": 11114
    },
    {
      "epoch": 0.38052036973639164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5075,
      "step": 11115
    },
    {
      "epoch": 0.38055460458747004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4793,
      "step": 11116
    },
    {
      "epoch": 0.38058883943854843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4701,
      "step": 11117
    },
    {
      "epoch": 0.38062307428962683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4934,
      "step": 11118
    },
    {
      "epoch": 0.3806573091407052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4707,
      "step": 11119
    },
    {
      "epoch": 0.3806915439917836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4616,
      "step": 11120
    },
    {
      "epoch": 0.380725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4589,
      "step": 11121
    },
    {
      "epoch": 0.3807600136939404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4707,
      "step": 11122
    },
    {
      "epoch": 0.3807942485450188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4809,
      "step": 11123
    },
    {
      "epoch": 0.3808284833960972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4855,
      "step": 11124
    },
    {
      "epoch": 0.3808627182471756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4901,
      "step": 11125
    },
    {
      "epoch": 0.380896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 11126
    },
    {
      "epoch": 0.3809311879493324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4819,
      "step": 11127
    },
    {
      "epoch": 0.3809654228004108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4987,
      "step": 11128
    },
    {
      "epoch": 0.38099965765148924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4748,
      "step": 11129
    },
    {
      "epoch": 0.38103389250256764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4756,
      "step": 11130
    },
    {
      "epoch": 0.38106812735364604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4794,
      "step": 11131
    },
    {
      "epoch": 0.38110236220472443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4645,
      "step": 11132
    },
    {
      "epoch": 0.38113659705580283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4865,
      "step": 11133
    },
    {
      "epoch": 0.3811708319068812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4753,
      "step": 11134
    },
    {
      "epoch": 0.3812050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4569,
      "step": 11135
    },
    {
      "epoch": 0.381239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.473,
      "step": 11136
    },
    {
      "epoch": 0.3812735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4793,
      "step": 11137
    },
    {
      "epoch": 0.3813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4826,
      "step": 11138
    },
    {
      "epoch": 0.3813420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4765,
      "step": 11139
    },
    {
      "epoch": 0.3813762410133516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4836,
      "step": 11140
    },
    {
      "epoch": 0.38141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4751,
      "step": 11141
    },
    {
      "epoch": 0.3814447107155084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4667,
      "step": 11142
    },
    {
      "epoch": 0.3814789455665868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4738,
      "step": 11143
    },
    {
      "epoch": 0.3815131804176652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.466,
      "step": 11144
    },
    {
      "epoch": 0.3815474152687436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4793,
      "step": 11145
    },
    {
      "epoch": 0.381581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4733,
      "step": 11146
    },
    {
      "epoch": 0.3816158849709004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.492,
      "step": 11147
    },
    {
      "epoch": 0.3816501198219788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4708,
      "step": 11148
    },
    {
      "epoch": 0.38168435467305717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4772,
      "step": 11149
    },
    {
      "epoch": 0.38171858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4729,
      "step": 11150
    },
    {
      "epoch": 0.38175282437521396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4608,
      "step": 11151
    },
    {
      "epoch": 0.38178705922629236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4716,
      "step": 11152
    },
    {
      "epoch": 0.38182129407737075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4738,
      "step": 11153
    },
    {
      "epoch": 0.38185552892844915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4714,
      "step": 11154
    },
    {
      "epoch": 0.38188976377952755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.475,
      "step": 11155
    },
    {
      "epoch": 0.38192399863060594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4727,
      "step": 11156
    },
    {
      "epoch": 0.38195823348168434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4694,
      "step": 11157
    },
    {
      "epoch": 0.38199246833276274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.463,
      "step": 11158
    },
    {
      "epoch": 0.38202670318384113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4891,
      "step": 11159
    },
    {
      "epoch": 0.38206093803491953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4755,
      "step": 11160
    },
    {
      "epoch": 0.3820951728859979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4799,
      "step": 11161
    },
    {
      "epoch": 0.3821294077370763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 11162
    },
    {
      "epoch": 0.3821636425881547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4795,
      "step": 11163
    },
    {
      "epoch": 0.3821978774392331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4866,
      "step": 11164
    },
    {
      "epoch": 0.3822321122903115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4814,
      "step": 11165
    },
    {
      "epoch": 0.38226634714138996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4772,
      "step": 11166
    },
    {
      "epoch": 0.38230058199246836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4606,
      "step": 11167
    },
    {
      "epoch": 0.38233481684354675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4714,
      "step": 11168
    },
    {
      "epoch": 0.38236905169462515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.471,
      "step": 11169
    },
    {
      "epoch": 0.38240328654570355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4691,
      "step": 11170
    },
    {
      "epoch": 0.38243752139678194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4923,
      "step": 11171
    },
    {
      "epoch": 0.38247175624786034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4777,
      "step": 11172
    },
    {
      "epoch": 0.38250599109893874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4528,
      "step": 11173
    },
    {
      "epoch": 0.38254022595001713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4712,
      "step": 11174
    },
    {
      "epoch": 0.38257446080109553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4577,
      "step": 11175
    },
    {
      "epoch": 0.3826086956521739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4803,
      "step": 11176
    },
    {
      "epoch": 0.3826429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4813,
      "step": 11177
    },
    {
      "epoch": 0.3826771653543307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.477,
      "step": 11178
    },
    {
      "epoch": 0.3827114002054091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4771,
      "step": 11179
    },
    {
      "epoch": 0.3827456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4803,
      "step": 11180
    },
    {
      "epoch": 0.3827798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4705,
      "step": 11181
    },
    {
      "epoch": 0.3828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4759,
      "step": 11182
    },
    {
      "epoch": 0.3828483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.479,
      "step": 11183
    },
    {
      "epoch": 0.3828825744608011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4901,
      "step": 11184
    },
    {
      "epoch": 0.3829168093118795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4901,
      "step": 11185
    },
    {
      "epoch": 0.3829510441629579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4619,
      "step": 11186
    },
    {
      "epoch": 0.3829852790140363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4841,
      "step": 11187
    },
    {
      "epoch": 0.3830195138651147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4682,
      "step": 11188
    },
    {
      "epoch": 0.3830537487161931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4575,
      "step": 11189
    },
    {
      "epoch": 0.3830879835672715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4849,
      "step": 11190
    },
    {
      "epoch": 0.38312221841834987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4808,
      "step": 11191
    },
    {
      "epoch": 0.38315645326942827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4763,
      "step": 11192
    },
    {
      "epoch": 0.38319068812050666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4758,
      "step": 11193
    },
    {
      "epoch": 0.38322492297158506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4878,
      "step": 11194
    },
    {
      "epoch": 0.38325915782266345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4678,
      "step": 11195
    },
    {
      "epoch": 0.38329339267374185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4673,
      "step": 11196
    },
    {
      "epoch": 0.38332762752482025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4691,
      "step": 11197
    },
    {
      "epoch": 0.38336186237589864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4644,
      "step": 11198
    },
    {
      "epoch": 0.38339609722697704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4664,
      "step": 11199
    },
    {
      "epoch": 0.38343033207805544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4699,
      "step": 11200
    },
    {
      "epoch": 0.38346456692913383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4787,
      "step": 11201
    },
    {
      "epoch": 0.38349880178021223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4642,
      "step": 11202
    },
    {
      "epoch": 0.3835330366312907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4833,
      "step": 11203
    },
    {
      "epoch": 0.3835672714823691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4785,
      "step": 11204
    },
    {
      "epoch": 0.3836015063334475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4622,
      "step": 11205
    },
    {
      "epoch": 0.38363574118452587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4671,
      "step": 11206
    },
    {
      "epoch": 0.38366997603560427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4892,
      "step": 11207
    },
    {
      "epoch": 0.38370421088668266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4736,
      "step": 11208
    },
    {
      "epoch": 0.38373844573776106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4526,
      "step": 11209
    },
    {
      "epoch": 0.38377268058883945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4904,
      "step": 11210
    },
    {
      "epoch": 0.38380691543991785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4965,
      "step": 11211
    },
    {
      "epoch": 0.38384115029099625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4668,
      "step": 11212
    },
    {
      "epoch": 0.38387538514207464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4628,
      "step": 11213
    },
    {
      "epoch": 0.38390961999315304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4763,
      "step": 11214
    },
    {
      "epoch": 0.38394385484423144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4523,
      "step": 11215
    },
    {
      "epoch": 0.38397808969530983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4828,
      "step": 11216
    },
    {
      "epoch": 0.38401232454638823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4694,
      "step": 11217
    },
    {
      "epoch": 0.3840465593974666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4741,
      "step": 11218
    },
    {
      "epoch": 0.384080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4642,
      "step": 11219
    },
    {
      "epoch": 0.3841150290996234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4675,
      "step": 11220
    },
    {
      "epoch": 0.3841492639507018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4681,
      "step": 11221
    },
    {
      "epoch": 0.3841834988017802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4727,
      "step": 11222
    },
    {
      "epoch": 0.3842177336528586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4895,
      "step": 11223
    },
    {
      "epoch": 0.384251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4835,
      "step": 11224
    },
    {
      "epoch": 0.3842862033550154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4771,
      "step": 11225
    },
    {
      "epoch": 0.3843204382060938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4643,
      "step": 11226
    },
    {
      "epoch": 0.3843546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.476,
      "step": 11227
    },
    {
      "epoch": 0.3843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4753,
      "step": 11228
    },
    {
      "epoch": 0.384423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4688,
      "step": 11229
    },
    {
      "epoch": 0.3844573776104074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4668,
      "step": 11230
    },
    {
      "epoch": 0.3844916124614858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4728,
      "step": 11231
    },
    {
      "epoch": 0.3845258473125642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4768,
      "step": 11232
    },
    {
      "epoch": 0.38456008216364257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4804,
      "step": 11233
    },
    {
      "epoch": 0.38459431701472097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4658,
      "step": 11234
    },
    {
      "epoch": 0.38462855186579936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4788,
      "step": 11235
    },
    {
      "epoch": 0.38466278671687776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4584,
      "step": 11236
    },
    {
      "epoch": 0.38469702156795615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4606,
      "step": 11237
    },
    {
      "epoch": 0.38473125641903455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4714,
      "step": 11238
    },
    {
      "epoch": 0.384765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4801,
      "step": 11239
    },
    {
      "epoch": 0.3847997261211914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4799,
      "step": 11240
    },
    {
      "epoch": 0.3848339609722698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4675,
      "step": 11241
    },
    {
      "epoch": 0.3848681958233482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4766,
      "step": 11242
    },
    {
      "epoch": 0.3849024306744266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4829,
      "step": 11243
    },
    {
      "epoch": 0.384936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4707,
      "step": 11244
    },
    {
      "epoch": 0.3849709003765834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4952,
      "step": 11245
    },
    {
      "epoch": 0.3850051352276618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.47,
      "step": 11246
    },
    {
      "epoch": 0.3850393700787402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4874,
      "step": 11247
    },
    {
      "epoch": 0.38507360492981857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4772,
      "step": 11248
    },
    {
      "epoch": 0.38510783978089697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4642,
      "step": 11249
    },
    {
      "epoch": 0.38514207463197536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.492,
      "step": 11250
    },
    {
      "epoch": 0.38517630948305376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4726,
      "step": 11251
    },
    {
      "epoch": 0.38521054433413215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4644,
      "step": 11252
    },
    {
      "epoch": 0.38524477918521055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4815,
      "step": 11253
    },
    {
      "epoch": 0.38527901403628895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4869,
      "step": 11254
    },
    {
      "epoch": 0.38531324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4751,
      "step": 11255
    },
    {
      "epoch": 0.38534748373844574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4748,
      "step": 11256
    },
    {
      "epoch": 0.38538171858952414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4777,
      "step": 11257
    },
    {
      "epoch": 0.38541595344060253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4744,
      "step": 11258
    },
    {
      "epoch": 0.38545018829168093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4747,
      "step": 11259
    },
    {
      "epoch": 0.3854844231427593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4753,
      "step": 11260
    },
    {
      "epoch": 0.3855186579938377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4781,
      "step": 11261
    },
    {
      "epoch": 0.3855528928449161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4777,
      "step": 11262
    },
    {
      "epoch": 0.3855871276959945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4692,
      "step": 11263
    },
    {
      "epoch": 0.3856213625470729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4692,
      "step": 11264
    },
    {
      "epoch": 0.3856555973981513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.471,
      "step": 11265
    },
    {
      "epoch": 0.3856898322492297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4755,
      "step": 11266
    },
    {
      "epoch": 0.3857240671003081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4658,
      "step": 11267
    },
    {
      "epoch": 0.3857583019513865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4709,
      "step": 11268
    },
    {
      "epoch": 0.3857925368024649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4663,
      "step": 11269
    },
    {
      "epoch": 0.3858267716535433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4916,
      "step": 11270
    },
    {
      "epoch": 0.3858610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4882,
      "step": 11271
    },
    {
      "epoch": 0.3858952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4719,
      "step": 11272
    },
    {
      "epoch": 0.3859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4734,
      "step": 11273
    },
    {
      "epoch": 0.3859637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4716,
      "step": 11274
    },
    {
      "epoch": 0.38599794590893527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4617,
      "step": 11275
    },
    {
      "epoch": 0.3860321807600137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4621,
      "step": 11276
    },
    {
      "epoch": 0.3860664156110921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4561,
      "step": 11277
    },
    {
      "epoch": 0.3861006504621705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4827,
      "step": 11278
    },
    {
      "epoch": 0.3861348853132489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.497,
      "step": 11279
    },
    {
      "epoch": 0.3861691201643273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4657,
      "step": 11280
    },
    {
      "epoch": 0.3862033550154057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4603,
      "step": 11281
    },
    {
      "epoch": 0.3862375898664841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4594,
      "step": 11282
    },
    {
      "epoch": 0.3862718247175625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4578,
      "step": 11283
    },
    {
      "epoch": 0.3863060595686409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4724,
      "step": 11284
    },
    {
      "epoch": 0.3863402944197193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.466,
      "step": 11285
    },
    {
      "epoch": 0.3863745292707977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4577,
      "step": 11286
    },
    {
      "epoch": 0.3864087641218761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4613,
      "step": 11287
    },
    {
      "epoch": 0.3864429989729545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4795,
      "step": 11288
    },
    {
      "epoch": 0.3864772338240329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4574,
      "step": 11289
    },
    {
      "epoch": 0.38651146867511127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4541,
      "step": 11290
    },
    {
      "epoch": 0.38654570352618967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4868,
      "step": 11291
    },
    {
      "epoch": 0.38657993837726806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4746,
      "step": 11292
    },
    {
      "epoch": 0.38661417322834646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4592,
      "step": 11293
    },
    {
      "epoch": 0.38664840807942485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4743,
      "step": 11294
    },
    {
      "epoch": 0.38668264293050325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4606,
      "step": 11295
    },
    {
      "epoch": 0.38671687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4613,
      "step": 11296
    },
    {
      "epoch": 0.38675111263266004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4535,
      "step": 11297
    },
    {
      "epoch": 0.38678534748373844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4526,
      "step": 11298
    },
    {
      "epoch": 0.38681958233481684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4694,
      "step": 11299
    },
    {
      "epoch": 0.38685381718589523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.47,
      "step": 11300
    },
    {
      "epoch": 0.38688805203697363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4715,
      "step": 11301
    },
    {
      "epoch": 0.386922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4667,
      "step": 11302
    },
    {
      "epoch": 0.3869565217391304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4768,
      "step": 11303
    },
    {
      "epoch": 0.3869907565902088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4708,
      "step": 11304
    },
    {
      "epoch": 0.3870249914412872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4665,
      "step": 11305
    },
    {
      "epoch": 0.3870592262923656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4813,
      "step": 11306
    },
    {
      "epoch": 0.387093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 11307
    },
    {
      "epoch": 0.3871276959945224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4647,
      "step": 11308
    },
    {
      "epoch": 0.3871619308456008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4549,
      "step": 11309
    },
    {
      "epoch": 0.3871961656966792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4624,
      "step": 11310
    },
    {
      "epoch": 0.3872304005477576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4785,
      "step": 11311
    },
    {
      "epoch": 0.387264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4657,
      "step": 11312
    },
    {
      "epoch": 0.38729887024991444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4733,
      "step": 11313
    },
    {
      "epoch": 0.38733310510099284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.469,
      "step": 11314
    },
    {
      "epoch": 0.38736733995207123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4733,
      "step": 11315
    },
    {
      "epoch": 0.38740157480314963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4723,
      "step": 11316
    },
    {
      "epoch": 0.387435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4643,
      "step": 11317
    },
    {
      "epoch": 0.3874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4863,
      "step": 11318
    },
    {
      "epoch": 0.3875042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4667,
      "step": 11319
    },
    {
      "epoch": 0.3875385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4811,
      "step": 11320
    },
    {
      "epoch": 0.3875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4975,
      "step": 11321
    },
    {
      "epoch": 0.38760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4625,
      "step": 11322
    },
    {
      "epoch": 0.3876412187606984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4656,
      "step": 11323
    },
    {
      "epoch": 0.3876754536117768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4664,
      "step": 11324
    },
    {
      "epoch": 0.3877096884628552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4846,
      "step": 11325
    },
    {
      "epoch": 0.3877439233139336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4727,
      "step": 11326
    },
    {
      "epoch": 0.387778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4678,
      "step": 11327
    },
    {
      "epoch": 0.3878123930160904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4639,
      "step": 11328
    },
    {
      "epoch": 0.3878466278671688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4677,
      "step": 11329
    },
    {
      "epoch": 0.3878808627182472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.466,
      "step": 11330
    },
    {
      "epoch": 0.3879150975693256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4819,
      "step": 11331
    },
    {
      "epoch": 0.38794933242040397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4715,
      "step": 11332
    },
    {
      "epoch": 0.38798356727148237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4736,
      "step": 11333
    },
    {
      "epoch": 0.38801780212256076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4627,
      "step": 11334
    },
    {
      "epoch": 0.38805203697363916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4591,
      "step": 11335
    },
    {
      "epoch": 0.38808627182471755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.485,
      "step": 11336
    },
    {
      "epoch": 0.38812050667579595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4791,
      "step": 11337
    },
    {
      "epoch": 0.38815474152687435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4628,
      "step": 11338
    },
    {
      "epoch": 0.38818897637795274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4699,
      "step": 11339
    },
    {
      "epoch": 0.38822321122903114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4671,
      "step": 11340
    },
    {
      "epoch": 0.38825744608010954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4783,
      "step": 11341
    },
    {
      "epoch": 0.38829168093118793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4517,
      "step": 11342
    },
    {
      "epoch": 0.38832591578226633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4627,
      "step": 11343
    },
    {
      "epoch": 0.3883601506333447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4733,
      "step": 11344
    },
    {
      "epoch": 0.3883943854844231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4658,
      "step": 11345
    },
    {
      "epoch": 0.3884286203355015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4525,
      "step": 11346
    },
    {
      "epoch": 0.3884628551865799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4718,
      "step": 11347
    },
    {
      "epoch": 0.3884970900376583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 11348
    },
    {
      "epoch": 0.3885313248887367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4866,
      "step": 11349
    },
    {
      "epoch": 0.38856555973981516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4675,
      "step": 11350
    },
    {
      "epoch": 0.38859979459089355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4617,
      "step": 11351
    },
    {
      "epoch": 0.38863402944197195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4575,
      "step": 11352
    },
    {
      "epoch": 0.38866826429305035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4556,
      "step": 11353
    },
    {
      "epoch": 0.38870249914412874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4585,
      "step": 11354
    },
    {
      "epoch": 0.38873673399520714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4557,
      "step": 11355
    },
    {
      "epoch": 0.38877096884628554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4738,
      "step": 11356
    },
    {
      "epoch": 0.38880520369736393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4789,
      "step": 11357
    },
    {
      "epoch": 0.38883943854844233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4845,
      "step": 11358
    },
    {
      "epoch": 0.3888736733995207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 11359
    },
    {
      "epoch": 0.3889079082505991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4538,
      "step": 11360
    },
    {
      "epoch": 0.3889421431016775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4678,
      "step": 11361
    },
    {
      "epoch": 0.3889763779527559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4724,
      "step": 11362
    },
    {
      "epoch": 0.3890106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4738,
      "step": 11363
    },
    {
      "epoch": 0.3890448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4564,
      "step": 11364
    },
    {
      "epoch": 0.3890790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4705,
      "step": 11365
    },
    {
      "epoch": 0.3891133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4721,
      "step": 11366
    },
    {
      "epoch": 0.3891475522081479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4605,
      "step": 11367
    },
    {
      "epoch": 0.3891817870592263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4808,
      "step": 11368
    },
    {
      "epoch": 0.3892160219103047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4724,
      "step": 11369
    },
    {
      "epoch": 0.3892502567613831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4855,
      "step": 11370
    },
    {
      "epoch": 0.3892844916124615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4891,
      "step": 11371
    },
    {
      "epoch": 0.3893187264635399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4827,
      "step": 11372
    },
    {
      "epoch": 0.3893529613146183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 11373
    },
    {
      "epoch": 0.38938719616569667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4901,
      "step": 11374
    },
    {
      "epoch": 0.38942143101677507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4638,
      "step": 11375
    },
    {
      "epoch": 0.38945566586785346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4642,
      "step": 11376
    },
    {
      "epoch": 0.38948990071893186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4716,
      "step": 11377
    },
    {
      "epoch": 0.38952413557001025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.457,
      "step": 11378
    },
    {
      "epoch": 0.38955837042108865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4644,
      "step": 11379
    },
    {
      "epoch": 0.38959260527216705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4733,
      "step": 11380
    },
    {
      "epoch": 0.38962684012324544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4402,
      "step": 11381
    },
    {
      "epoch": 0.38966107497432384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4915,
      "step": 11382
    },
    {
      "epoch": 0.38969530982540224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4468,
      "step": 11383
    },
    {
      "epoch": 0.38972954467648063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4851,
      "step": 11384
    },
    {
      "epoch": 0.38976377952755903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4686,
      "step": 11385
    },
    {
      "epoch": 0.3897980143786374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.5031,
      "step": 11386
    },
    {
      "epoch": 0.3898322492297159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4808,
      "step": 11387
    },
    {
      "epoch": 0.3898664840807943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4539,
      "step": 11388
    },
    {
      "epoch": 0.38990071893187267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4681,
      "step": 11389
    },
    {
      "epoch": 0.38993495378295107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4501,
      "step": 11390
    },
    {
      "epoch": 0.38996918863402946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4457,
      "step": 11391
    },
    {
      "epoch": 0.39000342348510786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.452,
      "step": 11392
    },
    {
      "epoch": 0.39003765833618625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4442,
      "step": 11393
    },
    {
      "epoch": 0.39007189318726465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4593,
      "step": 11394
    },
    {
      "epoch": 0.39010612803834305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4686,
      "step": 11395
    },
    {
      "epoch": 0.39014036288942144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4726,
      "step": 11396
    },
    {
      "epoch": 0.39017459774049984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4858,
      "step": 11397
    },
    {
      "epoch": 0.39020883259157824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4653,
      "step": 11398
    },
    {
      "epoch": 0.39024306744265663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4534,
      "step": 11399
    },
    {
      "epoch": 0.39027730229373503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 11400
    },
    {
      "epoch": 0.3903115371448134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4881,
      "step": 11401
    },
    {
      "epoch": 0.3903457719958918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4691,
      "step": 11402
    },
    {
      "epoch": 0.3903800068469702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4727,
      "step": 11403
    },
    {
      "epoch": 0.3904142416980486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4639,
      "step": 11404
    },
    {
      "epoch": 0.390448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4689,
      "step": 11405
    },
    {
      "epoch": 0.3904827114002054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4641,
      "step": 11406
    },
    {
      "epoch": 0.3905169462512838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4497,
      "step": 11407
    },
    {
      "epoch": 0.3905511811023622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4777,
      "step": 11408
    },
    {
      "epoch": 0.3905854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4632,
      "step": 11409
    },
    {
      "epoch": 0.390619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4727,
      "step": 11410
    },
    {
      "epoch": 0.3906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4714,
      "step": 11411
    },
    {
      "epoch": 0.3906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4623,
      "step": 11412
    },
    {
      "epoch": 0.3907223553577542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.466,
      "step": 11413
    },
    {
      "epoch": 0.3907565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.472,
      "step": 11414
    },
    {
      "epoch": 0.390790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.473,
      "step": 11415
    },
    {
      "epoch": 0.39082505991098937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4619,
      "step": 11416
    },
    {
      "epoch": 0.39085929476206777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4696,
      "step": 11417
    },
    {
      "epoch": 0.39089352961314616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4602,
      "step": 11418
    },
    {
      "epoch": 0.39092776446422456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4638,
      "step": 11419
    },
    {
      "epoch": 0.39096199931530295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4275,
      "step": 11420
    },
    {
      "epoch": 0.39099623416638135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.457,
      "step": 11421
    },
    {
      "epoch": 0.39103046901745975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4898,
      "step": 11422
    },
    {
      "epoch": 0.3910647038685382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4696,
      "step": 11423
    },
    {
      "epoch": 0.3910989387196166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4702,
      "step": 11424
    },
    {
      "epoch": 0.391133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4744,
      "step": 11425
    },
    {
      "epoch": 0.3911674084217734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.465,
      "step": 11426
    },
    {
      "epoch": 0.3912016432728518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4749,
      "step": 11427
    },
    {
      "epoch": 0.3912358781239302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 11428
    },
    {
      "epoch": 0.3912701129750086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4331,
      "step": 11429
    },
    {
      "epoch": 0.391304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4564,
      "step": 11430
    },
    {
      "epoch": 0.39133858267716537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4807,
      "step": 11431
    },
    {
      "epoch": 0.39137281752824377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4631,
      "step": 11432
    },
    {
      "epoch": 0.39140705237932216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4575,
      "step": 11433
    },
    {
      "epoch": 0.39144128723040056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4645,
      "step": 11434
    },
    {
      "epoch": 0.39147552208147895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4451,
      "step": 11435
    },
    {
      "epoch": 0.39150975693255735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4786,
      "step": 11436
    },
    {
      "epoch": 0.39154399178363575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4761,
      "step": 11437
    },
    {
      "epoch": 0.39157822663471414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4774,
      "step": 11438
    },
    {
      "epoch": 0.39161246148579254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.466,
      "step": 11439
    },
    {
      "epoch": 0.39164669633687094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4844,
      "step": 11440
    },
    {
      "epoch": 0.39168093118794933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4562,
      "step": 11441
    },
    {
      "epoch": 0.39171516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4548,
      "step": 11442
    },
    {
      "epoch": 0.3917494008901061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4749,
      "step": 11443
    },
    {
      "epoch": 0.3917836357411845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4691,
      "step": 11444
    },
    {
      "epoch": 0.3918178705922629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4479,
      "step": 11445
    },
    {
      "epoch": 0.3918521054433413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4513,
      "step": 11446
    },
    {
      "epoch": 0.3918863402944197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4562,
      "step": 11447
    },
    {
      "epoch": 0.3919205751454981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.451,
      "step": 11448
    },
    {
      "epoch": 0.3919548099965765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4556,
      "step": 11449
    },
    {
      "epoch": 0.3919890448476549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4692,
      "step": 11450
    },
    {
      "epoch": 0.3920232796987333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4772,
      "step": 11451
    },
    {
      "epoch": 0.3920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4717,
      "step": 11452
    },
    {
      "epoch": 0.3920917494008901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 11453
    },
    {
      "epoch": 0.3921259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 11454
    },
    {
      "epoch": 0.3921602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4816,
      "step": 11455
    },
    {
      "epoch": 0.3921944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4554,
      "step": 11456
    },
    {
      "epoch": 0.3922286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4771,
      "step": 11457
    },
    {
      "epoch": 0.39226292365628207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4805,
      "step": 11458
    },
    {
      "epoch": 0.39229715850736047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4729,
      "step": 11459
    },
    {
      "epoch": 0.3923313933584389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4596,
      "step": 11460
    },
    {
      "epoch": 0.3923656282095173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4722,
      "step": 11461
    },
    {
      "epoch": 0.3923998630605957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4533,
      "step": 11462
    },
    {
      "epoch": 0.3924340979116741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.471,
      "step": 11463
    },
    {
      "epoch": 0.3924683327627525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4587,
      "step": 11464
    },
    {
      "epoch": 0.3925025676138309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4611,
      "step": 11465
    },
    {
      "epoch": 0.3925368024649093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4711,
      "step": 11466
    },
    {
      "epoch": 0.3925710373159877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4767,
      "step": 11467
    },
    {
      "epoch": 0.3926052721670661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4551,
      "step": 11468
    },
    {
      "epoch": 0.3926395070181445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4481,
      "step": 11469
    },
    {
      "epoch": 0.3926737418692229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4601,
      "step": 11470
    },
    {
      "epoch": 0.3927079767203013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4525,
      "step": 11471
    },
    {
      "epoch": 0.3927422115713797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4715,
      "step": 11472
    },
    {
      "epoch": 0.39277644642245807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4647,
      "step": 11473
    },
    {
      "epoch": 0.39281068127353647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4438,
      "step": 11474
    },
    {
      "epoch": 0.39284491612461486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.476,
      "step": 11475
    },
    {
      "epoch": 0.39287915097569326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4644,
      "step": 11476
    },
    {
      "epoch": 0.39291338582677166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4508,
      "step": 11477
    },
    {
      "epoch": 0.39294762067785005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4595,
      "step": 11478
    },
    {
      "epoch": 0.39298185552892845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.484,
      "step": 11479
    },
    {
      "epoch": 0.39301609038000684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4488,
      "step": 11480
    },
    {
      "epoch": 0.39305032523108524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4663,
      "step": 11481
    },
    {
      "epoch": 0.39308456008216364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4663,
      "step": 11482
    },
    {
      "epoch": 0.39311879493324203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4467,
      "step": 11483
    },
    {
      "epoch": 0.39315302978432043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4558,
      "step": 11484
    },
    {
      "epoch": 0.3931872646353988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4526,
      "step": 11485
    },
    {
      "epoch": 0.3932214994864772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4589,
      "step": 11486
    },
    {
      "epoch": 0.3932557343375556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4538,
      "step": 11487
    },
    {
      "epoch": 0.393289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4508,
      "step": 11488
    },
    {
      "epoch": 0.3933242040397124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4641,
      "step": 11489
    },
    {
      "epoch": 0.3933584388907908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4487,
      "step": 11490
    },
    {
      "epoch": 0.3933926737418692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4628,
      "step": 11491
    },
    {
      "epoch": 0.3934269085929476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4761,
      "step": 11492
    },
    {
      "epoch": 0.393461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4722,
      "step": 11493
    },
    {
      "epoch": 0.3934953782951044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4493,
      "step": 11494
    },
    {
      "epoch": 0.3935296131461828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.484,
      "step": 11495
    },
    {
      "epoch": 0.3935638479972612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.458,
      "step": 11496
    },
    {
      "epoch": 0.39359808284833964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4406,
      "step": 11497
    },
    {
      "epoch": 0.39363231769941803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4583,
      "step": 11498
    },
    {
      "epoch": 0.39366655255049643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4541,
      "step": 11499
    },
    {
      "epoch": 0.3937007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4689,
      "step": 11500
    },
    {
      "epoch": 0.3937350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4688,
      "step": 11501
    },
    {
      "epoch": 0.3937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4758,
      "step": 11502
    },
    {
      "epoch": 0.39380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4663,
      "step": 11503
    },
    {
      "epoch": 0.3938377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4807,
      "step": 11504
    },
    {
      "epoch": 0.3938719616569668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4727,
      "step": 11505
    },
    {
      "epoch": 0.3939061965080452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4597,
      "step": 11506
    },
    {
      "epoch": 0.3939404313591236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4483,
      "step": 11507
    },
    {
      "epoch": 0.393974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4429,
      "step": 11508
    },
    {
      "epoch": 0.3940089010612804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4662,
      "step": 11509
    },
    {
      "epoch": 0.3940431359123588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4567,
      "step": 11510
    },
    {
      "epoch": 0.3940773707634372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.463,
      "step": 11511
    },
    {
      "epoch": 0.3941116056145156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.455,
      "step": 11512
    },
    {
      "epoch": 0.394145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.478,
      "step": 11513
    },
    {
      "epoch": 0.3941800753166724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4899,
      "step": 11514
    },
    {
      "epoch": 0.39421431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4676,
      "step": 11515
    },
    {
      "epoch": 0.39424854501882917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4733,
      "step": 11516
    },
    {
      "epoch": 0.39428277986990756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4714,
      "step": 11517
    },
    {
      "epoch": 0.39431701472098596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4643,
      "step": 11518
    },
    {
      "epoch": 0.39435124957206436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4435,
      "step": 11519
    },
    {
      "epoch": 0.39438548442314275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4596,
      "step": 11520
    },
    {
      "epoch": 0.39441971927422115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.475,
      "step": 11521
    },
    {
      "epoch": 0.39445395412529954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4688,
      "step": 11522
    },
    {
      "epoch": 0.39448818897637794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4483,
      "step": 11523
    },
    {
      "epoch": 0.39452242382745634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4764,
      "step": 11524
    },
    {
      "epoch": 0.39455665867853473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4652,
      "step": 11525
    },
    {
      "epoch": 0.39459089352961313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4695,
      "step": 11526
    },
    {
      "epoch": 0.3946251283806915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4543,
      "step": 11527
    },
    {
      "epoch": 0.3946593632317699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4754,
      "step": 11528
    },
    {
      "epoch": 0.3946935980828483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4779,
      "step": 11529
    },
    {
      "epoch": 0.3947278329339267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4611,
      "step": 11530
    },
    {
      "epoch": 0.3947620677850051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.474,
      "step": 11531
    },
    {
      "epoch": 0.3947963026360835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4677,
      "step": 11532
    },
    {
      "epoch": 0.3948305374871619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4636,
      "step": 11533
    },
    {
      "epoch": 0.39486477233824036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4535,
      "step": 11534
    },
    {
      "epoch": 0.39489900718931875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4509,
      "step": 11535
    },
    {
      "epoch": 0.39493324204039715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4692,
      "step": 11536
    },
    {
      "epoch": 0.39496747689147554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4457,
      "step": 11537
    },
    {
      "epoch": 0.39500171174255394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.46,
      "step": 11538
    },
    {
      "epoch": 0.39503594659363234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4768,
      "step": 11539
    },
    {
      "epoch": 0.39507018144471073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4703,
      "step": 11540
    },
    {
      "epoch": 0.39510441629578913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4725,
      "step": 11541
    },
    {
      "epoch": 0.3951386511468675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4753,
      "step": 11542
    },
    {
      "epoch": 0.3951728859979459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4501,
      "step": 11543
    },
    {
      "epoch": 0.3952071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4392,
      "step": 11544
    },
    {
      "epoch": 0.3952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4944,
      "step": 11545
    },
    {
      "epoch": 0.3952755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4497,
      "step": 11546
    },
    {
      "epoch": 0.3953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4645,
      "step": 11547
    },
    {
      "epoch": 0.3953440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4733,
      "step": 11548
    },
    {
      "epoch": 0.3953782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4607,
      "step": 11549
    },
    {
      "epoch": 0.3954125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4764,
      "step": 11550
    },
    {
      "epoch": 0.3954467648065731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4536,
      "step": 11551
    },
    {
      "epoch": 0.3954809996576515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4486,
      "step": 11552
    },
    {
      "epoch": 0.3955152345087299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4537,
      "step": 11553
    },
    {
      "epoch": 0.3955494693598083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4708,
      "step": 11554
    },
    {
      "epoch": 0.3955837042108867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4716,
      "step": 11555
    },
    {
      "epoch": 0.3956179390619651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4611,
      "step": 11556
    },
    {
      "epoch": 0.39565217391304347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4515,
      "step": 11557
    },
    {
      "epoch": 0.39568640876412187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.465,
      "step": 11558
    },
    {
      "epoch": 0.39572064361520026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4667,
      "step": 11559
    },
    {
      "epoch": 0.39575487846627866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4742,
      "step": 11560
    },
    {
      "epoch": 0.39578911331735706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4686,
      "step": 11561
    },
    {
      "epoch": 0.39582334816843545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.473,
      "step": 11562
    },
    {
      "epoch": 0.39585758301951385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4465,
      "step": 11563
    },
    {
      "epoch": 0.39589181787059224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4639,
      "step": 11564
    },
    {
      "epoch": 0.39592605272167064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4845,
      "step": 11565
    },
    {
      "epoch": 0.39596028757274904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4575,
      "step": 11566
    },
    {
      "epoch": 0.39599452242382743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.455,
      "step": 11567
    },
    {
      "epoch": 0.39602875727490583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.455,
      "step": 11568
    },
    {
      "epoch": 0.3960629921259842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4609,
      "step": 11569
    },
    {
      "epoch": 0.3960972269770627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4665,
      "step": 11570
    },
    {
      "epoch": 0.3961314618281411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4579,
      "step": 11571
    },
    {
      "epoch": 0.39616569667921947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4664,
      "step": 11572
    },
    {
      "epoch": 0.39619993153029787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.464,
      "step": 11573
    },
    {
      "epoch": 0.39623416638137626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4639,
      "step": 11574
    },
    {
      "epoch": 0.39626840123245466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4617,
      "step": 11575
    },
    {
      "epoch": 0.39630263608353306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4513,
      "step": 11576
    },
    {
      "epoch": 0.39633687093461145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4526,
      "step": 11577
    },
    {
      "epoch": 0.39637110578568985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4436,
      "step": 11578
    },
    {
      "epoch": 0.39640534063676824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4578,
      "step": 11579
    },
    {
      "epoch": 0.39643957548784664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4453,
      "step": 11580
    },
    {
      "epoch": 0.39647381033892504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4774,
      "step": 11581
    },
    {
      "epoch": 0.39650804519000343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.457,
      "step": 11582
    },
    {
      "epoch": 0.39654228004108183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4523,
      "step": 11583
    },
    {
      "epoch": 0.3965765148921602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.462,
      "step": 11584
    },
    {
      "epoch": 0.3966107497432386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4479,
      "step": 11585
    },
    {
      "epoch": 0.396644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4503,
      "step": 11586
    },
    {
      "epoch": 0.3966792194453954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.453,
      "step": 11587
    },
    {
      "epoch": 0.3967134542964738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4587,
      "step": 11588
    },
    {
      "epoch": 0.3967476891475522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4501,
      "step": 11589
    },
    {
      "epoch": 0.3967819239986306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4437,
      "step": 11590
    },
    {
      "epoch": 0.396816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4556,
      "step": 11591
    },
    {
      "epoch": 0.3968503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4487,
      "step": 11592
    },
    {
      "epoch": 0.3968846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4672,
      "step": 11593
    },
    {
      "epoch": 0.3969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.46,
      "step": 11594
    },
    {
      "epoch": 0.3969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4677,
      "step": 11595
    },
    {
      "epoch": 0.396987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4661,
      "step": 11596
    },
    {
      "epoch": 0.3970215679561794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.483,
      "step": 11597
    },
    {
      "epoch": 0.3970558028072578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4558,
      "step": 11598
    },
    {
      "epoch": 0.39709003765833617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.468,
      "step": 11599
    },
    {
      "epoch": 0.39712427250941457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4654,
      "step": 11600
    },
    {
      "epoch": 0.39715850736049296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4503,
      "step": 11601
    },
    {
      "epoch": 0.39719274221157136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4612,
      "step": 11602
    },
    {
      "epoch": 0.39722697706264976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4721,
      "step": 11603
    },
    {
      "epoch": 0.39726121191372815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 11604
    },
    {
      "epoch": 0.39729544676480655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4428,
      "step": 11605
    },
    {
      "epoch": 0.39732968161588494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4614,
      "step": 11606
    },
    {
      "epoch": 0.3973639164669634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.473,
      "step": 11607
    },
    {
      "epoch": 0.3973981513180418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4647,
      "step": 11608
    },
    {
      "epoch": 0.3974323861691202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4717,
      "step": 11609
    },
    {
      "epoch": 0.3974666210201986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4513,
      "step": 11610
    },
    {
      "epoch": 0.397500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4515,
      "step": 11611
    },
    {
      "epoch": 0.3975350907223554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4632,
      "step": 11612
    },
    {
      "epoch": 0.3975693255734338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4425,
      "step": 11613
    },
    {
      "epoch": 0.39760356042451217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4405,
      "step": 11614
    },
    {
      "epoch": 0.39763779527559057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4477,
      "step": 11615
    },
    {
      "epoch": 0.39767203012666896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4461,
      "step": 11616
    },
    {
      "epoch": 0.39770626497774736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4522,
      "step": 11617
    },
    {
      "epoch": 0.39774049982882576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4726,
      "step": 11618
    },
    {
      "epoch": 0.39777473467990415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4587,
      "step": 11619
    },
    {
      "epoch": 0.39780896953098255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4739,
      "step": 11620
    },
    {
      "epoch": 0.39784320438206094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 11621
    },
    {
      "epoch": 0.39787743923313934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4547,
      "step": 11622
    },
    {
      "epoch": 0.39791167408421774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4612,
      "step": 11623
    },
    {
      "epoch": 0.39794590893529613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4419,
      "step": 11624
    },
    {
      "epoch": 0.39798014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4827,
      "step": 11625
    },
    {
      "epoch": 0.3980143786374529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 11626
    },
    {
      "epoch": 0.3980486134885313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 11627
    },
    {
      "epoch": 0.3980828483396097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.473,
      "step": 11628
    },
    {
      "epoch": 0.3981170831906881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4539,
      "step": 11629
    },
    {
      "epoch": 0.3981513180417665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4425,
      "step": 11630
    },
    {
      "epoch": 0.3981855528928449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4736,
      "step": 11631
    },
    {
      "epoch": 0.3982197877439233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4586,
      "step": 11632
    },
    {
      "epoch": 0.3982540225950017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4576,
      "step": 11633
    },
    {
      "epoch": 0.3982882574460801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4486,
      "step": 11634
    },
    {
      "epoch": 0.3983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4569,
      "step": 11635
    },
    {
      "epoch": 0.3983567271482369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4675,
      "step": 11636
    },
    {
      "epoch": 0.3983909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4446,
      "step": 11637
    },
    {
      "epoch": 0.3984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4321,
      "step": 11638
    },
    {
      "epoch": 0.3984594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.476,
      "step": 11639
    },
    {
      "epoch": 0.3984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4548,
      "step": 11640
    },
    {
      "epoch": 0.39852790140362887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4601,
      "step": 11641
    },
    {
      "epoch": 0.39856213625470727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4602,
      "step": 11642
    },
    {
      "epoch": 0.39859637110578566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4667,
      "step": 11643
    },
    {
      "epoch": 0.3986306059568641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4702,
      "step": 11644
    },
    {
      "epoch": 0.3986648408079425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4645,
      "step": 11645
    },
    {
      "epoch": 0.3986990756590209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4608,
      "step": 11646
    },
    {
      "epoch": 0.3987333105100993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4466,
      "step": 11647
    },
    {
      "epoch": 0.3987675453611777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4436,
      "step": 11648
    },
    {
      "epoch": 0.3988017802122561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4624,
      "step": 11649
    },
    {
      "epoch": 0.3988360150633345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4771,
      "step": 11650
    },
    {
      "epoch": 0.3988702499144129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4649,
      "step": 11651
    },
    {
      "epoch": 0.3989044847654913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4592,
      "step": 11652
    },
    {
      "epoch": 0.3989387196165697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4686,
      "step": 11653
    },
    {
      "epoch": 0.3989729544676481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4763,
      "step": 11654
    },
    {
      "epoch": 0.3990071893187265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4468,
      "step": 11655
    },
    {
      "epoch": 0.39904142416980487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.444,
      "step": 11656
    },
    {
      "epoch": 0.39907565902088327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4807,
      "step": 11657
    },
    {
      "epoch": 0.39910989387196166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4728,
      "step": 11658
    },
    {
      "epoch": 0.39914412872304006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4634,
      "step": 11659
    },
    {
      "epoch": 0.39917836357411846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4744,
      "step": 11660
    },
    {
      "epoch": 0.39921259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4572,
      "step": 11661
    },
    {
      "epoch": 0.39924683327627525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4538,
      "step": 11662
    },
    {
      "epoch": 0.39928106812735364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.474,
      "step": 11663
    },
    {
      "epoch": 0.39931530297843204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4537,
      "step": 11664
    },
    {
      "epoch": 0.39934953782951044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4538,
      "step": 11665
    },
    {
      "epoch": 0.39938377268058883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4677,
      "step": 11666
    },
    {
      "epoch": 0.39941800753166723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4674,
      "step": 11667
    },
    {
      "epoch": 0.3994522423827456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4805,
      "step": 11668
    },
    {
      "epoch": 0.399486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4535,
      "step": 11669
    },
    {
      "epoch": 0.3995207120849024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4487,
      "step": 11670
    },
    {
      "epoch": 0.3995549469359808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4581,
      "step": 11671
    },
    {
      "epoch": 0.3995891817870592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4445,
      "step": 11672
    },
    {
      "epoch": 0.3996234166381376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4764,
      "step": 11673
    },
    {
      "epoch": 0.399657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4605,
      "step": 11674
    },
    {
      "epoch": 0.3996918863402944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4607,
      "step": 11675
    },
    {
      "epoch": 0.3997261211913728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4344,
      "step": 11676
    },
    {
      "epoch": 0.3997603560424512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4594,
      "step": 11677
    },
    {
      "epoch": 0.3997945908935296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4807,
      "step": 11678
    },
    {
      "epoch": 0.399828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4625,
      "step": 11679
    },
    {
      "epoch": 0.3998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.479,
      "step": 11680
    },
    {
      "epoch": 0.39989729544676483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4843,
      "step": 11681
    },
    {
      "epoch": 0.39993153029784323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4644,
      "step": 11682
    },
    {
      "epoch": 0.3999657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4722,
      "step": 11683
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4539,
      "step": 11684
    },
    {
      "epoch": 0.4000342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.45,
      "step": 11685
    },
    {
      "epoch": 0.4000684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4434,
      "step": 11686
    },
    {
      "epoch": 0.4001027045532352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4479,
      "step": 11687
    },
    {
      "epoch": 0.4001369394043136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 11688
    },
    {
      "epoch": 0.400171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4716,
      "step": 11689
    },
    {
      "epoch": 0.4002054091064704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4529,
      "step": 11690
    },
    {
      "epoch": 0.4002396439575488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 11691
    },
    {
      "epoch": 0.4002738788086272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4651,
      "step": 11692
    },
    {
      "epoch": 0.4003081136597056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4623,
      "step": 11693
    },
    {
      "epoch": 0.400342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.448,
      "step": 11694
    },
    {
      "epoch": 0.4003765833618624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4381,
      "step": 11695
    },
    {
      "epoch": 0.4004108182129408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4624,
      "step": 11696
    },
    {
      "epoch": 0.4004450530640192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.462,
      "step": 11697
    },
    {
      "epoch": 0.40047928791509757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4267,
      "step": 11698
    },
    {
      "epoch": 0.40051352276617597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4443,
      "step": 11699
    },
    {
      "epoch": 0.40054775761725436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4863,
      "step": 11700
    },
    {
      "epoch": 0.40058199246833276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4701,
      "step": 11701
    },
    {
      "epoch": 0.40061622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4609,
      "step": 11702
    },
    {
      "epoch": 0.40065046217048955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4475,
      "step": 11703
    },
    {
      "epoch": 0.40068469702156795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4303,
      "step": 11704
    },
    {
      "epoch": 0.40071893187264634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4529,
      "step": 11705
    },
    {
      "epoch": 0.40075316672372474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4609,
      "step": 11706
    },
    {
      "epoch": 0.40078740157480314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4464,
      "step": 11707
    },
    {
      "epoch": 0.40082163642588153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4654,
      "step": 11708
    },
    {
      "epoch": 0.40085587127695993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.46,
      "step": 11709
    },
    {
      "epoch": 0.4008901061280383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4432,
      "step": 11710
    },
    {
      "epoch": 0.4009243409791167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4533,
      "step": 11711
    },
    {
      "epoch": 0.4009585758301951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4683,
      "step": 11712
    },
    {
      "epoch": 0.4009928106812735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4604,
      "step": 11713
    },
    {
      "epoch": 0.4010270455323519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4638,
      "step": 11714
    },
    {
      "epoch": 0.4010612803834303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4728,
      "step": 11715
    },
    {
      "epoch": 0.4010955152345087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4371,
      "step": 11716
    },
    {
      "epoch": 0.4011297500855871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 11717
    },
    {
      "epoch": 0.40116398493666555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4376,
      "step": 11718
    },
    {
      "epoch": 0.40119821978774395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4569,
      "step": 11719
    },
    {
      "epoch": 0.40123245463882234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4536,
      "step": 11720
    },
    {
      "epoch": 0.40126668948990074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4592,
      "step": 11721
    },
    {
      "epoch": 0.40130092434097914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4525,
      "step": 11722
    },
    {
      "epoch": 0.40133515919205753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4583,
      "step": 11723
    },
    {
      "epoch": 0.40136939404313593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.462,
      "step": 11724
    },
    {
      "epoch": 0.4014036288942143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4681,
      "step": 11725
    },
    {
      "epoch": 0.4014378637452927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4556,
      "step": 11726
    },
    {
      "epoch": 0.4014720985963711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4566,
      "step": 11727
    },
    {
      "epoch": 0.4015063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4635,
      "step": 11728
    },
    {
      "epoch": 0.4015405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4606,
      "step": 11729
    },
    {
      "epoch": 0.4015748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4604,
      "step": 11730
    },
    {
      "epoch": 0.4016090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4532,
      "step": 11731
    },
    {
      "epoch": 0.4016432728517631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4609,
      "step": 11732
    },
    {
      "epoch": 0.4016775077028415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4437,
      "step": 11733
    },
    {
      "epoch": 0.4017117425539199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4642,
      "step": 11734
    },
    {
      "epoch": 0.4017459774049983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4534,
      "step": 11735
    },
    {
      "epoch": 0.4017802122560767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.461,
      "step": 11736
    },
    {
      "epoch": 0.4018144471071551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4785,
      "step": 11737
    },
    {
      "epoch": 0.4018486819582335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4462,
      "step": 11738
    },
    {
      "epoch": 0.4018829168093119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4533,
      "step": 11739
    },
    {
      "epoch": 0.40191715166039027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4513,
      "step": 11740
    },
    {
      "epoch": 0.40195138651146867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4608,
      "step": 11741
    },
    {
      "epoch": 0.40198562136254706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4639,
      "step": 11742
    },
    {
      "epoch": 0.40201985621362546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4677,
      "step": 11743
    },
    {
      "epoch": 0.40205409106470386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4461,
      "step": 11744
    },
    {
      "epoch": 0.40208832591578225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.443,
      "step": 11745
    },
    {
      "epoch": 0.40212256076686065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4729,
      "step": 11746
    },
    {
      "epoch": 0.40215679561793904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4502,
      "step": 11747
    },
    {
      "epoch": 0.40219103046901744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4454,
      "step": 11748
    },
    {
      "epoch": 0.40222526532009584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4509,
      "step": 11749
    },
    {
      "epoch": 0.40225950017117423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4489,
      "step": 11750
    },
    {
      "epoch": 0.40229373502225263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4472,
      "step": 11751
    },
    {
      "epoch": 0.402327969873331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4452,
      "step": 11752
    },
    {
      "epoch": 0.4023622047244094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4682,
      "step": 11753
    },
    {
      "epoch": 0.4023964395754879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4661,
      "step": 11754
    },
    {
      "epoch": 0.40243067442656627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4517,
      "step": 11755
    },
    {
      "epoch": 0.40246490927764467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4492,
      "step": 11756
    },
    {
      "epoch": 0.40249914412872306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4608,
      "step": 11757
    },
    {
      "epoch": 0.40253337897980146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4614,
      "step": 11758
    },
    {
      "epoch": 0.40256761383087986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.473,
      "step": 11759
    },
    {
      "epoch": 0.40260184868195825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4619,
      "step": 11760
    },
    {
      "epoch": 0.40263608353303665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4519,
      "step": 11761
    },
    {
      "epoch": 0.40267031838411504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 11762
    },
    {
      "epoch": 0.40270455323519344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4619,
      "step": 11763
    },
    {
      "epoch": 0.40273878808627184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4643,
      "step": 11764
    },
    {
      "epoch": 0.40277302293735023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4548,
      "step": 11765
    },
    {
      "epoch": 0.40280725778842863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.469,
      "step": 11766
    },
    {
      "epoch": 0.402841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4465,
      "step": 11767
    },
    {
      "epoch": 0.4028757274905854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4555,
      "step": 11768
    },
    {
      "epoch": 0.4029099623416638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4717,
      "step": 11769
    },
    {
      "epoch": 0.4029441971927422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.457,
      "step": 11770
    },
    {
      "epoch": 0.4029784320438206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4591,
      "step": 11771
    },
    {
      "epoch": 0.403012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4589,
      "step": 11772
    },
    {
      "epoch": 0.4030469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4652,
      "step": 11773
    },
    {
      "epoch": 0.4030811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4479,
      "step": 11774
    },
    {
      "epoch": 0.4031153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4652,
      "step": 11775
    },
    {
      "epoch": 0.4031496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4479,
      "step": 11776
    },
    {
      "epoch": 0.403183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4776,
      "step": 11777
    },
    {
      "epoch": 0.4032180760013694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4677,
      "step": 11778
    },
    {
      "epoch": 0.4032523108524478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4516,
      "step": 11779
    },
    {
      "epoch": 0.4032865457035262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.456,
      "step": 11780
    },
    {
      "epoch": 0.4033207805546046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4727,
      "step": 11781
    },
    {
      "epoch": 0.40335501540568297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4517,
      "step": 11782
    },
    {
      "epoch": 0.40338925025676137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4761,
      "step": 11783
    },
    {
      "epoch": 0.40342348510783976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4562,
      "step": 11784
    },
    {
      "epoch": 0.40345771995891816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4468,
      "step": 11785
    },
    {
      "epoch": 0.40349195480999656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4564,
      "step": 11786
    },
    {
      "epoch": 0.40352618966107495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.464,
      "step": 11787
    },
    {
      "epoch": 0.40356042451215335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4635,
      "step": 11788
    },
    {
      "epoch": 0.40359465936323174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4559,
      "step": 11789
    },
    {
      "epoch": 0.40362889421431014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4489,
      "step": 11790
    },
    {
      "epoch": 0.4036631290653886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.468,
      "step": 11791
    },
    {
      "epoch": 0.403697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4641,
      "step": 11792
    },
    {
      "epoch": 0.4037315987675454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4547,
      "step": 11793
    },
    {
      "epoch": 0.4037658336186238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4659,
      "step": 11794
    },
    {
      "epoch": 0.4038000684697022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4522,
      "step": 11795
    },
    {
      "epoch": 0.4038343033207806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4635,
      "step": 11796
    },
    {
      "epoch": 0.40386853817185897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4831,
      "step": 11797
    },
    {
      "epoch": 0.40390277302293737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4602,
      "step": 11798
    },
    {
      "epoch": 0.40393700787401576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4706,
      "step": 11799
    },
    {
      "epoch": 0.40397124272509416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4383,
      "step": 11800
    },
    {
      "epoch": 0.40400547757617256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4595,
      "step": 11801
    },
    {
      "epoch": 0.40403971242725095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4591,
      "step": 11802
    },
    {
      "epoch": 0.40407394727832935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4735,
      "step": 11803
    },
    {
      "epoch": 0.40410818212940774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4484,
      "step": 11804
    },
    {
      "epoch": 0.40414241698048614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.445,
      "step": 11805
    },
    {
      "epoch": 0.40417665183156454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4617,
      "step": 11806
    },
    {
      "epoch": 0.40421088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4495,
      "step": 11807
    },
    {
      "epoch": 0.40424512153372133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4543,
      "step": 11808
    },
    {
      "epoch": 0.4042793563847997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4639,
      "step": 11809
    },
    {
      "epoch": 0.4043135912358781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4654,
      "step": 11810
    },
    {
      "epoch": 0.4043478260869565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4603,
      "step": 11811
    },
    {
      "epoch": 0.4043820609380349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4598,
      "step": 11812
    },
    {
      "epoch": 0.4044162957891133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4421,
      "step": 11813
    },
    {
      "epoch": 0.4044505306401917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4445,
      "step": 11814
    },
    {
      "epoch": 0.4044847654912701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4591,
      "step": 11815
    },
    {
      "epoch": 0.4045190003423485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4534,
      "step": 11816
    },
    {
      "epoch": 0.4045532351934269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4434,
      "step": 11817
    },
    {
      "epoch": 0.4045874700445053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4625,
      "step": 11818
    },
    {
      "epoch": 0.4046217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4506,
      "step": 11819
    },
    {
      "epoch": 0.4046559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4576,
      "step": 11820
    },
    {
      "epoch": 0.4046901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4545,
      "step": 11821
    },
    {
      "epoch": 0.4047244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4515,
      "step": 11822
    },
    {
      "epoch": 0.4047586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 11823
    },
    {
      "epoch": 0.40479287915097567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4632,
      "step": 11824
    },
    {
      "epoch": 0.40482711400205407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4381,
      "step": 11825
    },
    {
      "epoch": 0.40486134885313246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4667,
      "step": 11826
    },
    {
      "epoch": 0.40489558370421086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4448,
      "step": 11827
    },
    {
      "epoch": 0.4049298185552893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.441,
      "step": 11828
    },
    {
      "epoch": 0.4049640534063677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4374,
      "step": 11829
    },
    {
      "epoch": 0.4049982882574461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4508,
      "step": 11830
    },
    {
      "epoch": 0.4050325231085245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4576,
      "step": 11831
    },
    {
      "epoch": 0.4050667579596029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4372,
      "step": 11832
    },
    {
      "epoch": 0.4051009928106813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4555,
      "step": 11833
    },
    {
      "epoch": 0.4051352276617597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4453,
      "step": 11834
    },
    {
      "epoch": 0.4051694625128381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4357,
      "step": 11835
    },
    {
      "epoch": 0.4052036973639165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4432,
      "step": 11836
    },
    {
      "epoch": 0.4052379322149949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4422,
      "step": 11837
    },
    {
      "epoch": 0.4052721670660733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4575,
      "step": 11838
    },
    {
      "epoch": 0.40530640191715167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 11839
    },
    {
      "epoch": 0.40534063676823007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4351,
      "step": 11840
    },
    {
      "epoch": 0.40537487161930846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4697,
      "step": 11841
    },
    {
      "epoch": 0.40540910647038686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4508,
      "step": 11842
    },
    {
      "epoch": 0.40544334132146526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4255,
      "step": 11843
    },
    {
      "epoch": 0.40547757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 11844
    },
    {
      "epoch": 0.40551181102362205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4578,
      "step": 11845
    },
    {
      "epoch": 0.40554604587470044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 11846
    },
    {
      "epoch": 0.40558028072577884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4664,
      "step": 11847
    },
    {
      "epoch": 0.40561451557685724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4533,
      "step": 11848
    },
    {
      "epoch": 0.40564875042793563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4742,
      "step": 11849
    },
    {
      "epoch": 0.40568298527901403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4573,
      "step": 11850
    },
    {
      "epoch": 0.4057172201300924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4619,
      "step": 11851
    },
    {
      "epoch": 0.4057514549811708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4446,
      "step": 11852
    },
    {
      "epoch": 0.4057856898322492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4424,
      "step": 11853
    },
    {
      "epoch": 0.4058199246833276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4556,
      "step": 11854
    },
    {
      "epoch": 0.405854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.455,
      "step": 11855
    },
    {
      "epoch": 0.4058883943854844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.446,
      "step": 11856
    },
    {
      "epoch": 0.4059226292365628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4613,
      "step": 11857
    },
    {
      "epoch": 0.4059568640876412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4551,
      "step": 11858
    },
    {
      "epoch": 0.4059910989387196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4478,
      "step": 11859
    },
    {
      "epoch": 0.406025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 11860
    },
    {
      "epoch": 0.4060595686408764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4536,
      "step": 11861
    },
    {
      "epoch": 0.4060938034919548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4314,
      "step": 11862
    },
    {
      "epoch": 0.4061280383430332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4302,
      "step": 11863
    },
    {
      "epoch": 0.4061622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4407,
      "step": 11864
    },
    {
      "epoch": 0.40619650804519003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4579,
      "step": 11865
    },
    {
      "epoch": 0.4062307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4677,
      "step": 11866
    },
    {
      "epoch": 0.4062649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4491,
      "step": 11867
    },
    {
      "epoch": 0.4062992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4461,
      "step": 11868
    },
    {
      "epoch": 0.4063334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4566,
      "step": 11869
    },
    {
      "epoch": 0.406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4594,
      "step": 11870
    },
    {
      "epoch": 0.4064019171516604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4636,
      "step": 11871
    },
    {
      "epoch": 0.4064361520027388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4413,
      "step": 11872
    },
    {
      "epoch": 0.4064703868538172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4602,
      "step": 11873
    },
    {
      "epoch": 0.4065046217048956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4551,
      "step": 11874
    },
    {
      "epoch": 0.406538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4526,
      "step": 11875
    },
    {
      "epoch": 0.4065730914070524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4378,
      "step": 11876
    },
    {
      "epoch": 0.4066073262581308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4563,
      "step": 11877
    },
    {
      "epoch": 0.4066415611092092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4589,
      "step": 11878
    },
    {
      "epoch": 0.4066757959602876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4562,
      "step": 11879
    },
    {
      "epoch": 0.406710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.457,
      "step": 11880
    },
    {
      "epoch": 0.40674426566244437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.455,
      "step": 11881
    },
    {
      "epoch": 0.40677850051352277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4519,
      "step": 11882
    },
    {
      "epoch": 0.40681273536460116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4454,
      "step": 11883
    },
    {
      "epoch": 0.40684697021567956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.447,
      "step": 11884
    },
    {
      "epoch": 0.40688120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4203,
      "step": 11885
    },
    {
      "epoch": 0.40691543991783635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4465,
      "step": 11886
    },
    {
      "epoch": 0.40694967476891475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4651,
      "step": 11887
    },
    {
      "epoch": 0.40698390961999314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4365,
      "step": 11888
    },
    {
      "epoch": 0.40701814447107154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4349,
      "step": 11889
    },
    {
      "epoch": 0.40705237932214994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4366,
      "step": 11890
    },
    {
      "epoch": 0.40708661417322833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4592,
      "step": 11891
    },
    {
      "epoch": 0.40712084902430673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4378,
      "step": 11892
    },
    {
      "epoch": 0.4071550838753851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4423,
      "step": 11893
    },
    {
      "epoch": 0.4071893187264635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4559,
      "step": 11894
    },
    {
      "epoch": 0.4072235535775419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.446,
      "step": 11895
    },
    {
      "epoch": 0.4072577884286203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.454,
      "step": 11896
    },
    {
      "epoch": 0.4072920232796987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4636,
      "step": 11897
    },
    {
      "epoch": 0.4073262581307771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4356,
      "step": 11898
    },
    {
      "epoch": 0.4073604929818555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4647,
      "step": 11899
    },
    {
      "epoch": 0.4073947278329339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.469,
      "step": 11900
    },
    {
      "epoch": 0.4074289626840123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4487,
      "step": 11901
    },
    {
      "epoch": 0.40746319753509075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4516,
      "step": 11902
    },
    {
      "epoch": 0.40749743238616914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4387,
      "step": 11903
    },
    {
      "epoch": 0.40753166723724754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4381,
      "step": 11904
    },
    {
      "epoch": 0.40756590208832594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4271,
      "step": 11905
    },
    {
      "epoch": 0.40760013693940433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4601,
      "step": 11906
    },
    {
      "epoch": 0.40763437179048273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4428,
      "step": 11907
    },
    {
      "epoch": 0.4076686066415611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4598,
      "step": 11908
    },
    {
      "epoch": 0.4077028414926395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4485,
      "step": 11909
    },
    {
      "epoch": 0.4077370763437179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4788,
      "step": 11910
    },
    {
      "epoch": 0.4077713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4385,
      "step": 11911
    },
    {
      "epoch": 0.4078055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4416,
      "step": 11912
    },
    {
      "epoch": 0.4078397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4533,
      "step": 11913
    },
    {
      "epoch": 0.4078740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4409,
      "step": 11914
    },
    {
      "epoch": 0.4079082505991099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4562,
      "step": 11915
    },
    {
      "epoch": 0.4079424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4626,
      "step": 11916
    },
    {
      "epoch": 0.4079767203012667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4553,
      "step": 11917
    },
    {
      "epoch": 0.4080109551523451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4533,
      "step": 11918
    },
    {
      "epoch": 0.4080451900034235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.439,
      "step": 11919
    },
    {
      "epoch": 0.4080794248545019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4461,
      "step": 11920
    },
    {
      "epoch": 0.4081136597055803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4454,
      "step": 11921
    },
    {
      "epoch": 0.4081478945566587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4647,
      "step": 11922
    },
    {
      "epoch": 0.40818212940773707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 11923
    },
    {
      "epoch": 0.40821636425881547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4603,
      "step": 11924
    },
    {
      "epoch": 0.40825059910989386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4461,
      "step": 11925
    },
    {
      "epoch": 0.40828483396097226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4634,
      "step": 11926
    },
    {
      "epoch": 0.40831906881205066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4435,
      "step": 11927
    },
    {
      "epoch": 0.40835330366312905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4561,
      "step": 11928
    },
    {
      "epoch": 0.40838753851420745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4258,
      "step": 11929
    },
    {
      "epoch": 0.40842177336528585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4768,
      "step": 11930
    },
    {
      "epoch": 0.40845600821636424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4779,
      "step": 11931
    },
    {
      "epoch": 0.40849024306744264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4539,
      "step": 11932
    },
    {
      "epoch": 0.40852447791852103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.457,
      "step": 11933
    },
    {
      "epoch": 0.40855871276959943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4464,
      "step": 11934
    },
    {
      "epoch": 0.4085929476206778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4637,
      "step": 11935
    },
    {
      "epoch": 0.4086271824717562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4707,
      "step": 11936
    },
    {
      "epoch": 0.4086614173228346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4587,
      "step": 11937
    },
    {
      "epoch": 0.40869565217391307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4626,
      "step": 11938
    },
    {
      "epoch": 0.40872988702499147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 11939
    },
    {
      "epoch": 0.40876412187606986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4566,
      "step": 11940
    },
    {
      "epoch": 0.40879835672714826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.445,
      "step": 11941
    },
    {
      "epoch": 0.40883259157822666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4539,
      "step": 11942
    },
    {
      "epoch": 0.40886682642930505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4684,
      "step": 11943
    },
    {
      "epoch": 0.40890106128038345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4578,
      "step": 11944
    },
    {
      "epoch": 0.40893529613146185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4564,
      "step": 11945
    },
    {
      "epoch": 0.40896953098254024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4561,
      "step": 11946
    },
    {
      "epoch": 0.40900376583361864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4404,
      "step": 11947
    },
    {
      "epoch": 0.40903800068469703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4651,
      "step": 11948
    },
    {
      "epoch": 0.40907223553577543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4448,
      "step": 11949
    },
    {
      "epoch": 0.4091064703868538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.446,
      "step": 11950
    },
    {
      "epoch": 0.4091407052379322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4514,
      "step": 11951
    },
    {
      "epoch": 0.4091749400890106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4382,
      "step": 11952
    },
    {
      "epoch": 0.409209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4559,
      "step": 11953
    },
    {
      "epoch": 0.4092434097911674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.444,
      "step": 11954
    },
    {
      "epoch": 0.4092776446422458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4213,
      "step": 11955
    },
    {
      "epoch": 0.4093118794933242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4616,
      "step": 11956
    },
    {
      "epoch": 0.4093461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4481,
      "step": 11957
    },
    {
      "epoch": 0.409380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4705,
      "step": 11958
    },
    {
      "epoch": 0.4094145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4461,
      "step": 11959
    },
    {
      "epoch": 0.4094488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4381,
      "step": 11960
    },
    {
      "epoch": 0.4094830537487162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4644,
      "step": 11961
    },
    {
      "epoch": 0.4095172885997946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4714,
      "step": 11962
    },
    {
      "epoch": 0.409551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4569,
      "step": 11963
    },
    {
      "epoch": 0.4095857583019514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4426,
      "step": 11964
    },
    {
      "epoch": 0.40961999315302977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4578,
      "step": 11965
    },
    {
      "epoch": 0.40965422800410817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4528,
      "step": 11966
    },
    {
      "epoch": 0.40968846285518656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.441,
      "step": 11967
    },
    {
      "epoch": 0.40972269770626496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4439,
      "step": 11968
    },
    {
      "epoch": 0.40975693255734336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4646,
      "step": 11969
    },
    {
      "epoch": 0.40979116740842175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4379,
      "step": 11970
    },
    {
      "epoch": 0.40982540225950015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4709,
      "step": 11971
    },
    {
      "epoch": 0.40985963711057855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4416,
      "step": 11972
    },
    {
      "epoch": 0.40989387196165694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.443,
      "step": 11973
    },
    {
      "epoch": 0.40992810681273534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4572,
      "step": 11974
    },
    {
      "epoch": 0.4099623416638138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4312,
      "step": 11975
    },
    {
      "epoch": 0.4099965765148922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4619,
      "step": 11976
    },
    {
      "epoch": 0.4100308113659706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.432,
      "step": 11977
    },
    {
      "epoch": 0.410065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4425,
      "step": 11978
    },
    {
      "epoch": 0.4100992810681274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4553,
      "step": 11979
    },
    {
      "epoch": 0.41013351591920577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4845,
      "step": 11980
    },
    {
      "epoch": 0.41016775077028417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4488,
      "step": 11981
    },
    {
      "epoch": 0.41020198562136256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4525,
      "step": 11982
    },
    {
      "epoch": 0.41023622047244096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4313,
      "step": 11983
    },
    {
      "epoch": 0.41027045532351936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4647,
      "step": 11984
    },
    {
      "epoch": 0.41030469017459775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4464,
      "step": 11985
    },
    {
      "epoch": 0.41033892502567615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4622,
      "step": 11986
    },
    {
      "epoch": 0.41037315987675455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4337,
      "step": 11987
    },
    {
      "epoch": 0.41040739472783294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4606,
      "step": 11988
    },
    {
      "epoch": 0.41044162957891134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 11989
    },
    {
      "epoch": 0.41047586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4522,
      "step": 11990
    },
    {
      "epoch": 0.41051009928106813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4509,
      "step": 11991
    },
    {
      "epoch": 0.4105443341321465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4575,
      "step": 11992
    },
    {
      "epoch": 0.4105785689832249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4763,
      "step": 11993
    },
    {
      "epoch": 0.4106128038343033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4527,
      "step": 11994
    },
    {
      "epoch": 0.4106470386853817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4566,
      "step": 11995
    },
    {
      "epoch": 0.4106812735364601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4404,
      "step": 11996
    },
    {
      "epoch": 0.4107155083875385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4465,
      "step": 11997
    },
    {
      "epoch": 0.4107497432386169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.447,
      "step": 11998
    },
    {
      "epoch": 0.4107839780896953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4375,
      "step": 11999
    },
    {
      "epoch": 0.4108182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4437,
      "step": 12000
    },
    {
      "epoch": 0.4108524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4376,
      "step": 12001
    },
    {
      "epoch": 0.4108866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4532,
      "step": 12002
    },
    {
      "epoch": 0.4109209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4378,
      "step": 12003
    },
    {
      "epoch": 0.4109551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4548,
      "step": 12004
    },
    {
      "epoch": 0.4109893871961657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4305,
      "step": 12005
    },
    {
      "epoch": 0.4110236220472441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4398,
      "step": 12006
    },
    {
      "epoch": 0.41105785689832247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4337,
      "step": 12007
    },
    {
      "epoch": 0.41109209174940087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4375,
      "step": 12008
    },
    {
      "epoch": 0.41112632660047926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4704,
      "step": 12009
    },
    {
      "epoch": 0.41116056145155766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4484,
      "step": 12010
    },
    {
      "epoch": 0.41119479630263606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4574,
      "step": 12011
    },
    {
      "epoch": 0.4112290311537145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4727,
      "step": 12012
    },
    {
      "epoch": 0.4112632660047929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4457,
      "step": 12013
    },
    {
      "epoch": 0.4112975008558713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4419,
      "step": 12014
    },
    {
      "epoch": 0.4113317357069497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4558,
      "step": 12015
    },
    {
      "epoch": 0.4113659705580281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4295,
      "step": 12016
    },
    {
      "epoch": 0.4114002054091065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.458,
      "step": 12017
    },
    {
      "epoch": 0.4114344402601849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4395,
      "step": 12018
    },
    {
      "epoch": 0.4114686751112633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4583,
      "step": 12019
    },
    {
      "epoch": 0.4115029099623417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4405,
      "step": 12020
    },
    {
      "epoch": 0.4115371448134201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4421,
      "step": 12021
    },
    {
      "epoch": 0.41157137966449847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4437,
      "step": 12022
    },
    {
      "epoch": 0.41160561451557687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4496,
      "step": 12023
    },
    {
      "epoch": 0.41163984936665526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4484,
      "step": 12024
    },
    {
      "epoch": 0.41167408421773366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4383,
      "step": 12025
    },
    {
      "epoch": 0.41170831906881206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4407,
      "step": 12026
    },
    {
      "epoch": 0.41174255391989045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4339,
      "step": 12027
    },
    {
      "epoch": 0.41177678877096885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4345,
      "step": 12028
    },
    {
      "epoch": 0.41181102362204725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4299,
      "step": 12029
    },
    {
      "epoch": 0.41184525847312564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4567,
      "step": 12030
    },
    {
      "epoch": 0.41187949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4536,
      "step": 12031
    },
    {
      "epoch": 0.41191372817528243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4566,
      "step": 12032
    },
    {
      "epoch": 0.41194796302636083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4584,
      "step": 12033
    },
    {
      "epoch": 0.4119821978774392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4597,
      "step": 12034
    },
    {
      "epoch": 0.4120164327285176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4399,
      "step": 12035
    },
    {
      "epoch": 0.412050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4437,
      "step": 12036
    },
    {
      "epoch": 0.4120849024306744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4491,
      "step": 12037
    },
    {
      "epoch": 0.4121191372817528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4512,
      "step": 12038
    },
    {
      "epoch": 0.4121533721328312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4646,
      "step": 12039
    },
    {
      "epoch": 0.4121876069839096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4489,
      "step": 12040
    },
    {
      "epoch": 0.412221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4661,
      "step": 12041
    },
    {
      "epoch": 0.4122560766860664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4356,
      "step": 12042
    },
    {
      "epoch": 0.4122903115371448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4668,
      "step": 12043
    },
    {
      "epoch": 0.4123245463882232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4478,
      "step": 12044
    },
    {
      "epoch": 0.4123587812393016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4351,
      "step": 12045
    },
    {
      "epoch": 0.41239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4543,
      "step": 12046
    },
    {
      "epoch": 0.4124272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4452,
      "step": 12047
    },
    {
      "epoch": 0.4124614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4415,
      "step": 12048
    },
    {
      "epoch": 0.4124957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4586,
      "step": 12049
    },
    {
      "epoch": 0.4125299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4304,
      "step": 12050
    },
    {
      "epoch": 0.412564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4373,
      "step": 12051
    },
    {
      "epoch": 0.4125984251968504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.441,
      "step": 12052
    },
    {
      "epoch": 0.4126326600479288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4194,
      "step": 12053
    },
    {
      "epoch": 0.4126668948990072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4459,
      "step": 12054
    },
    {
      "epoch": 0.4127011297500856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4457,
      "step": 12055
    },
    {
      "epoch": 0.412735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4456,
      "step": 12056
    },
    {
      "epoch": 0.4127695994522424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4518,
      "step": 12057
    },
    {
      "epoch": 0.4128038343033208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4486,
      "step": 12058
    },
    {
      "epoch": 0.4128380691543992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4584,
      "step": 12059
    },
    {
      "epoch": 0.4128723040054776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4482,
      "step": 12060
    },
    {
      "epoch": 0.412906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4535,
      "step": 12061
    },
    {
      "epoch": 0.4129407737076344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4407,
      "step": 12062
    },
    {
      "epoch": 0.4129750085587128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4591,
      "step": 12063
    },
    {
      "epoch": 0.41300924340979117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4512,
      "step": 12064
    },
    {
      "epoch": 0.41304347826086957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4406,
      "step": 12065
    },
    {
      "epoch": 0.41307771311194796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4446,
      "step": 12066
    },
    {
      "epoch": 0.41311194796302636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4448,
      "step": 12067
    },
    {
      "epoch": 0.41314618281410476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.44,
      "step": 12068
    },
    {
      "epoch": 0.41318041766518315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4462,
      "step": 12069
    },
    {
      "epoch": 0.41321465251626155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4365,
      "step": 12070
    },
    {
      "epoch": 0.41324888736733995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4313,
      "step": 12071
    },
    {
      "epoch": 0.41328312221841834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4592,
      "step": 12072
    },
    {
      "epoch": 0.41331735706949674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4598,
      "step": 12073
    },
    {
      "epoch": 0.41335159192057513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4334,
      "step": 12074
    },
    {
      "epoch": 0.41338582677165353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4437,
      "step": 12075
    },
    {
      "epoch": 0.4134200616227319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4532,
      "step": 12076
    },
    {
      "epoch": 0.4134542964738103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4476,
      "step": 12077
    },
    {
      "epoch": 0.4134885313248887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4313,
      "step": 12078
    },
    {
      "epoch": 0.4135227661759671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4381,
      "step": 12079
    },
    {
      "epoch": 0.4135570010270455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4234,
      "step": 12080
    },
    {
      "epoch": 0.4135912358781239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4368,
      "step": 12081
    },
    {
      "epoch": 0.4136254707292023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4417,
      "step": 12082
    },
    {
      "epoch": 0.4136597055802807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4533,
      "step": 12083
    },
    {
      "epoch": 0.4136939404313591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4478,
      "step": 12084
    },
    {
      "epoch": 0.41372817528243755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4638,
      "step": 12085
    },
    {
      "epoch": 0.41376241013351595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.454,
      "step": 12086
    },
    {
      "epoch": 0.41379664498459434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4459,
      "step": 12087
    },
    {
      "epoch": 0.41383087983567274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4585,
      "step": 12088
    },
    {
      "epoch": 0.41386511468675113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.459,
      "step": 12089
    },
    {
      "epoch": 0.41389934953782953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4594,
      "step": 12090
    },
    {
      "epoch": 0.4139335843889079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4398,
      "step": 12091
    },
    {
      "epoch": 0.4139678192399863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4327,
      "step": 12092
    },
    {
      "epoch": 0.4140020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 12093
    },
    {
      "epoch": 0.4140362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4362,
      "step": 12094
    },
    {
      "epoch": 0.4140705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4348,
      "step": 12095
    },
    {
      "epoch": 0.4141047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4571,
      "step": 12096
    },
    {
      "epoch": 0.4141389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4429,
      "step": 12097
    },
    {
      "epoch": 0.4141732283464567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4432,
      "step": 12098
    },
    {
      "epoch": 0.4142074631975351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4226,
      "step": 12099
    },
    {
      "epoch": 0.4142416980486135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4391,
      "step": 12100
    },
    {
      "epoch": 0.4142759328996919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4489,
      "step": 12101
    },
    {
      "epoch": 0.4143101677507703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4468,
      "step": 12102
    },
    {
      "epoch": 0.4143444026018487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4434,
      "step": 12103
    },
    {
      "epoch": 0.4143786374529271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4392,
      "step": 12104
    },
    {
      "epoch": 0.4144128723040055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4562,
      "step": 12105
    },
    {
      "epoch": 0.41444710715508387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4544,
      "step": 12106
    },
    {
      "epoch": 0.41448134200616227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4541,
      "step": 12107
    },
    {
      "epoch": 0.41451557685724066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4422,
      "step": 12108
    },
    {
      "epoch": 0.41454981170831906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4681,
      "step": 12109
    },
    {
      "epoch": 0.41458404655939746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4525,
      "step": 12110
    },
    {
      "epoch": 0.41461828141047585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4186,
      "step": 12111
    },
    {
      "epoch": 0.41465251626155425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4589,
      "step": 12112
    },
    {
      "epoch": 0.41468675111263265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4394,
      "step": 12113
    },
    {
      "epoch": 0.41472098596371104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4541,
      "step": 12114
    },
    {
      "epoch": 0.41475522081478944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4396,
      "step": 12115
    },
    {
      "epoch": 0.41478945566586783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4633,
      "step": 12116
    },
    {
      "epoch": 0.41482369051694623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4631,
      "step": 12117
    },
    {
      "epoch": 0.4148579253680246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4616,
      "step": 12118
    },
    {
      "epoch": 0.414892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4446,
      "step": 12119
    },
    {
      "epoch": 0.4149263950701814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4465,
      "step": 12120
    },
    {
      "epoch": 0.4149606299212598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 12121
    },
    {
      "epoch": 0.41499486477233827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4202,
      "step": 12122
    },
    {
      "epoch": 0.41502909962341666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4385,
      "step": 12123
    },
    {
      "epoch": 0.41506333447449506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4572,
      "step": 12124
    },
    {
      "epoch": 0.41509756932557346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4286,
      "step": 12125
    },
    {
      "epoch": 0.41513180417665185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4351,
      "step": 12126
    },
    {
      "epoch": 0.41516603902773025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4583,
      "step": 12127
    },
    {
      "epoch": 0.41520027387880865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.445,
      "step": 12128
    },
    {
      "epoch": 0.41523450872988704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4473,
      "step": 12129
    },
    {
      "epoch": 0.41526874358096544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4328,
      "step": 12130
    },
    {
      "epoch": 0.41530297843204383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4386,
      "step": 12131
    },
    {
      "epoch": 0.41533721328312223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4255,
      "step": 12132
    },
    {
      "epoch": 0.4153714481342006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4497,
      "step": 12133
    },
    {
      "epoch": 0.415405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4619,
      "step": 12134
    },
    {
      "epoch": 0.4154399178363574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4569,
      "step": 12135
    },
    {
      "epoch": 0.4154741526874358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4561,
      "step": 12136
    },
    {
      "epoch": 0.4155083875385142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4493,
      "step": 12137
    },
    {
      "epoch": 0.4155426223895926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4557,
      "step": 12138
    },
    {
      "epoch": 0.415576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4346,
      "step": 12139
    },
    {
      "epoch": 0.4156110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4473,
      "step": 12140
    },
    {
      "epoch": 0.4156453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4339,
      "step": 12141
    },
    {
      "epoch": 0.4156795617939062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4294,
      "step": 12142
    },
    {
      "epoch": 0.4157137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.443,
      "step": 12143
    },
    {
      "epoch": 0.415748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4437,
      "step": 12144
    },
    {
      "epoch": 0.4157822663471414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4296,
      "step": 12145
    },
    {
      "epoch": 0.4158165011982198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4417,
      "step": 12146
    },
    {
      "epoch": 0.4158507360492982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4512,
      "step": 12147
    },
    {
      "epoch": 0.41588497090037657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4208,
      "step": 12148
    },
    {
      "epoch": 0.41591920575145497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4381,
      "step": 12149
    },
    {
      "epoch": 0.41595344060253336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4309,
      "step": 12150
    },
    {
      "epoch": 0.41598767545361176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4424,
      "step": 12151
    },
    {
      "epoch": 0.41602191030469016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4337,
      "step": 12152
    },
    {
      "epoch": 0.41605614515576855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4421,
      "step": 12153
    },
    {
      "epoch": 0.41609038000684695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4329,
      "step": 12154
    },
    {
      "epoch": 0.41612461485792535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4471,
      "step": 12155
    },
    {
      "epoch": 0.41615884970900374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4517,
      "step": 12156
    },
    {
      "epoch": 0.41619308456008214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4419,
      "step": 12157
    },
    {
      "epoch": 0.41622731941116053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4533,
      "step": 12158
    },
    {
      "epoch": 0.416261554262239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4522,
      "step": 12159
    },
    {
      "epoch": 0.4162957891133174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4535,
      "step": 12160
    },
    {
      "epoch": 0.4163300239643958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4285,
      "step": 12161
    },
    {
      "epoch": 0.4163642588154742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4645,
      "step": 12162
    },
    {
      "epoch": 0.41639849366655257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4446,
      "step": 12163
    },
    {
      "epoch": 0.41643272851763097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4504,
      "step": 12164
    },
    {
      "epoch": 0.41646696336870936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4514,
      "step": 12165
    },
    {
      "epoch": 0.41650119821978776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4307,
      "step": 12166
    },
    {
      "epoch": 0.41653543307086616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4377,
      "step": 12167
    },
    {
      "epoch": 0.41656966792194455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4276,
      "step": 12168
    },
    {
      "epoch": 0.41660390277302295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4493,
      "step": 12169
    },
    {
      "epoch": 0.41663813762410135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4395,
      "step": 12170
    },
    {
      "epoch": 0.41667237247517974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.447,
      "step": 12171
    },
    {
      "epoch": 0.41670660732625814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4376,
      "step": 12172
    },
    {
      "epoch": 0.41674084217733653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4271,
      "step": 12173
    },
    {
      "epoch": 0.41677507702841493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4354,
      "step": 12174
    },
    {
      "epoch": 0.4168093118794933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4339,
      "step": 12175
    },
    {
      "epoch": 0.4168435467305717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4345,
      "step": 12176
    },
    {
      "epoch": 0.4168777815816501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4351,
      "step": 12177
    },
    {
      "epoch": 0.4169120164327285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4504,
      "step": 12178
    },
    {
      "epoch": 0.4169462512838069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4375,
      "step": 12179
    },
    {
      "epoch": 0.4169804861348853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4391,
      "step": 12180
    },
    {
      "epoch": 0.4170147209859637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4431,
      "step": 12181
    },
    {
      "epoch": 0.4170489558370421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4355,
      "step": 12182
    },
    {
      "epoch": 0.4170831906881205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4249,
      "step": 12183
    },
    {
      "epoch": 0.4171174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4373,
      "step": 12184
    },
    {
      "epoch": 0.4171516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4144,
      "step": 12185
    },
    {
      "epoch": 0.4171858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.442,
      "step": 12186
    },
    {
      "epoch": 0.4172201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.438,
      "step": 12187
    },
    {
      "epoch": 0.4172543649435125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4396,
      "step": 12188
    },
    {
      "epoch": 0.4172885997945909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4341,
      "step": 12189
    },
    {
      "epoch": 0.41732283464566927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4481,
      "step": 12190
    },
    {
      "epoch": 0.41735706949674767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4395,
      "step": 12191
    },
    {
      "epoch": 0.41739130434782606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4486,
      "step": 12192
    },
    {
      "epoch": 0.41742553919890446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4585,
      "step": 12193
    },
    {
      "epoch": 0.41745977404998286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4626,
      "step": 12194
    },
    {
      "epoch": 0.41749400890106125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4324,
      "step": 12195
    },
    {
      "epoch": 0.4175282437521397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4246,
      "step": 12196
    },
    {
      "epoch": 0.4175624786032181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4514,
      "step": 12197
    },
    {
      "epoch": 0.4175967134542965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.419,
      "step": 12198
    },
    {
      "epoch": 0.4176309483053749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4454,
      "step": 12199
    },
    {
      "epoch": 0.4176651831564533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4175,
      "step": 12200
    },
    {
      "epoch": 0.4176994180075317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4469,
      "step": 12201
    },
    {
      "epoch": 0.4177336528586101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4438,
      "step": 12202
    },
    {
      "epoch": 0.4177678877096885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4504,
      "step": 12203
    },
    {
      "epoch": 0.4178021225607669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4491,
      "step": 12204
    },
    {
      "epoch": 0.41783635741184527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4445,
      "step": 12205
    },
    {
      "epoch": 0.41787059226292367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4616,
      "step": 12206
    },
    {
      "epoch": 0.41790482711400206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4497,
      "step": 12207
    },
    {
      "epoch": 0.41793906196508046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4458,
      "step": 12208
    },
    {
      "epoch": 0.41797329681615886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 12209
    },
    {
      "epoch": 0.41800753166723725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4439,
      "step": 12210
    },
    {
      "epoch": 0.41804176651831565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4518,
      "step": 12211
    },
    {
      "epoch": 0.41807600136939405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4318,
      "step": 12212
    },
    {
      "epoch": 0.41811023622047244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4389,
      "step": 12213
    },
    {
      "epoch": 0.41814447107155084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4368,
      "step": 12214
    },
    {
      "epoch": 0.41817870592262923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4362,
      "step": 12215
    },
    {
      "epoch": 0.41821294077370763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4484,
      "step": 12216
    },
    {
      "epoch": 0.418247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4457,
      "step": 12217
    },
    {
      "epoch": 0.4182814104758644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.439,
      "step": 12218
    },
    {
      "epoch": 0.4183156453269428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4142,
      "step": 12219
    },
    {
      "epoch": 0.4183498801780212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4424,
      "step": 12220
    },
    {
      "epoch": 0.4183841150290996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4284,
      "step": 12221
    },
    {
      "epoch": 0.418418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4639,
      "step": 12222
    },
    {
      "epoch": 0.4184525847312564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4407,
      "step": 12223
    },
    {
      "epoch": 0.4184868195823348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4403,
      "step": 12224
    },
    {
      "epoch": 0.4185210544334132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4194,
      "step": 12225
    },
    {
      "epoch": 0.4185552892844916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4566,
      "step": 12226
    },
    {
      "epoch": 0.41858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4516,
      "step": 12227
    },
    {
      "epoch": 0.4186237589866484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4298,
      "step": 12228
    },
    {
      "epoch": 0.4186579938377268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4415,
      "step": 12229
    },
    {
      "epoch": 0.4186922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4175,
      "step": 12230
    },
    {
      "epoch": 0.4187264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4415,
      "step": 12231
    },
    {
      "epoch": 0.41876069839096197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4405,
      "step": 12232
    },
    {
      "epoch": 0.4187949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4323,
      "step": 12233
    },
    {
      "epoch": 0.4188291680931188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4361,
      "step": 12234
    },
    {
      "epoch": 0.4188634029441972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4288,
      "step": 12235
    },
    {
      "epoch": 0.4188976377952756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4455,
      "step": 12236
    },
    {
      "epoch": 0.418931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4407,
      "step": 12237
    },
    {
      "epoch": 0.4189661074974324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4506,
      "step": 12238
    },
    {
      "epoch": 0.4190003423485108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4482,
      "step": 12239
    },
    {
      "epoch": 0.4190345771995892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4343,
      "step": 12240
    },
    {
      "epoch": 0.4190688120506676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4473,
      "step": 12241
    },
    {
      "epoch": 0.419103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4172,
      "step": 12242
    },
    {
      "epoch": 0.4191372817528244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4279,
      "step": 12243
    },
    {
      "epoch": 0.4191715166039028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4513,
      "step": 12244
    },
    {
      "epoch": 0.4192057514549812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4536,
      "step": 12245
    },
    {
      "epoch": 0.4192399863060596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4696,
      "step": 12246
    },
    {
      "epoch": 0.41927422115713797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4419,
      "step": 12247
    },
    {
      "epoch": 0.41930845600821637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4456,
      "step": 12248
    },
    {
      "epoch": 0.41934269085929476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4258,
      "step": 12249
    },
    {
      "epoch": 0.41937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4434,
      "step": 12250
    },
    {
      "epoch": 0.41941116056145156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.446,
      "step": 12251
    },
    {
      "epoch": 0.41944539541252995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.427,
      "step": 12252
    },
    {
      "epoch": 0.41947963026360835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4494,
      "step": 12253
    },
    {
      "epoch": 0.41951386511468675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4426,
      "step": 12254
    },
    {
      "epoch": 0.41954809996576514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4472,
      "step": 12255
    },
    {
      "epoch": 0.41958233481684354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4456,
      "step": 12256
    },
    {
      "epoch": 0.41961656966792193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4191,
      "step": 12257
    },
    {
      "epoch": 0.41965080451900033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4429,
      "step": 12258
    },
    {
      "epoch": 0.4196850393700787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4467,
      "step": 12259
    },
    {
      "epoch": 0.4197192742211571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4522,
      "step": 12260
    },
    {
      "epoch": 0.4197535090722355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4504,
      "step": 12261
    },
    {
      "epoch": 0.4197877439233139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4549,
      "step": 12262
    },
    {
      "epoch": 0.4198219787743923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4486,
      "step": 12263
    },
    {
      "epoch": 0.4198562136254707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4357,
      "step": 12264
    },
    {
      "epoch": 0.4198904484765491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4413,
      "step": 12265
    },
    {
      "epoch": 0.4199246833276275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4288,
      "step": 12266
    },
    {
      "epoch": 0.4199589181787059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4278,
      "step": 12267
    },
    {
      "epoch": 0.4199931530297843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4501,
      "step": 12268
    },
    {
      "epoch": 0.42002738788086275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4478,
      "step": 12269
    },
    {
      "epoch": 0.42006162273194114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4278,
      "step": 12270
    },
    {
      "epoch": 0.42009585758301954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4508,
      "step": 12271
    },
    {
      "epoch": 0.42013009243409793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 12272
    },
    {
      "epoch": 0.42016432728517633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4437,
      "step": 12273
    },
    {
      "epoch": 0.4201985621362547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.434,
      "step": 12274
    },
    {
      "epoch": 0.4202327969873331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4446,
      "step": 12275
    },
    {
      "epoch": 0.4202670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4188,
      "step": 12276
    },
    {
      "epoch": 0.4203012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4224,
      "step": 12277
    },
    {
      "epoch": 0.4203355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4408,
      "step": 12278
    },
    {
      "epoch": 0.4203697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4498,
      "step": 12279
    },
    {
      "epoch": 0.4204039712427251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4395,
      "step": 12280
    },
    {
      "epoch": 0.4204382060938035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4399,
      "step": 12281
    },
    {
      "epoch": 0.4204724409448819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4324,
      "step": 12282
    },
    {
      "epoch": 0.4205066757959603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4438,
      "step": 12283
    },
    {
      "epoch": 0.4205409106470387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4468,
      "step": 12284
    },
    {
      "epoch": 0.4205751454981171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4578,
      "step": 12285
    },
    {
      "epoch": 0.4206093803491955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4367,
      "step": 12286
    },
    {
      "epoch": 0.4206436152002739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4507,
      "step": 12287
    },
    {
      "epoch": 0.4206778500513523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4475,
      "step": 12288
    },
    {
      "epoch": 0.42071208490243067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4416,
      "step": 12289
    },
    {
      "epoch": 0.42074631975350907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4124,
      "step": 12290
    },
    {
      "epoch": 0.42078055460458746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.45,
      "step": 12291
    },
    {
      "epoch": 0.42081478945566586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4304,
      "step": 12292
    },
    {
      "epoch": 0.42084902430674426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4161,
      "step": 12293
    },
    {
      "epoch": 0.42088325915782265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.437,
      "step": 12294
    },
    {
      "epoch": 0.42091749400890105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4196,
      "step": 12295
    },
    {
      "epoch": 0.42095172885997945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4237,
      "step": 12296
    },
    {
      "epoch": 0.42098596371105784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4533,
      "step": 12297
    },
    {
      "epoch": 0.42102019856213624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4592,
      "step": 12298
    },
    {
      "epoch": 0.42105443341321463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4574,
      "step": 12299
    },
    {
      "epoch": 0.42108866826429303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4601,
      "step": 12300
    },
    {
      "epoch": 0.4211229031153714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4291,
      "step": 12301
    },
    {
      "epoch": 0.4211571379664498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4357,
      "step": 12302
    },
    {
      "epoch": 0.4211913728175282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4377,
      "step": 12303
    },
    {
      "epoch": 0.4212256076686066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4291,
      "step": 12304
    },
    {
      "epoch": 0.421259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4458,
      "step": 12305
    },
    {
      "epoch": 0.42129407737076346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4548,
      "step": 12306
    },
    {
      "epoch": 0.42132831222184186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4451,
      "step": 12307
    },
    {
      "epoch": 0.42136254707292026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4228,
      "step": 12308
    },
    {
      "epoch": 0.42139678192399865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4384,
      "step": 12309
    },
    {
      "epoch": 0.42143101677507705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4589,
      "step": 12310
    },
    {
      "epoch": 0.42146525162615545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4381,
      "step": 12311
    },
    {
      "epoch": 0.42149948647723384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4325,
      "step": 12312
    },
    {
      "epoch": 0.42153372132831224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.445,
      "step": 12313
    },
    {
      "epoch": 0.42156795617939063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.442,
      "step": 12314
    },
    {
      "epoch": 0.42160219103046903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4442,
      "step": 12315
    },
    {
      "epoch": 0.4216364258815474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4406,
      "step": 12316
    },
    {
      "epoch": 0.4216706607326258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4383,
      "step": 12317
    },
    {
      "epoch": 0.4217048955837042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.432,
      "step": 12318
    },
    {
      "epoch": 0.4217391304347826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4616,
      "step": 12319
    },
    {
      "epoch": 0.421773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4478,
      "step": 12320
    },
    {
      "epoch": 0.4218076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.439,
      "step": 12321
    },
    {
      "epoch": 0.4218418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4614,
      "step": 12322
    },
    {
      "epoch": 0.4218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.448,
      "step": 12323
    },
    {
      "epoch": 0.4219103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4582,
      "step": 12324
    },
    {
      "epoch": 0.421944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4387,
      "step": 12325
    },
    {
      "epoch": 0.4219787743923314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4412,
      "step": 12326
    },
    {
      "epoch": 0.4220130092434098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.459,
      "step": 12327
    },
    {
      "epoch": 0.4220472440944882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4203,
      "step": 12328
    },
    {
      "epoch": 0.4220814789455666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4467,
      "step": 12329
    },
    {
      "epoch": 0.422115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 12330
    },
    {
      "epoch": 0.42214994864772337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4417,
      "step": 12331
    },
    {
      "epoch": 0.42218418349880177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4682,
      "step": 12332
    },
    {
      "epoch": 0.42221841834988016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4482,
      "step": 12333
    },
    {
      "epoch": 0.42225265320095856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4468,
      "step": 12334
    },
    {
      "epoch": 0.42228688805203696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4147,
      "step": 12335
    },
    {
      "epoch": 0.42232112290311535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4445,
      "step": 12336
    },
    {
      "epoch": 0.42235535775419375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4459,
      "step": 12337
    },
    {
      "epoch": 0.42238959260527215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4211,
      "step": 12338
    },
    {
      "epoch": 0.42242382745635054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4357,
      "step": 12339
    },
    {
      "epoch": 0.42245806230742894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4545,
      "step": 12340
    },
    {
      "epoch": 0.42249229715850734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4465,
      "step": 12341
    },
    {
      "epoch": 0.42252653200958573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.434,
      "step": 12342
    },
    {
      "epoch": 0.4225607668606642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4328,
      "step": 12343
    },
    {
      "epoch": 0.4225950017117426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4307,
      "step": 12344
    },
    {
      "epoch": 0.422629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4523,
      "step": 12345
    },
    {
      "epoch": 0.42266347141389937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4352,
      "step": 12346
    },
    {
      "epoch": 0.42269770626497777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4386,
      "step": 12347
    },
    {
      "epoch": 0.42273194111605616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4365,
      "step": 12348
    },
    {
      "epoch": 0.42276617596713456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.444,
      "step": 12349
    },
    {
      "epoch": 0.42280041081821296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4348,
      "step": 12350
    },
    {
      "epoch": 0.42283464566929135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4446,
      "step": 12351
    },
    {
      "epoch": 0.42286888052036975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4484,
      "step": 12352
    },
    {
      "epoch": 0.42290311537144815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4417,
      "step": 12353
    },
    {
      "epoch": 0.42293735022252654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4175,
      "step": 12354
    },
    {
      "epoch": 0.42297158507360494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4385,
      "step": 12355
    },
    {
      "epoch": 0.42300581992468334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4169,
      "step": 12356
    },
    {
      "epoch": 0.42304005477576173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4382,
      "step": 12357
    },
    {
      "epoch": 0.4230742896268401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4468,
      "step": 12358
    },
    {
      "epoch": 0.4231085244779185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4486,
      "step": 12359
    },
    {
      "epoch": 0.4231427593289969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4268,
      "step": 12360
    },
    {
      "epoch": 0.4231769941800753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.443,
      "step": 12361
    },
    {
      "epoch": 0.4232112290311537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4347,
      "step": 12362
    },
    {
      "epoch": 0.4232454638822321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4304,
      "step": 12363
    },
    {
      "epoch": 0.4232796987333105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4358,
      "step": 12364
    },
    {
      "epoch": 0.4233139335843889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4567,
      "step": 12365
    },
    {
      "epoch": 0.4233481684354673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.42,
      "step": 12366
    },
    {
      "epoch": 0.4233824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4324,
      "step": 12367
    },
    {
      "epoch": 0.4234166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.45,
      "step": 12368
    },
    {
      "epoch": 0.4234508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4346,
      "step": 12369
    },
    {
      "epoch": 0.4234851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4213,
      "step": 12370
    },
    {
      "epoch": 0.4235193426908593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4279,
      "step": 12371
    },
    {
      "epoch": 0.4235535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 12372
    },
    {
      "epoch": 0.42358781239301607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4315,
      "step": 12373
    },
    {
      "epoch": 0.42362204724409447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4432,
      "step": 12374
    },
    {
      "epoch": 0.42365628209517286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4364,
      "step": 12375
    },
    {
      "epoch": 0.42369051694625126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4288,
      "step": 12376
    },
    {
      "epoch": 0.42372475179732966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4443,
      "step": 12377
    },
    {
      "epoch": 0.42375898664840805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4259,
      "step": 12378
    },
    {
      "epoch": 0.42379322149948645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4722,
      "step": 12379
    },
    {
      "epoch": 0.4238274563505649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.441,
      "step": 12380
    },
    {
      "epoch": 0.4238616912016433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4369,
      "step": 12381
    },
    {
      "epoch": 0.4238959260527217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4507,
      "step": 12382
    },
    {
      "epoch": 0.4239301609038001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4341,
      "step": 12383
    },
    {
      "epoch": 0.4239643957548785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4268,
      "step": 12384
    },
    {
      "epoch": 0.4239986306059569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4091,
      "step": 12385
    },
    {
      "epoch": 0.4240328654570353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4436,
      "step": 12386
    },
    {
      "epoch": 0.4240671003081137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4404,
      "step": 12387
    },
    {
      "epoch": 0.42410133515919207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.413,
      "step": 12388
    },
    {
      "epoch": 0.42413557001027047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4216,
      "step": 12389
    },
    {
      "epoch": 0.42416980486134886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.437,
      "step": 12390
    },
    {
      "epoch": 0.42420403971242726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4435,
      "step": 12391
    },
    {
      "epoch": 0.42423827456350566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4329,
      "step": 12392
    },
    {
      "epoch": 0.42427250941458405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4589,
      "step": 12393
    },
    {
      "epoch": 0.42430674426566245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4337,
      "step": 12394
    },
    {
      "epoch": 0.42434097911674085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4296,
      "step": 12395
    },
    {
      "epoch": 0.42437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4257,
      "step": 12396
    },
    {
      "epoch": 0.42440944881889764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4458,
      "step": 12397
    },
    {
      "epoch": 0.42444368366997604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4266,
      "step": 12398
    },
    {
      "epoch": 0.42447791852105443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4475,
      "step": 12399
    },
    {
      "epoch": 0.4245121533721328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4531,
      "step": 12400
    },
    {
      "epoch": 0.4245463882232112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4255,
      "step": 12401
    },
    {
      "epoch": 0.4245806230742896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4373,
      "step": 12402
    },
    {
      "epoch": 0.424614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4476,
      "step": 12403
    },
    {
      "epoch": 0.4246490927764464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.437,
      "step": 12404
    },
    {
      "epoch": 0.4246833276275248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4396,
      "step": 12405
    },
    {
      "epoch": 0.4247175624786032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4448,
      "step": 12406
    },
    {
      "epoch": 0.4247517973296816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4371,
      "step": 12407
    },
    {
      "epoch": 0.42478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4362,
      "step": 12408
    },
    {
      "epoch": 0.4248202670318384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4371,
      "step": 12409
    },
    {
      "epoch": 0.4248545018829168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4547,
      "step": 12410
    },
    {
      "epoch": 0.4248887367339952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4513,
      "step": 12411
    },
    {
      "epoch": 0.4249229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4326,
      "step": 12412
    },
    {
      "epoch": 0.424957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4229,
      "step": 12413
    },
    {
      "epoch": 0.4249914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4229,
      "step": 12414
    },
    {
      "epoch": 0.4250256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.439,
      "step": 12415
    },
    {
      "epoch": 0.42505991098938717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4423,
      "step": 12416
    },
    {
      "epoch": 0.4250941458404656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4243,
      "step": 12417
    },
    {
      "epoch": 0.425128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4238,
      "step": 12418
    },
    {
      "epoch": 0.4251626155426224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4375,
      "step": 12419
    },
    {
      "epoch": 0.4251968503937008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4342,
      "step": 12420
    },
    {
      "epoch": 0.4252310852447792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4513,
      "step": 12421
    },
    {
      "epoch": 0.4252653200958576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4335,
      "step": 12422
    },
    {
      "epoch": 0.425299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4406,
      "step": 12423
    },
    {
      "epoch": 0.4253337897980144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4478,
      "step": 12424
    },
    {
      "epoch": 0.4253680246490928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4385,
      "step": 12425
    },
    {
      "epoch": 0.4254022595001712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4478,
      "step": 12426
    },
    {
      "epoch": 0.4254364943512496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4484,
      "step": 12427
    },
    {
      "epoch": 0.425470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.431,
      "step": 12428
    },
    {
      "epoch": 0.4255049640534064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4601,
      "step": 12429
    },
    {
      "epoch": 0.4255391989044848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4512,
      "step": 12430
    },
    {
      "epoch": 0.42557343375556317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4598,
      "step": 12431
    },
    {
      "epoch": 0.42560766860664156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4423,
      "step": 12432
    },
    {
      "epoch": 0.42564190345771996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4389,
      "step": 12433
    },
    {
      "epoch": 0.42567613830879836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4605,
      "step": 12434
    },
    {
      "epoch": 0.42571037315987675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4377,
      "step": 12435
    },
    {
      "epoch": 0.42574460801095515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4431,
      "step": 12436
    },
    {
      "epoch": 0.42577884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4348,
      "step": 12437
    },
    {
      "epoch": 0.42581307771311194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.437,
      "step": 12438
    },
    {
      "epoch": 0.42584731256419034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4749,
      "step": 12439
    },
    {
      "epoch": 0.42588154741526874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4305,
      "step": 12440
    },
    {
      "epoch": 0.42591578226634713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4214,
      "step": 12441
    },
    {
      "epoch": 0.42595001711742553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4529,
      "step": 12442
    },
    {
      "epoch": 0.4259842519685039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4363,
      "step": 12443
    },
    {
      "epoch": 0.4260184868195823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4397,
      "step": 12444
    },
    {
      "epoch": 0.4260527216706607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4518,
      "step": 12445
    },
    {
      "epoch": 0.4260869565217391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4423,
      "step": 12446
    },
    {
      "epoch": 0.4261211913728175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4412,
      "step": 12447
    },
    {
      "epoch": 0.4261554262238959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4409,
      "step": 12448
    },
    {
      "epoch": 0.4261896610749743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4346,
      "step": 12449
    },
    {
      "epoch": 0.4262238959260527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4424,
      "step": 12450
    },
    {
      "epoch": 0.4262581307771311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4433,
      "step": 12451
    },
    {
      "epoch": 0.4262923656282095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4436,
      "step": 12452
    },
    {
      "epoch": 0.42632660047928794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4235,
      "step": 12453
    },
    {
      "epoch": 0.42636083533036634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4289,
      "step": 12454
    },
    {
      "epoch": 0.42639507018144474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4257,
      "step": 12455
    },
    {
      "epoch": 0.42642930503252313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4338,
      "step": 12456
    },
    {
      "epoch": 0.42646353988360153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4388,
      "step": 12457
    },
    {
      "epoch": 0.4264977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4287,
      "step": 12458
    },
    {
      "epoch": 0.4265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4315,
      "step": 12459
    },
    {
      "epoch": 0.4265662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4334,
      "step": 12460
    },
    {
      "epoch": 0.4266004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4255,
      "step": 12461
    },
    {
      "epoch": 0.4266347141389935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4135,
      "step": 12462
    },
    {
      "epoch": 0.4266689489900719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4345,
      "step": 12463
    },
    {
      "epoch": 0.4267031838411503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4453,
      "step": 12464
    },
    {
      "epoch": 0.4267374186922287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4371,
      "step": 12465
    },
    {
      "epoch": 0.4267716535433071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4182,
      "step": 12466
    },
    {
      "epoch": 0.4268058883943855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4194,
      "step": 12467
    },
    {
      "epoch": 0.4268401232454639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4312,
      "step": 12468
    },
    {
      "epoch": 0.4268743580965423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4087,
      "step": 12469
    },
    {
      "epoch": 0.4269085929476207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4146,
      "step": 12470
    },
    {
      "epoch": 0.4269428277986991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4298,
      "step": 12471
    },
    {
      "epoch": 0.4269770626497775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4484,
      "step": 12472
    },
    {
      "epoch": 0.42701129750085587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4456,
      "step": 12473
    },
    {
      "epoch": 0.42704553235193426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4396,
      "step": 12474
    },
    {
      "epoch": 0.42707976720301266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.419,
      "step": 12475
    },
    {
      "epoch": 0.42711400205409106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4334,
      "step": 12476
    },
    {
      "epoch": 0.42714823690516945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4294,
      "step": 12477
    },
    {
      "epoch": 0.42718247175624785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.428,
      "step": 12478
    },
    {
      "epoch": 0.42721670660732625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4613,
      "step": 12479
    },
    {
      "epoch": 0.42725094145840464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4433,
      "step": 12480
    },
    {
      "epoch": 0.42728517630948304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4234,
      "step": 12481
    },
    {
      "epoch": 0.42731941116056144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4241,
      "step": 12482
    },
    {
      "epoch": 0.42735364601163983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4312,
      "step": 12483
    },
    {
      "epoch": 0.42738788086271823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4387,
      "step": 12484
    },
    {
      "epoch": 0.4274221157137966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4238,
      "step": 12485
    },
    {
      "epoch": 0.427456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.429,
      "step": 12486
    },
    {
      "epoch": 0.4274905854159534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4376,
      "step": 12487
    },
    {
      "epoch": 0.4275248202670318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4266,
      "step": 12488
    },
    {
      "epoch": 0.4275590551181102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4303,
      "step": 12489
    },
    {
      "epoch": 0.42759328996918866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4251,
      "step": 12490
    },
    {
      "epoch": 0.42762752482026706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4462,
      "step": 12491
    },
    {
      "epoch": 0.42766175967134545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4468,
      "step": 12492
    },
    {
      "epoch": 0.42769599452242385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4462,
      "step": 12493
    },
    {
      "epoch": 0.42773022937350225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4312,
      "step": 12494
    },
    {
      "epoch": 0.42776446422458064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4249,
      "step": 12495
    },
    {
      "epoch": 0.42779869907565904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4342,
      "step": 12496
    },
    {
      "epoch": 0.42783293392673744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4407,
      "step": 12497
    },
    {
      "epoch": 0.42786716877781583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4333,
      "step": 12498
    },
    {
      "epoch": 0.42790140362889423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4351,
      "step": 12499
    },
    {
      "epoch": 0.4279356384799726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4503,
      "step": 12500
    },
    {
      "epoch": 0.427969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.436,
      "step": 12501
    },
    {
      "epoch": 0.4280041081821294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4318,
      "step": 12502
    },
    {
      "epoch": 0.4280383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.423,
      "step": 12503
    },
    {
      "epoch": 0.4280725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4591,
      "step": 12504
    },
    {
      "epoch": 0.4281068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4266,
      "step": 12505
    },
    {
      "epoch": 0.428141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4088,
      "step": 12506
    },
    {
      "epoch": 0.4281752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4287,
      "step": 12507
    },
    {
      "epoch": 0.4282095172885998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.423,
      "step": 12508
    },
    {
      "epoch": 0.4282437521396782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4208,
      "step": 12509
    },
    {
      "epoch": 0.4282779869907566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4147,
      "step": 12510
    },
    {
      "epoch": 0.428312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4258,
      "step": 12511
    },
    {
      "epoch": 0.4283464566929134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4276,
      "step": 12512
    },
    {
      "epoch": 0.4283806915439918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4327,
      "step": 12513
    },
    {
      "epoch": 0.4284149263950702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4385,
      "step": 12514
    },
    {
      "epoch": 0.42844916124614857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4456,
      "step": 12515
    },
    {
      "epoch": 0.42848339609722697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4505,
      "step": 12516
    },
    {
      "epoch": 0.42851763094830536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4266,
      "step": 12517
    },
    {
      "epoch": 0.42855186579938376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4408,
      "step": 12518
    },
    {
      "epoch": 0.42858610065046215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4254,
      "step": 12519
    },
    {
      "epoch": 0.42862033550154055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4433,
      "step": 12520
    },
    {
      "epoch": 0.42865457035261895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4346,
      "step": 12521
    },
    {
      "epoch": 0.42868880520369734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.44,
      "step": 12522
    },
    {
      "epoch": 0.42872304005477574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4204,
      "step": 12523
    },
    {
      "epoch": 0.42875727490585414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.432,
      "step": 12524
    },
    {
      "epoch": 0.42879150975693253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4397,
      "step": 12525
    },
    {
      "epoch": 0.42882574460801093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4655,
      "step": 12526
    },
    {
      "epoch": 0.4288599794590894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4255,
      "step": 12527
    },
    {
      "epoch": 0.4288942143101678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4281,
      "step": 12528
    },
    {
      "epoch": 0.4289284491612462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4263,
      "step": 12529
    },
    {
      "epoch": 0.42896268401232457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.441,
      "step": 12530
    },
    {
      "epoch": 0.42899691886340297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4529,
      "step": 12531
    },
    {
      "epoch": 0.42903115371448136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4249,
      "step": 12532
    },
    {
      "epoch": 0.42906538856555976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4449,
      "step": 12533
    },
    {
      "epoch": 0.42909962341663815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.413,
      "step": 12534
    },
    {
      "epoch": 0.42913385826771655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4211,
      "step": 12535
    },
    {
      "epoch": 0.42916809311879495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.439,
      "step": 12536
    },
    {
      "epoch": 0.42920232796987334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4269,
      "step": 12537
    },
    {
      "epoch": 0.42923656282095174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.416,
      "step": 12538
    },
    {
      "epoch": 0.42927079767203014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4318,
      "step": 12539
    },
    {
      "epoch": 0.42930503252310853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4341,
      "step": 12540
    },
    {
      "epoch": 0.42933926737418693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4267,
      "step": 12541
    },
    {
      "epoch": 0.4293735022252653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4377,
      "step": 12542
    },
    {
      "epoch": 0.4294077370763437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4307,
      "step": 12543
    },
    {
      "epoch": 0.4294419719274221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4168,
      "step": 12544
    },
    {
      "epoch": 0.4294762067785005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4173,
      "step": 12545
    },
    {
      "epoch": 0.4295104416295789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4356,
      "step": 12546
    },
    {
      "epoch": 0.4295446764806573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4146,
      "step": 12547
    },
    {
      "epoch": 0.4295789113317357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4097,
      "step": 12548
    },
    {
      "epoch": 0.4296131461828141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4277,
      "step": 12549
    },
    {
      "epoch": 0.4296473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4459,
      "step": 12550
    },
    {
      "epoch": 0.4296816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.458,
      "step": 12551
    },
    {
      "epoch": 0.4297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.426,
      "step": 12552
    },
    {
      "epoch": 0.4297500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4234,
      "step": 12553
    },
    {
      "epoch": 0.4297843204382061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4533,
      "step": 12554
    },
    {
      "epoch": 0.4298185552892845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4207,
      "step": 12555
    },
    {
      "epoch": 0.4298527901403629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4463,
      "step": 12556
    },
    {
      "epoch": 0.42988702499144127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4254,
      "step": 12557
    },
    {
      "epoch": 0.42992125984251967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.427,
      "step": 12558
    },
    {
      "epoch": 0.42995549469359806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4166,
      "step": 12559
    },
    {
      "epoch": 0.42998972954467646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.435,
      "step": 12560
    },
    {
      "epoch": 0.43002396439575485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4339,
      "step": 12561
    },
    {
      "epoch": 0.43005819924683325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.435,
      "step": 12562
    },
    {
      "epoch": 0.43009243409791165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.458,
      "step": 12563
    },
    {
      "epoch": 0.4301266689489901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.435,
      "step": 12564
    },
    {
      "epoch": 0.4301609038000685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4411,
      "step": 12565
    },
    {
      "epoch": 0.4301951386511469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4276,
      "step": 12566
    },
    {
      "epoch": 0.4302293735022253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4422,
      "step": 12567
    },
    {
      "epoch": 0.4302636083533037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4354,
      "step": 12568
    },
    {
      "epoch": 0.4302978432043821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4428,
      "step": 12569
    },
    {
      "epoch": 0.4303320780554605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4316,
      "step": 12570
    },
    {
      "epoch": 0.4303663129065389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4532,
      "step": 12571
    },
    {
      "epoch": 0.43040054775761727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4367,
      "step": 12572
    },
    {
      "epoch": 0.43043478260869567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.429,
      "step": 12573
    },
    {
      "epoch": 0.43046901745977406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4413,
      "step": 12574
    },
    {
      "epoch": 0.43050325231085246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4296,
      "step": 12575
    },
    {
      "epoch": 0.43053748716193085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4528,
      "step": 12576
    },
    {
      "epoch": 0.43057172201300925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.426,
      "step": 12577
    },
    {
      "epoch": 0.43060595686408765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4356,
      "step": 12578
    },
    {
      "epoch": 0.43064019171516604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4,
      "step": 12579
    },
    {
      "epoch": 0.43067442656624444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4211,
      "step": 12580
    },
    {
      "epoch": 0.43070866141732284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.417,
      "step": 12581
    },
    {
      "epoch": 0.43074289626840123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4252,
      "step": 12582
    },
    {
      "epoch": 0.43077713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4336,
      "step": 12583
    },
    {
      "epoch": 0.430811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4179,
      "step": 12584
    },
    {
      "epoch": 0.4308456008216364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.434,
      "step": 12585
    },
    {
      "epoch": 0.4308798356727148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4086,
      "step": 12586
    },
    {
      "epoch": 0.4309140705237932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4223,
      "step": 12587
    },
    {
      "epoch": 0.4309483053748716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4436,
      "step": 12588
    },
    {
      "epoch": 0.43098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4389,
      "step": 12589
    },
    {
      "epoch": 0.4310167750770284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.429,
      "step": 12590
    },
    {
      "epoch": 0.4310510099281068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4316,
      "step": 12591
    },
    {
      "epoch": 0.4310852447791852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.415,
      "step": 12592
    },
    {
      "epoch": 0.4311194796302636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4192,
      "step": 12593
    },
    {
      "epoch": 0.431153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4154,
      "step": 12594
    },
    {
      "epoch": 0.4311879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4428,
      "step": 12595
    },
    {
      "epoch": 0.4312221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4373,
      "step": 12596
    },
    {
      "epoch": 0.4312564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4202,
      "step": 12597
    },
    {
      "epoch": 0.4312906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4413,
      "step": 12598
    },
    {
      "epoch": 0.43132488873673397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4453,
      "step": 12599
    },
    {
      "epoch": 0.43135912358781237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4193,
      "step": 12600
    },
    {
      "epoch": 0.4313933584388908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4162,
      "step": 12601
    },
    {
      "epoch": 0.4314275932899692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4241,
      "step": 12602
    },
    {
      "epoch": 0.4314618281410476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4273,
      "step": 12603
    },
    {
      "epoch": 0.431496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4442,
      "step": 12604
    },
    {
      "epoch": 0.4315302978432044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4139,
      "step": 12605
    },
    {
      "epoch": 0.4315645326942828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4179,
      "step": 12606
    },
    {
      "epoch": 0.4315987675453612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4258,
      "step": 12607
    },
    {
      "epoch": 0.4316330023964396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4294,
      "step": 12608
    },
    {
      "epoch": 0.431667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4149,
      "step": 12609
    },
    {
      "epoch": 0.4317014720985964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4302,
      "step": 12610
    },
    {
      "epoch": 0.4317357069496748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4426,
      "step": 12611
    },
    {
      "epoch": 0.4317699418007532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4189,
      "step": 12612
    },
    {
      "epoch": 0.4318041766518316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4361,
      "step": 12613
    },
    {
      "epoch": 0.43183841150290997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4457,
      "step": 12614
    },
    {
      "epoch": 0.43187264635398837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4269,
      "step": 12615
    },
    {
      "epoch": 0.43190688120506676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.437,
      "step": 12616
    },
    {
      "epoch": 0.43194111605614516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4219,
      "step": 12617
    },
    {
      "epoch": 0.43197535090722355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4463,
      "step": 12618
    },
    {
      "epoch": 0.43200958575830195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4349,
      "step": 12619
    },
    {
      "epoch": 0.43204382060938035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4226,
      "step": 12620
    },
    {
      "epoch": 0.43207805546045874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4385,
      "step": 12621
    },
    {
      "epoch": 0.43211229031153714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4227,
      "step": 12622
    },
    {
      "epoch": 0.43214652516261554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4277,
      "step": 12623
    },
    {
      "epoch": 0.43218076001369393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4158,
      "step": 12624
    },
    {
      "epoch": 0.43221499486477233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4282,
      "step": 12625
    },
    {
      "epoch": 0.4322492297158507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.431,
      "step": 12626
    },
    {
      "epoch": 0.4322834645669291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4394,
      "step": 12627
    },
    {
      "epoch": 0.4323176994180075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4276,
      "step": 12628
    },
    {
      "epoch": 0.4323519342690859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4194,
      "step": 12629
    },
    {
      "epoch": 0.4323861691201643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4144,
      "step": 12630
    },
    {
      "epoch": 0.4324204039712427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.427,
      "step": 12631
    },
    {
      "epoch": 0.4324546388223211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4362,
      "step": 12632
    },
    {
      "epoch": 0.4324888736733995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4176,
      "step": 12633
    },
    {
      "epoch": 0.4325231085244779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.428,
      "step": 12634
    },
    {
      "epoch": 0.4325573433755563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4199,
      "step": 12635
    },
    {
      "epoch": 0.4325915782266347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4272,
      "step": 12636
    },
    {
      "epoch": 0.43262581307771314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4266,
      "step": 12637
    },
    {
      "epoch": 0.43266004792879154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4238,
      "step": 12638
    },
    {
      "epoch": 0.43269428277986993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4269,
      "step": 12639
    },
    {
      "epoch": 0.43272851763094833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4128,
      "step": 12640
    },
    {
      "epoch": 0.4327627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4302,
      "step": 12641
    },
    {
      "epoch": 0.4327969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4334,
      "step": 12642
    },
    {
      "epoch": 0.4328312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.414,
      "step": 12643
    },
    {
      "epoch": 0.4328654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4351,
      "step": 12644
    },
    {
      "epoch": 0.4328996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4282,
      "step": 12645
    },
    {
      "epoch": 0.4329339267374187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.42,
      "step": 12646
    },
    {
      "epoch": 0.4329681615884971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4224,
      "step": 12647
    },
    {
      "epoch": 0.4330023964395755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.429,
      "step": 12648
    },
    {
      "epoch": 0.4330366312906539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4271,
      "step": 12649
    },
    {
      "epoch": 0.4330708661417323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4261,
      "step": 12650
    },
    {
      "epoch": 0.4331051009928107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4173,
      "step": 12651
    },
    {
      "epoch": 0.4331393358438891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.415,
      "step": 12652
    },
    {
      "epoch": 0.4331735706949675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.422,
      "step": 12653
    },
    {
      "epoch": 0.4332078055460459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4511,
      "step": 12654
    },
    {
      "epoch": 0.4332420403971243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.43,
      "step": 12655
    },
    {
      "epoch": 0.43327627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4482,
      "step": 12656
    },
    {
      "epoch": 0.43331051009928107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 12657
    },
    {
      "epoch": 0.43334474495035946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4507,
      "step": 12658
    },
    {
      "epoch": 0.43337897980143786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4285,
      "step": 12659
    },
    {
      "epoch": 0.43341321465251625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4236,
      "step": 12660
    },
    {
      "epoch": 0.43344744950359465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4232,
      "step": 12661
    },
    {
      "epoch": 0.43348168435467305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4313,
      "step": 12662
    },
    {
      "epoch": 0.43351591920575144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4268,
      "step": 12663
    },
    {
      "epoch": 0.43355015405682984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4265,
      "step": 12664
    },
    {
      "epoch": 0.43358438890790824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4761,
      "step": 12665
    },
    {
      "epoch": 0.43361862375898663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4304,
      "step": 12666
    },
    {
      "epoch": 0.43365285861006503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4203,
      "step": 12667
    },
    {
      "epoch": 0.4336870934611434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4368,
      "step": 12668
    },
    {
      "epoch": 0.4337213283122218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4226,
      "step": 12669
    },
    {
      "epoch": 0.4337555631633002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4318,
      "step": 12670
    },
    {
      "epoch": 0.4337897980143786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4451,
      "step": 12671
    },
    {
      "epoch": 0.433824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4359,
      "step": 12672
    },
    {
      "epoch": 0.4338582677165354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4254,
      "step": 12673
    },
    {
      "epoch": 0.43389250256761386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4246,
      "step": 12674
    },
    {
      "epoch": 0.43392673741869225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4205,
      "step": 12675
    },
    {
      "epoch": 0.43396097226977065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4105,
      "step": 12676
    },
    {
      "epoch": 0.43399520712084905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4338,
      "step": 12677
    },
    {
      "epoch": 0.43402944197192744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4394,
      "step": 12678
    },
    {
      "epoch": 0.43406367682300584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4136,
      "step": 12679
    },
    {
      "epoch": 0.43409791167408424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.417,
      "step": 12680
    },
    {
      "epoch": 0.43413214652516263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4199,
      "step": 12681
    },
    {
      "epoch": 0.43416638137624103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4392,
      "step": 12682
    },
    {
      "epoch": 0.4342006162273194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4356,
      "step": 12683
    },
    {
      "epoch": 0.4342348510783978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4325,
      "step": 12684
    },
    {
      "epoch": 0.4342690859294762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4136,
      "step": 12685
    },
    {
      "epoch": 0.4343033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4046,
      "step": 12686
    },
    {
      "epoch": 0.434337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4209,
      "step": 12687
    },
    {
      "epoch": 0.4343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4403,
      "step": 12688
    },
    {
      "epoch": 0.4344060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4387,
      "step": 12689
    },
    {
      "epoch": 0.4344402601848682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4107,
      "step": 12690
    },
    {
      "epoch": 0.4344744950359466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4388,
      "step": 12691
    },
    {
      "epoch": 0.434508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4357,
      "step": 12692
    },
    {
      "epoch": 0.4345429647381034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4267,
      "step": 12693
    },
    {
      "epoch": 0.4345771995891818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4131,
      "step": 12694
    },
    {
      "epoch": 0.4346114344402602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.429,
      "step": 12695
    },
    {
      "epoch": 0.4346456692913386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4011,
      "step": 12696
    },
    {
      "epoch": 0.434679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4289,
      "step": 12697
    },
    {
      "epoch": 0.43471413899349537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4384,
      "step": 12698
    },
    {
      "epoch": 0.43474837384457377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4247,
      "step": 12699
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4099,
      "step": 12700
    },
    {
      "epoch": 0.43481684354673056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4137,
      "step": 12701
    },
    {
      "epoch": 0.43485107839780895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4101,
      "step": 12702
    },
    {
      "epoch": 0.43488531324888735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4227,
      "step": 12703
    },
    {
      "epoch": 0.43491954809996575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4274,
      "step": 12704
    },
    {
      "epoch": 0.43495378295104414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4261,
      "step": 12705
    },
    {
      "epoch": 0.43498801780212254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4198,
      "step": 12706
    },
    {
      "epoch": 0.43502225265320094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4127,
      "step": 12707
    },
    {
      "epoch": 0.43505648750427933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4528,
      "step": 12708
    },
    {
      "epoch": 0.43509072235535773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4234,
      "step": 12709
    },
    {
      "epoch": 0.4351249572064361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4331,
      "step": 12710
    },
    {
      "epoch": 0.4351591920575146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4265,
      "step": 12711
    },
    {
      "epoch": 0.435193426908593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4554,
      "step": 12712
    },
    {
      "epoch": 0.43522766175967137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4315,
      "step": 12713
    },
    {
      "epoch": 0.43526189661074977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4343,
      "step": 12714
    },
    {
      "epoch": 0.43529613146182816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4126,
      "step": 12715
    },
    {
      "epoch": 0.43533036631290656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4403,
      "step": 12716
    },
    {
      "epoch": 0.43536460116398495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4417,
      "step": 12717
    },
    {
      "epoch": 0.43539883601506335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4237,
      "step": 12718
    },
    {
      "epoch": 0.43543307086614175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4563,
      "step": 12719
    },
    {
      "epoch": 0.43546730571722014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4351,
      "step": 12720
    },
    {
      "epoch": 0.43550154056829854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4359,
      "step": 12721
    },
    {
      "epoch": 0.43553577541937694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4387,
      "step": 12722
    },
    {
      "epoch": 0.43557001027045533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4254,
      "step": 12723
    },
    {
      "epoch": 0.43560424512153373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.431,
      "step": 12724
    },
    {
      "epoch": 0.4356384799726121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4304,
      "step": 12725
    },
    {
      "epoch": 0.4356727148236905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4207,
      "step": 12726
    },
    {
      "epoch": 0.4357069496747689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4562,
      "step": 12727
    },
    {
      "epoch": 0.4357411845258473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.424,
      "step": 12728
    },
    {
      "epoch": 0.4357754193769257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4664,
      "step": 12729
    },
    {
      "epoch": 0.4358096542280041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4053,
      "step": 12730
    },
    {
      "epoch": 0.4358438890790825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4179,
      "step": 12731
    },
    {
      "epoch": 0.4358781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4273,
      "step": 12732
    },
    {
      "epoch": 0.4359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4324,
      "step": 12733
    },
    {
      "epoch": 0.4359465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 12734
    },
    {
      "epoch": 0.4359808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4459,
      "step": 12735
    },
    {
      "epoch": 0.4360150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4144,
      "step": 12736
    },
    {
      "epoch": 0.4360492981855529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4111,
      "step": 12737
    },
    {
      "epoch": 0.4360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4307,
      "step": 12738
    },
    {
      "epoch": 0.4361177678877097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.436,
      "step": 12739
    },
    {
      "epoch": 0.43615200273878807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4353,
      "step": 12740
    },
    {
      "epoch": 0.43618623758986647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4217,
      "step": 12741
    },
    {
      "epoch": 0.43622047244094486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4204,
      "step": 12742
    },
    {
      "epoch": 0.43625470729202326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4255,
      "step": 12743
    },
    {
      "epoch": 0.43628894214310165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4185,
      "step": 12744
    },
    {
      "epoch": 0.43632317699418005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4129,
      "step": 12745
    },
    {
      "epoch": 0.43635741184525845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4128,
      "step": 12746
    },
    {
      "epoch": 0.43639164669633684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4206,
      "step": 12747
    },
    {
      "epoch": 0.4364258815474153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4124,
      "step": 12748
    },
    {
      "epoch": 0.4364601163984937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4192,
      "step": 12749
    },
    {
      "epoch": 0.4364943512495721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4345,
      "step": 12750
    },
    {
      "epoch": 0.4365285861006505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4216,
      "step": 12751
    },
    {
      "epoch": 0.4365628209517289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4342,
      "step": 12752
    },
    {
      "epoch": 0.4365970558028073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4274,
      "step": 12753
    },
    {
      "epoch": 0.4366312906538857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4221,
      "step": 12754
    },
    {
      "epoch": 0.43666552550496407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.428,
      "step": 12755
    },
    {
      "epoch": 0.43669976035604247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4247,
      "step": 12756
    },
    {
      "epoch": 0.43673399520712086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4281,
      "step": 12757
    },
    {
      "epoch": 0.43676823005819926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4411,
      "step": 12758
    },
    {
      "epoch": 0.43680246490927765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4177,
      "step": 12759
    },
    {
      "epoch": 0.43683669976035605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4323,
      "step": 12760
    },
    {
      "epoch": 0.43687093461143445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4173,
      "step": 12761
    },
    {
      "epoch": 0.43690516946251284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4111,
      "step": 12762
    },
    {
      "epoch": 0.43693940431359124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4266,
      "step": 12763
    },
    {
      "epoch": 0.43697363916466964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4371,
      "step": 12764
    },
    {
      "epoch": 0.43700787401574803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4269,
      "step": 12765
    },
    {
      "epoch": 0.43704210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4376,
      "step": 12766
    },
    {
      "epoch": 0.4370763437179048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4381,
      "step": 12767
    },
    {
      "epoch": 0.4371105785689832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4128,
      "step": 12768
    },
    {
      "epoch": 0.4371448134200616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.432,
      "step": 12769
    },
    {
      "epoch": 0.43717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4244,
      "step": 12770
    },
    {
      "epoch": 0.4372132831222184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4345,
      "step": 12771
    },
    {
      "epoch": 0.4372475179732968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4429,
      "step": 12772
    },
    {
      "epoch": 0.4372817528243752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.431,
      "step": 12773
    },
    {
      "epoch": 0.4373159876754536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4072,
      "step": 12774
    },
    {
      "epoch": 0.437350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4133,
      "step": 12775
    },
    {
      "epoch": 0.4373844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4366,
      "step": 12776
    },
    {
      "epoch": 0.4374186922286888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.439,
      "step": 12777
    },
    {
      "epoch": 0.4374529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4149,
      "step": 12778
    },
    {
      "epoch": 0.4374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.439,
      "step": 12779
    },
    {
      "epoch": 0.437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4341,
      "step": 12780
    },
    {
      "epoch": 0.4375556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4195,
      "step": 12781
    },
    {
      "epoch": 0.43758986648408077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4291,
      "step": 12782
    },
    {
      "epoch": 0.43762410133515917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.405,
      "step": 12783
    },
    {
      "epoch": 0.4376583361862376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4215,
      "step": 12784
    },
    {
      "epoch": 0.437692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4245,
      "step": 12785
    },
    {
      "epoch": 0.4377268058883944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4156,
      "step": 12786
    },
    {
      "epoch": 0.4377610407394728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4357,
      "step": 12787
    },
    {
      "epoch": 0.4377952755905512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4232,
      "step": 12788
    },
    {
      "epoch": 0.4378295104416296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4269,
      "step": 12789
    },
    {
      "epoch": 0.437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4024,
      "step": 12790
    },
    {
      "epoch": 0.4378979801437864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4126,
      "step": 12791
    },
    {
      "epoch": 0.4379322149948648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4213,
      "step": 12792
    },
    {
      "epoch": 0.4379664498459432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.428,
      "step": 12793
    },
    {
      "epoch": 0.4380006846970216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4181,
      "step": 12794
    },
    {
      "epoch": 0.4380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4201,
      "step": 12795
    },
    {
      "epoch": 0.4380691543991784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4161,
      "step": 12796
    },
    {
      "epoch": 0.43810338925025677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4355,
      "step": 12797
    },
    {
      "epoch": 0.43813762410133517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4365,
      "step": 12798
    },
    {
      "epoch": 0.43817185895241356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4437,
      "step": 12799
    },
    {
      "epoch": 0.43820609380349196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4431,
      "step": 12800
    },
    {
      "epoch": 0.43824032865457035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4147,
      "step": 12801
    },
    {
      "epoch": 0.43827456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4132,
      "step": 12802
    },
    {
      "epoch": 0.43830879835672715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4141,
      "step": 12803
    },
    {
      "epoch": 0.43834303320780554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4285,
      "step": 12804
    },
    {
      "epoch": 0.43837726805888394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4335,
      "step": 12805
    },
    {
      "epoch": 0.43841150290996234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.45,
      "step": 12806
    },
    {
      "epoch": 0.43844573776104073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4387,
      "step": 12807
    },
    {
      "epoch": 0.43847997261211913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4287,
      "step": 12808
    },
    {
      "epoch": 0.4385142074631975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4094,
      "step": 12809
    },
    {
      "epoch": 0.4385484423142759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4438,
      "step": 12810
    },
    {
      "epoch": 0.4385826771653543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4202,
      "step": 12811
    },
    {
      "epoch": 0.4386169120164327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4092,
      "step": 12812
    },
    {
      "epoch": 0.4386511468675111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4371,
      "step": 12813
    },
    {
      "epoch": 0.4386853817185895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4285,
      "step": 12814
    },
    {
      "epoch": 0.4387196165696679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4171,
      "step": 12815
    },
    {
      "epoch": 0.4387538514207463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.424,
      "step": 12816
    },
    {
      "epoch": 0.4387880862718247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4233,
      "step": 12817
    },
    {
      "epoch": 0.4388223211229031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4258,
      "step": 12818
    },
    {
      "epoch": 0.4388565559739815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4225,
      "step": 12819
    },
    {
      "epoch": 0.4388907908250599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4326,
      "step": 12820
    },
    {
      "epoch": 0.43892502567613834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4258,
      "step": 12821
    },
    {
      "epoch": 0.43895926052721673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4237,
      "step": 12822
    },
    {
      "epoch": 0.43899349537829513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4074,
      "step": 12823
    },
    {
      "epoch": 0.4390277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4252,
      "step": 12824
    },
    {
      "epoch": 0.4390619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4231,
      "step": 12825
    },
    {
      "epoch": 0.4390961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4187,
      "step": 12826
    },
    {
      "epoch": 0.4391304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4293,
      "step": 12827
    },
    {
      "epoch": 0.4391646696336871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4325,
      "step": 12828
    },
    {
      "epoch": 0.4391989044847655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4388,
      "step": 12829
    },
    {
      "epoch": 0.4392331393358439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4217,
      "step": 12830
    },
    {
      "epoch": 0.4392673741869223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4254,
      "step": 12831
    },
    {
      "epoch": 0.4393016090380007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4246,
      "step": 12832
    },
    {
      "epoch": 0.4393358438890791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.416,
      "step": 12833
    },
    {
      "epoch": 0.4393700787401575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4016,
      "step": 12834
    },
    {
      "epoch": 0.4394043135912359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4409,
      "step": 12835
    },
    {
      "epoch": 0.4394385484423143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4155,
      "step": 12836
    },
    {
      "epoch": 0.4394727832933927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4109,
      "step": 12837
    },
    {
      "epoch": 0.4395070181444711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4243,
      "step": 12838
    },
    {
      "epoch": 0.43954125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4229,
      "step": 12839
    },
    {
      "epoch": 0.43957548784662787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4154,
      "step": 12840
    },
    {
      "epoch": 0.43960972269770626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4287,
      "step": 12841
    },
    {
      "epoch": 0.43964395754878466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.439,
      "step": 12842
    },
    {
      "epoch": 0.43967819239986305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4127,
      "step": 12843
    },
    {
      "epoch": 0.43971242725094145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4309,
      "step": 12844
    },
    {
      "epoch": 0.43974666210201985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4258,
      "step": 12845
    },
    {
      "epoch": 0.43978089695309824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4205,
      "step": 12846
    },
    {
      "epoch": 0.43981513180417664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4304,
      "step": 12847
    },
    {
      "epoch": 0.43984936665525504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4222,
      "step": 12848
    },
    {
      "epoch": 0.43988360150633343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4301,
      "step": 12849
    },
    {
      "epoch": 0.43991783635741183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4208,
      "step": 12850
    },
    {
      "epoch": 0.4399520712084902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4046,
      "step": 12851
    },
    {
      "epoch": 0.4399863060595686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4203,
      "step": 12852
    },
    {
      "epoch": 0.440020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4202,
      "step": 12853
    },
    {
      "epoch": 0.4400547757617254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4173,
      "step": 12854
    },
    {
      "epoch": 0.4400890106128038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4299,
      "step": 12855
    },
    {
      "epoch": 0.4401232454638822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4119,
      "step": 12856
    },
    {
      "epoch": 0.4401574803149606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4369,
      "step": 12857
    },
    {
      "epoch": 0.44019171516603905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4187,
      "step": 12858
    },
    {
      "epoch": 0.44022595001711745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4039,
      "step": 12859
    },
    {
      "epoch": 0.44026018486819585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4205,
      "step": 12860
    },
    {
      "epoch": 0.44029441971927424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4359,
      "step": 12861
    },
    {
      "epoch": 0.44032865457035264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3946,
      "step": 12862
    },
    {
      "epoch": 0.44036288942143104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4359,
      "step": 12863
    },
    {
      "epoch": 0.44039712427250943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4486,
      "step": 12864
    },
    {
      "epoch": 0.44043135912358783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.432,
      "step": 12865
    },
    {
      "epoch": 0.4404655939746662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.446,
      "step": 12866
    },
    {
      "epoch": 0.4404998288257446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.41,
      "step": 12867
    },
    {
      "epoch": 0.440534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4453,
      "step": 12868
    },
    {
      "epoch": 0.4405682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4232,
      "step": 12869
    },
    {
      "epoch": 0.4406025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4356,
      "step": 12870
    },
    {
      "epoch": 0.4406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4059,
      "step": 12871
    },
    {
      "epoch": 0.4406710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4152,
      "step": 12872
    },
    {
      "epoch": 0.440705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.44,
      "step": 12873
    },
    {
      "epoch": 0.4407394727832934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4305,
      "step": 12874
    },
    {
      "epoch": 0.4407737076343718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4248,
      "step": 12875
    },
    {
      "epoch": 0.4408079424854502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4144,
      "step": 12876
    },
    {
      "epoch": 0.4408421773365286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4123,
      "step": 12877
    },
    {
      "epoch": 0.440876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4153,
      "step": 12878
    },
    {
      "epoch": 0.4409106470386854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4252,
      "step": 12879
    },
    {
      "epoch": 0.4409448818897638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4221,
      "step": 12880
    },
    {
      "epoch": 0.44097911674084217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.403,
      "step": 12881
    },
    {
      "epoch": 0.44101335159192057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4211,
      "step": 12882
    },
    {
      "epoch": 0.44104758644299896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4296,
      "step": 12883
    },
    {
      "epoch": 0.44108182129407736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4302,
      "step": 12884
    },
    {
      "epoch": 0.44111605614515575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4277,
      "step": 12885
    },
    {
      "epoch": 0.44115029099623415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4382,
      "step": 12886
    },
    {
      "epoch": 0.44118452584731255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4189,
      "step": 12887
    },
    {
      "epoch": 0.44121876069839094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4336,
      "step": 12888
    },
    {
      "epoch": 0.44125299554946934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4453,
      "step": 12889
    },
    {
      "epoch": 0.44128723040054774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4257,
      "step": 12890
    },
    {
      "epoch": 0.44132146525162613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4006,
      "step": 12891
    },
    {
      "epoch": 0.44135570010270453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4112,
      "step": 12892
    },
    {
      "epoch": 0.4413899349537829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4053,
      "step": 12893
    },
    {
      "epoch": 0.4414241698048613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4207,
      "step": 12894
    },
    {
      "epoch": 0.4414584046559398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4235,
      "step": 12895
    },
    {
      "epoch": 0.44149263950701817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4237,
      "step": 12896
    },
    {
      "epoch": 0.44152687435809657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4365,
      "step": 12897
    },
    {
      "epoch": 0.44156110920917496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4144,
      "step": 12898
    },
    {
      "epoch": 0.44159534406025336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4213,
      "step": 12899
    },
    {
      "epoch": 0.44162957891133175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4003,
      "step": 12900
    },
    {
      "epoch": 0.44166381376241015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4006,
      "step": 12901
    },
    {
      "epoch": 0.44169804861348855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4303,
      "step": 12902
    },
    {
      "epoch": 0.44173228346456694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4078,
      "step": 12903
    },
    {
      "epoch": 0.44176651831564534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4314,
      "step": 12904
    },
    {
      "epoch": 0.44180075316672374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4312,
      "step": 12905
    },
    {
      "epoch": 0.44183498801780213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.423,
      "step": 12906
    },
    {
      "epoch": 0.44186922286888053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4361,
      "step": 12907
    },
    {
      "epoch": 0.4419034577199589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4172,
      "step": 12908
    },
    {
      "epoch": 0.4419376925710373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4335,
      "step": 12909
    },
    {
      "epoch": 0.4419719274221157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4175,
      "step": 12910
    },
    {
      "epoch": 0.4420061622731941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4312,
      "step": 12911
    },
    {
      "epoch": 0.4420403971242725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4379,
      "step": 12912
    },
    {
      "epoch": 0.4420746319753509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4394,
      "step": 12913
    },
    {
      "epoch": 0.4421088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4091,
      "step": 12914
    },
    {
      "epoch": 0.4421431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4227,
      "step": 12915
    },
    {
      "epoch": 0.4421773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4116,
      "step": 12916
    },
    {
      "epoch": 0.4422115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4193,
      "step": 12917
    },
    {
      "epoch": 0.4422458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4276,
      "step": 12918
    },
    {
      "epoch": 0.4422800410818213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4196,
      "step": 12919
    },
    {
      "epoch": 0.4423142759328997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4174,
      "step": 12920
    },
    {
      "epoch": 0.4423485107839781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4324,
      "step": 12921
    },
    {
      "epoch": 0.4423827456350565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4296,
      "step": 12922
    },
    {
      "epoch": 0.44241698048613487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4378,
      "step": 12923
    },
    {
      "epoch": 0.44245121533721327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4343,
      "step": 12924
    },
    {
      "epoch": 0.44248545018829166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4317,
      "step": 12925
    },
    {
      "epoch": 0.44251968503937006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4139,
      "step": 12926
    },
    {
      "epoch": 0.44255391989044846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4268,
      "step": 12927
    },
    {
      "epoch": 0.44258815474152685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4305,
      "step": 12928
    },
    {
      "epoch": 0.44262238959260525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4221,
      "step": 12929
    },
    {
      "epoch": 0.44265662444368364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4269,
      "step": 12930
    },
    {
      "epoch": 0.44269085929476204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4085,
      "step": 12931
    },
    {
      "epoch": 0.4427250941458405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4112,
      "step": 12932
    },
    {
      "epoch": 0.4427593289969189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4263,
      "step": 12933
    },
    {
      "epoch": 0.4427935638479973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4082,
      "step": 12934
    },
    {
      "epoch": 0.4428277986990757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4227,
      "step": 12935
    },
    {
      "epoch": 0.4428620335501541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4355,
      "step": 12936
    },
    {
      "epoch": 0.4428962684012325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4186,
      "step": 12937
    },
    {
      "epoch": 0.44293050325231087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4157,
      "step": 12938
    },
    {
      "epoch": 0.44296473810338927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4215,
      "step": 12939
    },
    {
      "epoch": 0.44299897295446766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4356,
      "step": 12940
    },
    {
      "epoch": 0.44303320780554606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4428,
      "step": 12941
    },
    {
      "epoch": 0.44306744265662446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4143,
      "step": 12942
    },
    {
      "epoch": 0.44310167750770285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4328,
      "step": 12943
    },
    {
      "epoch": 0.44313591235878125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4332,
      "step": 12944
    },
    {
      "epoch": 0.44317014720985964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4248,
      "step": 12945
    },
    {
      "epoch": 0.44320438206093804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4072,
      "step": 12946
    },
    {
      "epoch": 0.44323861691201644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4108,
      "step": 12947
    },
    {
      "epoch": 0.44327285176309483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4153,
      "step": 12948
    },
    {
      "epoch": 0.44330708661417323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4121,
      "step": 12949
    },
    {
      "epoch": 0.4433413214652516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4296,
      "step": 12950
    },
    {
      "epoch": 0.44337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4218,
      "step": 12951
    },
    {
      "epoch": 0.4434097911674084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4045,
      "step": 12952
    },
    {
      "epoch": 0.4434440260184868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4211,
      "step": 12953
    },
    {
      "epoch": 0.4434782608695652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4147,
      "step": 12954
    },
    {
      "epoch": 0.4435124957206436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4231,
      "step": 12955
    },
    {
      "epoch": 0.443546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4217,
      "step": 12956
    },
    {
      "epoch": 0.4435809654228004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4175,
      "step": 12957
    },
    {
      "epoch": 0.4436152002738788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4353,
      "step": 12958
    },
    {
      "epoch": 0.4436494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4089,
      "step": 12959
    },
    {
      "epoch": 0.4436836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4199,
      "step": 12960
    },
    {
      "epoch": 0.443717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4329,
      "step": 12961
    },
    {
      "epoch": 0.4437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4031,
      "step": 12962
    },
    {
      "epoch": 0.4437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4136,
      "step": 12963
    },
    {
      "epoch": 0.4438206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4213,
      "step": 12964
    },
    {
      "epoch": 0.44385484423142757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4174,
      "step": 12965
    },
    {
      "epoch": 0.44388907908250597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4451,
      "step": 12966
    },
    {
      "epoch": 0.44392331393358436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.426,
      "step": 12967
    },
    {
      "epoch": 0.4439575487846628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4069,
      "step": 12968
    },
    {
      "epoch": 0.4439917836357412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4101,
      "step": 12969
    },
    {
      "epoch": 0.4440260184868196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4183,
      "step": 12970
    },
    {
      "epoch": 0.444060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4382,
      "step": 12971
    },
    {
      "epoch": 0.4440944881889764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.421,
      "step": 12972
    },
    {
      "epoch": 0.4441287230400548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4302,
      "step": 12973
    },
    {
      "epoch": 0.4441629578911332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4335,
      "step": 12974
    },
    {
      "epoch": 0.4441971927422116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4254,
      "step": 12975
    },
    {
      "epoch": 0.44423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3959,
      "step": 12976
    },
    {
      "epoch": 0.4442656624443684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.402,
      "step": 12977
    },
    {
      "epoch": 0.4442998972954468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4254,
      "step": 12978
    },
    {
      "epoch": 0.4443341321465252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4139,
      "step": 12979
    },
    {
      "epoch": 0.44436836699760357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4423,
      "step": 12980
    },
    {
      "epoch": 0.44440260184868197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4077,
      "step": 12981
    },
    {
      "epoch": 0.44443683669976036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4218,
      "step": 12982
    },
    {
      "epoch": 0.44447107155083876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4274,
      "step": 12983
    },
    {
      "epoch": 0.44450530640191716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3914,
      "step": 12984
    },
    {
      "epoch": 0.44453954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4233,
      "step": 12985
    },
    {
      "epoch": 0.44457377610407395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4135,
      "step": 12986
    },
    {
      "epoch": 0.44460801095515234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4426,
      "step": 12987
    },
    {
      "epoch": 0.44464224580623074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4303,
      "step": 12988
    },
    {
      "epoch": 0.44467648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4312,
      "step": 12989
    },
    {
      "epoch": 0.44471071550838753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4353,
      "step": 12990
    },
    {
      "epoch": 0.44474495035946593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4131,
      "step": 12991
    },
    {
      "epoch": 0.4447791852105443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4373,
      "step": 12992
    },
    {
      "epoch": 0.4448134200616227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4169,
      "step": 12993
    },
    {
      "epoch": 0.4448476549127011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4191,
      "step": 12994
    },
    {
      "epoch": 0.4448818897637795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4208,
      "step": 12995
    },
    {
      "epoch": 0.4449161246148579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4092,
      "step": 12996
    },
    {
      "epoch": 0.4449503594659363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4357,
      "step": 12997
    },
    {
      "epoch": 0.4449845943170147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4349,
      "step": 12998
    },
    {
      "epoch": 0.4450188291680931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4368,
      "step": 12999
    },
    {
      "epoch": 0.4450530640191715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4165,
      "step": 13000
    },
    {
      "epoch": 0.4450872988702499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4148,
      "step": 13001
    },
    {
      "epoch": 0.4451215337213283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4122,
      "step": 13002
    },
    {
      "epoch": 0.4451557685724067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4263,
      "step": 13003
    },
    {
      "epoch": 0.4451900034234851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4337,
      "step": 13004
    },
    {
      "epoch": 0.44522423827456353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4171,
      "step": 13005
    },
    {
      "epoch": 0.44525847312564193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4045,
      "step": 13006
    },
    {
      "epoch": 0.4452927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4172,
      "step": 13007
    },
    {
      "epoch": 0.4453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4198,
      "step": 13008
    },
    {
      "epoch": 0.4453611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4298,
      "step": 13009
    },
    {
      "epoch": 0.4453954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.424,
      "step": 13010
    },
    {
      "epoch": 0.4454296473810339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4188,
      "step": 13011
    },
    {
      "epoch": 0.4454638822321123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4133,
      "step": 13012
    },
    {
      "epoch": 0.4454981170831907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4251,
      "step": 13013
    },
    {
      "epoch": 0.4455323519342691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4295,
      "step": 13014
    },
    {
      "epoch": 0.4455665867853475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4223,
      "step": 13015
    },
    {
      "epoch": 0.4456008216364259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4184,
      "step": 13016
    },
    {
      "epoch": 0.4456350564875043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.427,
      "step": 13017
    },
    {
      "epoch": 0.4456692913385827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4396,
      "step": 13018
    },
    {
      "epoch": 0.4457035261896611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4305,
      "step": 13019
    },
    {
      "epoch": 0.4457377610407395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4273,
      "step": 13020
    },
    {
      "epoch": 0.4457719958918179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4274,
      "step": 13021
    },
    {
      "epoch": 0.44580623074289627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4327,
      "step": 13022
    },
    {
      "epoch": 0.44584046559397467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4233,
      "step": 13023
    },
    {
      "epoch": 0.44587470044505306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4131,
      "step": 13024
    },
    {
      "epoch": 0.44590893529613146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4322,
      "step": 13025
    },
    {
      "epoch": 0.44594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4174,
      "step": 13026
    },
    {
      "epoch": 0.44597740499828825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4182,
      "step": 13027
    },
    {
      "epoch": 0.44601163984936665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4146,
      "step": 13028
    },
    {
      "epoch": 0.44604587470044504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4294,
      "step": 13029
    },
    {
      "epoch": 0.44608010955152344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4159,
      "step": 13030
    },
    {
      "epoch": 0.44611434440260184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4347,
      "step": 13031
    },
    {
      "epoch": 0.44614857925368023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4189,
      "step": 13032
    },
    {
      "epoch": 0.44618281410475863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4164,
      "step": 13033
    },
    {
      "epoch": 0.446217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4287,
      "step": 13034
    },
    {
      "epoch": 0.4462512838069154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4198,
      "step": 13035
    },
    {
      "epoch": 0.4462855186579938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4031,
      "step": 13036
    },
    {
      "epoch": 0.4463197535090722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4251,
      "step": 13037
    },
    {
      "epoch": 0.4463539883601506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.424,
      "step": 13038
    },
    {
      "epoch": 0.446388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4338,
      "step": 13039
    },
    {
      "epoch": 0.4464224580623074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3955,
      "step": 13040
    },
    {
      "epoch": 0.4464566929133858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4049,
      "step": 13041
    },
    {
      "epoch": 0.44649092776446425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4214,
      "step": 13042
    },
    {
      "epoch": 0.44652516261554265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3969,
      "step": 13043
    },
    {
      "epoch": 0.44655939746662104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4283,
      "step": 13044
    },
    {
      "epoch": 0.44659363231769944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4176,
      "step": 13045
    },
    {
      "epoch": 0.44662786716877784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3985,
      "step": 13046
    },
    {
      "epoch": 0.44666210201985623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4229,
      "step": 13047
    },
    {
      "epoch": 0.44669633687093463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4301,
      "step": 13048
    },
    {
      "epoch": 0.446730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4337,
      "step": 13049
    },
    {
      "epoch": 0.4467648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4256,
      "step": 13050
    },
    {
      "epoch": 0.4467990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4205,
      "step": 13051
    },
    {
      "epoch": 0.4468332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.431,
      "step": 13052
    },
    {
      "epoch": 0.4468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4278,
      "step": 13053
    },
    {
      "epoch": 0.446901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4457,
      "step": 13054
    },
    {
      "epoch": 0.4469359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3991,
      "step": 13055
    },
    {
      "epoch": 0.4469702156795618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4196,
      "step": 13056
    },
    {
      "epoch": 0.4470044505306402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4117,
      "step": 13057
    },
    {
      "epoch": 0.4470386853817186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4058,
      "step": 13058
    },
    {
      "epoch": 0.447072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4135,
      "step": 13059
    },
    {
      "epoch": 0.4471071550838754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.405,
      "step": 13060
    },
    {
      "epoch": 0.4471413899349538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4096,
      "step": 13061
    },
    {
      "epoch": 0.4471756247860322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4047,
      "step": 13062
    },
    {
      "epoch": 0.4472098596371106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4342,
      "step": 13063
    },
    {
      "epoch": 0.44724409448818897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4108,
      "step": 13064
    },
    {
      "epoch": 0.44727832933926737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4116,
      "step": 13065
    },
    {
      "epoch": 0.44731256419034576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4001,
      "step": 13066
    },
    {
      "epoch": 0.44734679904142416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4159,
      "step": 13067
    },
    {
      "epoch": 0.44738103389250256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4294,
      "step": 13068
    },
    {
      "epoch": 0.44741526874358095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4153,
      "step": 13069
    },
    {
      "epoch": 0.44744950359465935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4226,
      "step": 13070
    },
    {
      "epoch": 0.44748373844573774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4473,
      "step": 13071
    },
    {
      "epoch": 0.44751797329681614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3911,
      "step": 13072
    },
    {
      "epoch": 0.44755220814789454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4366,
      "step": 13073
    },
    {
      "epoch": 0.44758644299897293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4017,
      "step": 13074
    },
    {
      "epoch": 0.44762067785005133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4216,
      "step": 13075
    },
    {
      "epoch": 0.4476549127011297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4244,
      "step": 13076
    },
    {
      "epoch": 0.4476891475522081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4132,
      "step": 13077
    },
    {
      "epoch": 0.4477233824032865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4367,
      "step": 13078
    },
    {
      "epoch": 0.44775761725436497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4097,
      "step": 13079
    },
    {
      "epoch": 0.44779185210544337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.423,
      "step": 13080
    },
    {
      "epoch": 0.44782608695652176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4274,
      "step": 13081
    },
    {
      "epoch": 0.44786032180760016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4145,
      "step": 13082
    },
    {
      "epoch": 0.44789455665867856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4144,
      "step": 13083
    },
    {
      "epoch": 0.44792879150975695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4112,
      "step": 13084
    },
    {
      "epoch": 0.44796302636083535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4097,
      "step": 13085
    },
    {
      "epoch": 0.44799726121191374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4185,
      "step": 13086
    },
    {
      "epoch": 0.44803149606299214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.426,
      "step": 13087
    },
    {
      "epoch": 0.44806573091407054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4028,
      "step": 13088
    },
    {
      "epoch": 0.44809996576514893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4085,
      "step": 13089
    },
    {
      "epoch": 0.44813420061622733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4219,
      "step": 13090
    },
    {
      "epoch": 0.4481684354673057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4174,
      "step": 13091
    },
    {
      "epoch": 0.4482026703183841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4318,
      "step": 13092
    },
    {
      "epoch": 0.4482369051694625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4261,
      "step": 13093
    },
    {
      "epoch": 0.4482711400205409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4136,
      "step": 13094
    },
    {
      "epoch": 0.4483053748716193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4224,
      "step": 13095
    },
    {
      "epoch": 0.4483396097226977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4117,
      "step": 13096
    },
    {
      "epoch": 0.4483738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4199,
      "step": 13097
    },
    {
      "epoch": 0.4484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4122,
      "step": 13098
    },
    {
      "epoch": 0.4484423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4226,
      "step": 13099
    },
    {
      "epoch": 0.4484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4081,
      "step": 13100
    },
    {
      "epoch": 0.4485107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4401,
      "step": 13101
    },
    {
      "epoch": 0.4485450188291681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4122,
      "step": 13102
    },
    {
      "epoch": 0.4485792536802465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.418,
      "step": 13103
    },
    {
      "epoch": 0.4486134885313249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4273,
      "step": 13104
    },
    {
      "epoch": 0.4486477233824033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4127,
      "step": 13105
    },
    {
      "epoch": 0.44868195823348167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4071,
      "step": 13106
    },
    {
      "epoch": 0.44871619308456007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4066,
      "step": 13107
    },
    {
      "epoch": 0.44875042793563846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4135,
      "step": 13108
    },
    {
      "epoch": 0.44878466278671686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4066,
      "step": 13109
    },
    {
      "epoch": 0.44881889763779526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4273,
      "step": 13110
    },
    {
      "epoch": 0.44885313248887365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4106,
      "step": 13111
    },
    {
      "epoch": 0.44888736733995205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4236,
      "step": 13112
    },
    {
      "epoch": 0.44892160219103044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4155,
      "step": 13113
    },
    {
      "epoch": 0.44895583704210884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4278,
      "step": 13114
    },
    {
      "epoch": 0.44899007189318724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.42,
      "step": 13115
    },
    {
      "epoch": 0.4490243067442657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.398,
      "step": 13116
    },
    {
      "epoch": 0.4490585415953441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4199,
      "step": 13117
    },
    {
      "epoch": 0.4490927764464225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.402,
      "step": 13118
    },
    {
      "epoch": 0.4491270112975009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4149,
      "step": 13119
    },
    {
      "epoch": 0.4491612461485793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4161,
      "step": 13120
    },
    {
      "epoch": 0.44919548099965767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4377,
      "step": 13121
    },
    {
      "epoch": 0.44922971585073607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4111,
      "step": 13122
    },
    {
      "epoch": 0.44926395070181446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4214,
      "step": 13123
    },
    {
      "epoch": 0.44929818555289286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4212,
      "step": 13124
    },
    {
      "epoch": 0.44933242040397126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4114,
      "step": 13125
    },
    {
      "epoch": 0.44936665525504965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4058,
      "step": 13126
    },
    {
      "epoch": 0.44940089010612805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4184,
      "step": 13127
    },
    {
      "epoch": 0.44943512495720644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4043,
      "step": 13128
    },
    {
      "epoch": 0.44946935980828484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4233,
      "step": 13129
    },
    {
      "epoch": 0.44950359465936324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4273,
      "step": 13130
    },
    {
      "epoch": 0.44953782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4248,
      "step": 13131
    },
    {
      "epoch": 0.44957206436152003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4215,
      "step": 13132
    },
    {
      "epoch": 0.4496062992125984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4232,
      "step": 13133
    },
    {
      "epoch": 0.4496405340636768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3974,
      "step": 13134
    },
    {
      "epoch": 0.4496747689147552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.407,
      "step": 13135
    },
    {
      "epoch": 0.4497090037658336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3979,
      "step": 13136
    },
    {
      "epoch": 0.449743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4222,
      "step": 13137
    },
    {
      "epoch": 0.4497774734679904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4086,
      "step": 13138
    },
    {
      "epoch": 0.4498117083190688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4084,
      "step": 13139
    },
    {
      "epoch": 0.4498459431701472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4149,
      "step": 13140
    },
    {
      "epoch": 0.4498801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4091,
      "step": 13141
    },
    {
      "epoch": 0.449914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3985,
      "step": 13142
    },
    {
      "epoch": 0.4499486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4183,
      "step": 13143
    },
    {
      "epoch": 0.4499828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4243,
      "step": 13144
    },
    {
      "epoch": 0.4500171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4162,
      "step": 13145
    },
    {
      "epoch": 0.4500513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4205,
      "step": 13146
    },
    {
      "epoch": 0.450085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4024,
      "step": 13147
    },
    {
      "epoch": 0.45011982197877437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4179,
      "step": 13148
    },
    {
      "epoch": 0.45015405682985277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4164,
      "step": 13149
    },
    {
      "epoch": 0.45018829168093116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4307,
      "step": 13150
    },
    {
      "epoch": 0.45022252653200956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4103,
      "step": 13151
    },
    {
      "epoch": 0.450256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4313,
      "step": 13152
    },
    {
      "epoch": 0.4502909962341664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4072,
      "step": 13153
    },
    {
      "epoch": 0.4503252310852448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4111,
      "step": 13154
    },
    {
      "epoch": 0.4503594659363232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4277,
      "step": 13155
    },
    {
      "epoch": 0.4503937007874016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4063,
      "step": 13156
    },
    {
      "epoch": 0.45042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4266,
      "step": 13157
    },
    {
      "epoch": 0.4504621704895584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4343,
      "step": 13158
    },
    {
      "epoch": 0.4504964053406368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4332,
      "step": 13159
    },
    {
      "epoch": 0.4505306401917152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4193,
      "step": 13160
    },
    {
      "epoch": 0.4505648750427936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4245,
      "step": 13161
    },
    {
      "epoch": 0.450599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4279,
      "step": 13162
    },
    {
      "epoch": 0.45063334474495037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4104,
      "step": 13163
    },
    {
      "epoch": 0.45066757959602877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4127,
      "step": 13164
    },
    {
      "epoch": 0.45070181444710716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.395,
      "step": 13165
    },
    {
      "epoch": 0.45073604929818556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4078,
      "step": 13166
    },
    {
      "epoch": 0.45077028414926396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4241,
      "step": 13167
    },
    {
      "epoch": 0.45080451900034235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4146,
      "step": 13168
    },
    {
      "epoch": 0.45083875385142075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4157,
      "step": 13169
    },
    {
      "epoch": 0.45087298870249914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4086,
      "step": 13170
    },
    {
      "epoch": 0.45090722355357754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4147,
      "step": 13171
    },
    {
      "epoch": 0.45094145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.408,
      "step": 13172
    },
    {
      "epoch": 0.45097569325573433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.415,
      "step": 13173
    },
    {
      "epoch": 0.45100992810681273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3914,
      "step": 13174
    },
    {
      "epoch": 0.4510441629578911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4092,
      "step": 13175
    },
    {
      "epoch": 0.4510783978089695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4133,
      "step": 13176
    },
    {
      "epoch": 0.4511126326600479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4095,
      "step": 13177
    },
    {
      "epoch": 0.4511468675111263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4252,
      "step": 13178
    },
    {
      "epoch": 0.4511811023622047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.406,
      "step": 13179
    },
    {
      "epoch": 0.4512153372132831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4261,
      "step": 13180
    },
    {
      "epoch": 0.4512495720643615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4343,
      "step": 13181
    },
    {
      "epoch": 0.4512838069154399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4028,
      "step": 13182
    },
    {
      "epoch": 0.4513180417665183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4193,
      "step": 13183
    },
    {
      "epoch": 0.4513522766175967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.408,
      "step": 13184
    },
    {
      "epoch": 0.4513865114686751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4143,
      "step": 13185
    },
    {
      "epoch": 0.4514207463197535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4296,
      "step": 13186
    },
    {
      "epoch": 0.4514549811708319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4162,
      "step": 13187
    },
    {
      "epoch": 0.4514892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4241,
      "step": 13188
    },
    {
      "epoch": 0.45152345087298873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4052,
      "step": 13189
    },
    {
      "epoch": 0.4515576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4178,
      "step": 13190
    },
    {
      "epoch": 0.4515919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4221,
      "step": 13191
    },
    {
      "epoch": 0.4516261554262239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4023,
      "step": 13192
    },
    {
      "epoch": 0.4516603902773023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4215,
      "step": 13193
    },
    {
      "epoch": 0.4516946251283807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4004,
      "step": 13194
    },
    {
      "epoch": 0.4517288599794591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4206,
      "step": 13195
    },
    {
      "epoch": 0.4517630948305375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4102,
      "step": 13196
    },
    {
      "epoch": 0.4517973296816159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4122,
      "step": 13197
    },
    {
      "epoch": 0.4518315645326943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4181,
      "step": 13198
    },
    {
      "epoch": 0.4518657993837727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4193,
      "step": 13199
    },
    {
      "epoch": 0.4519000342348511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4171,
      "step": 13200
    },
    {
      "epoch": 0.4519342690859295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4132,
      "step": 13201
    },
    {
      "epoch": 0.4519685039370079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4027,
      "step": 13202
    },
    {
      "epoch": 0.4520027387880863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3954,
      "step": 13203
    },
    {
      "epoch": 0.4520369736391647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4036,
      "step": 13204
    },
    {
      "epoch": 0.45207120849024307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.419,
      "step": 13205
    },
    {
      "epoch": 0.45210544334132147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4153,
      "step": 13206
    },
    {
      "epoch": 0.45213967819239986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4164,
      "step": 13207
    },
    {
      "epoch": 0.45217391304347826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4307,
      "step": 13208
    },
    {
      "epoch": 0.45220814789455666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4091,
      "step": 13209
    },
    {
      "epoch": 0.45224238274563505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4152,
      "step": 13210
    },
    {
      "epoch": 0.45227661759671345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4078,
      "step": 13211
    },
    {
      "epoch": 0.45231085244779184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3973,
      "step": 13212
    },
    {
      "epoch": 0.45234508729887024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4394,
      "step": 13213
    },
    {
      "epoch": 0.45237932214994864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.407,
      "step": 13214
    },
    {
      "epoch": 0.45241355700102703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4106,
      "step": 13215
    },
    {
      "epoch": 0.45244779185210543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3925,
      "step": 13216
    },
    {
      "epoch": 0.4524820267031838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4271,
      "step": 13217
    },
    {
      "epoch": 0.4525162615542622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4412,
      "step": 13218
    },
    {
      "epoch": 0.4525504964053406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4097,
      "step": 13219
    },
    {
      "epoch": 0.452584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4183,
      "step": 13220
    },
    {
      "epoch": 0.4526189661074974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4044,
      "step": 13221
    },
    {
      "epoch": 0.4526532009585758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4121,
      "step": 13222
    },
    {
      "epoch": 0.4526874358096542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4233,
      "step": 13223
    },
    {
      "epoch": 0.4527216706607326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4184,
      "step": 13224
    },
    {
      "epoch": 0.452755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4153,
      "step": 13225
    },
    {
      "epoch": 0.45279014036288945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4274,
      "step": 13226
    },
    {
      "epoch": 0.45282437521396784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.424,
      "step": 13227
    },
    {
      "epoch": 0.45285861006504624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4098,
      "step": 13228
    },
    {
      "epoch": 0.45289284491612464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4361,
      "step": 13229
    },
    {
      "epoch": 0.45292707976720303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4168,
      "step": 13230
    },
    {
      "epoch": 0.45296131461828143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4203,
      "step": 13231
    },
    {
      "epoch": 0.4529955494693598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3973,
      "step": 13232
    },
    {
      "epoch": 0.4530297843204382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4124,
      "step": 13233
    },
    {
      "epoch": 0.4530640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4154,
      "step": 13234
    },
    {
      "epoch": 0.453098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4213,
      "step": 13235
    },
    {
      "epoch": 0.4531324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4132,
      "step": 13236
    },
    {
      "epoch": 0.4531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4059,
      "step": 13237
    },
    {
      "epoch": 0.4532009585758302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4298,
      "step": 13238
    },
    {
      "epoch": 0.4532351934269086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4188,
      "step": 13239
    },
    {
      "epoch": 0.453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3969,
      "step": 13240
    },
    {
      "epoch": 0.4533036631290654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4258,
      "step": 13241
    },
    {
      "epoch": 0.4533378979801438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4295,
      "step": 13242
    },
    {
      "epoch": 0.4533721328312222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3956,
      "step": 13243
    },
    {
      "epoch": 0.4534063676823006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4074,
      "step": 13244
    },
    {
      "epoch": 0.453440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3967,
      "step": 13245
    },
    {
      "epoch": 0.4534748373844574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4212,
      "step": 13246
    },
    {
      "epoch": 0.45350907223553577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4056,
      "step": 13247
    },
    {
      "epoch": 0.45354330708661417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3991,
      "step": 13248
    },
    {
      "epoch": 0.45357754193769256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.423,
      "step": 13249
    },
    {
      "epoch": 0.45361177678877096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4147,
      "step": 13250
    },
    {
      "epoch": 0.45364601163984936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3964,
      "step": 13251
    },
    {
      "epoch": 0.45368024649092775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3975,
      "step": 13252
    },
    {
      "epoch": 0.45371448134200615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4224,
      "step": 13253
    },
    {
      "epoch": 0.45374871619308454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4142,
      "step": 13254
    },
    {
      "epoch": 0.45378295104416294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4311,
      "step": 13255
    },
    {
      "epoch": 0.45381718589524134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4044,
      "step": 13256
    },
    {
      "epoch": 0.45385142074631973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4119,
      "step": 13257
    },
    {
      "epoch": 0.45388565559739813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4074,
      "step": 13258
    },
    {
      "epoch": 0.4539198904484765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4099,
      "step": 13259
    },
    {
      "epoch": 0.4539541252995549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4098,
      "step": 13260
    },
    {
      "epoch": 0.4539883601506333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4064,
      "step": 13261
    },
    {
      "epoch": 0.4540225950017117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4211,
      "step": 13262
    },
    {
      "epoch": 0.45405682985279017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4241,
      "step": 13263
    },
    {
      "epoch": 0.45409106470386856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4095,
      "step": 13264
    },
    {
      "epoch": 0.45412529955494696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4418,
      "step": 13265
    },
    {
      "epoch": 0.45415953440602536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4428,
      "step": 13266
    },
    {
      "epoch": 0.45419376925710375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4265,
      "step": 13267
    },
    {
      "epoch": 0.45422800410818215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4352,
      "step": 13268
    },
    {
      "epoch": 0.45426223895926054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.419,
      "step": 13269
    },
    {
      "epoch": 0.45429647381033894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4177,
      "step": 13270
    },
    {
      "epoch": 0.45433070866141734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4116,
      "step": 13271
    },
    {
      "epoch": 0.45436494351249573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4033,
      "step": 13272
    },
    {
      "epoch": 0.45439917836357413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4191,
      "step": 13273
    },
    {
      "epoch": 0.4544334132146525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4025,
      "step": 13274
    },
    {
      "epoch": 0.4544676480657309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4135,
      "step": 13275
    },
    {
      "epoch": 0.4545018829168093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4106,
      "step": 13276
    },
    {
      "epoch": 0.4545361177678877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4019,
      "step": 13277
    },
    {
      "epoch": 0.4545703526189661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4027,
      "step": 13278
    },
    {
      "epoch": 0.4546045874700445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4342,
      "step": 13279
    },
    {
      "epoch": 0.4546388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4177,
      "step": 13280
    },
    {
      "epoch": 0.4546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4083,
      "step": 13281
    },
    {
      "epoch": 0.4547072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4056,
      "step": 13282
    },
    {
      "epoch": 0.4547415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4002,
      "step": 13283
    },
    {
      "epoch": 0.4547757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4088,
      "step": 13284
    },
    {
      "epoch": 0.4548099965765149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4125,
      "step": 13285
    },
    {
      "epoch": 0.4548442314275933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4056,
      "step": 13286
    },
    {
      "epoch": 0.4548784662786717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4032,
      "step": 13287
    },
    {
      "epoch": 0.4549127011297501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4113,
      "step": 13288
    },
    {
      "epoch": 0.45494693598082847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4192,
      "step": 13289
    },
    {
      "epoch": 0.45498117083190687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.412,
      "step": 13290
    },
    {
      "epoch": 0.45501540568298526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4152,
      "step": 13291
    },
    {
      "epoch": 0.45504964053406366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4121,
      "step": 13292
    },
    {
      "epoch": 0.45508387538514206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4176,
      "step": 13293
    },
    {
      "epoch": 0.45511811023622045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3912,
      "step": 13294
    },
    {
      "epoch": 0.45515234508729885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4047,
      "step": 13295
    },
    {
      "epoch": 0.45518657993837724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4074,
      "step": 13296
    },
    {
      "epoch": 0.45522081478945564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4014,
      "step": 13297
    },
    {
      "epoch": 0.45525504964053404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3865,
      "step": 13298
    },
    {
      "epoch": 0.45528928449161243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3888,
      "step": 13299
    },
    {
      "epoch": 0.4553235193426909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3823,
      "step": 13300
    },
    {
      "epoch": 0.4553577541937693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4028,
      "step": 13301
    },
    {
      "epoch": 0.4553919890448477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4113,
      "step": 13302
    },
    {
      "epoch": 0.4554262238959261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4136,
      "step": 13303
    },
    {
      "epoch": 0.45546045874700447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4127,
      "step": 13304
    },
    {
      "epoch": 0.45549469359808287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3853,
      "step": 13305
    },
    {
      "epoch": 0.45552892844916126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4166,
      "step": 13306
    },
    {
      "epoch": 0.45556316330023966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4128,
      "step": 13307
    },
    {
      "epoch": 0.45559739815131806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4074,
      "step": 13308
    },
    {
      "epoch": 0.45563163300239645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4189,
      "step": 13309
    },
    {
      "epoch": 0.45566586785347485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4218,
      "step": 13310
    },
    {
      "epoch": 0.45570010270455324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4058,
      "step": 13311
    },
    {
      "epoch": 0.45573433755563164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4058,
      "step": 13312
    },
    {
      "epoch": 0.45576857240671004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4157,
      "step": 13313
    },
    {
      "epoch": 0.45580280725778843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4095,
      "step": 13314
    },
    {
      "epoch": 0.45583704210886683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4287,
      "step": 13315
    },
    {
      "epoch": 0.4558712769599452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4102,
      "step": 13316
    },
    {
      "epoch": 0.4559055118110236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4056,
      "step": 13317
    },
    {
      "epoch": 0.455939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4291,
      "step": 13318
    },
    {
      "epoch": 0.4559739815131804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4039,
      "step": 13319
    },
    {
      "epoch": 0.4560082163642588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4313,
      "step": 13320
    },
    {
      "epoch": 0.4560424512153372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4193,
      "step": 13321
    },
    {
      "epoch": 0.4560766860664156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4109,
      "step": 13322
    },
    {
      "epoch": 0.456110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4042,
      "step": 13323
    },
    {
      "epoch": 0.4561451557685724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3965,
      "step": 13324
    },
    {
      "epoch": 0.4561793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.42,
      "step": 13325
    },
    {
      "epoch": 0.4562136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4184,
      "step": 13326
    },
    {
      "epoch": 0.4562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4262,
      "step": 13327
    },
    {
      "epoch": 0.456282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4139,
      "step": 13328
    },
    {
      "epoch": 0.4563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.422,
      "step": 13329
    },
    {
      "epoch": 0.4563505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4127,
      "step": 13330
    },
    {
      "epoch": 0.45638479972612117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4038,
      "step": 13331
    },
    {
      "epoch": 0.45641903457719957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4189,
      "step": 13332
    },
    {
      "epoch": 0.45645326942827796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3956,
      "step": 13333
    },
    {
      "epoch": 0.45648750427935636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3958,
      "step": 13334
    },
    {
      "epoch": 0.45652173913043476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4177,
      "step": 13335
    },
    {
      "epoch": 0.4565559739815132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4016,
      "step": 13336
    },
    {
      "epoch": 0.4565902088325916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4411,
      "step": 13337
    },
    {
      "epoch": 0.45662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.389,
      "step": 13338
    },
    {
      "epoch": 0.4566586785347484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4062,
      "step": 13339
    },
    {
      "epoch": 0.4566929133858268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4215,
      "step": 13340
    },
    {
      "epoch": 0.4567271482369052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4326,
      "step": 13341
    },
    {
      "epoch": 0.4567613830879836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4002,
      "step": 13342
    },
    {
      "epoch": 0.456795617939062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4105,
      "step": 13343
    },
    {
      "epoch": 0.4568298527901404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4029,
      "step": 13344
    },
    {
      "epoch": 0.4568640876412188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.397,
      "step": 13345
    },
    {
      "epoch": 0.45689832249229717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.389,
      "step": 13346
    },
    {
      "epoch": 0.45693255734337557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.412,
      "step": 13347
    },
    {
      "epoch": 0.45696679219445396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3841,
      "step": 13348
    },
    {
      "epoch": 0.45700102704553236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4223,
      "step": 13349
    },
    {
      "epoch": 0.45703526189661076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3981,
      "step": 13350
    },
    {
      "epoch": 0.45706949674768915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4107,
      "step": 13351
    },
    {
      "epoch": 0.45710373159876755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4143,
      "step": 13352
    },
    {
      "epoch": 0.45713796644984594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4008,
      "step": 13353
    },
    {
      "epoch": 0.45717220130092434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4085,
      "step": 13354
    },
    {
      "epoch": 0.45720643615200274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4055,
      "step": 13355
    },
    {
      "epoch": 0.45724067100308113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4279,
      "step": 13356
    },
    {
      "epoch": 0.45727490585415953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4177,
      "step": 13357
    },
    {
      "epoch": 0.4573091407052379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4197,
      "step": 13358
    },
    {
      "epoch": 0.4573433755563163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4154,
      "step": 13359
    },
    {
      "epoch": 0.4573776104073947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4136,
      "step": 13360
    },
    {
      "epoch": 0.4574118452584731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4061,
      "step": 13361
    },
    {
      "epoch": 0.4574460801095515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4238,
      "step": 13362
    },
    {
      "epoch": 0.4574803149606299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4073,
      "step": 13363
    },
    {
      "epoch": 0.4575145498117083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4541,
      "step": 13364
    },
    {
      "epoch": 0.4575487846627867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.421,
      "step": 13365
    },
    {
      "epoch": 0.4575830195138651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4016,
      "step": 13366
    },
    {
      "epoch": 0.4576172543649435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3913,
      "step": 13367
    },
    {
      "epoch": 0.4576514892160219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4137,
      "step": 13368
    },
    {
      "epoch": 0.4576857240671003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3952,
      "step": 13369
    },
    {
      "epoch": 0.4577199589181787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3884,
      "step": 13370
    },
    {
      "epoch": 0.4577541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4076,
      "step": 13371
    },
    {
      "epoch": 0.4577884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4022,
      "step": 13372
    },
    {
      "epoch": 0.4578226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4195,
      "step": 13373
    },
    {
      "epoch": 0.4578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4,
      "step": 13374
    },
    {
      "epoch": 0.4578911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4221,
      "step": 13375
    },
    {
      "epoch": 0.4579253680246491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3981,
      "step": 13376
    },
    {
      "epoch": 0.4579596028757275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4182,
      "step": 13377
    },
    {
      "epoch": 0.4579938377268059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4099,
      "step": 13378
    },
    {
      "epoch": 0.4580280725778843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4037,
      "step": 13379
    },
    {
      "epoch": 0.4580623074289627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3944,
      "step": 13380
    },
    {
      "epoch": 0.4580965422800411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3908,
      "step": 13381
    },
    {
      "epoch": 0.4581307771311195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4146,
      "step": 13382
    },
    {
      "epoch": 0.4581650119821979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3989,
      "step": 13383
    },
    {
      "epoch": 0.4581992468332763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4017,
      "step": 13384
    },
    {
      "epoch": 0.4582334816843547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4269,
      "step": 13385
    },
    {
      "epoch": 0.4582677165354331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4089,
      "step": 13386
    },
    {
      "epoch": 0.4583019513865115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4009,
      "step": 13387
    },
    {
      "epoch": 0.45833618623758987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4136,
      "step": 13388
    },
    {
      "epoch": 0.45837042108866827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4142,
      "step": 13389
    },
    {
      "epoch": 0.45840465593974666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4032,
      "step": 13390
    },
    {
      "epoch": 0.45843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4045,
      "step": 13391
    },
    {
      "epoch": 0.45847312564190346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4222,
      "step": 13392
    },
    {
      "epoch": 0.45850736049298185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4135,
      "step": 13393
    },
    {
      "epoch": 0.45854159534406025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4073,
      "step": 13394
    },
    {
      "epoch": 0.45857583019513865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4034,
      "step": 13395
    },
    {
      "epoch": 0.45861006504621704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4121,
      "step": 13396
    },
    {
      "epoch": 0.45864429989729544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4055,
      "step": 13397
    },
    {
      "epoch": 0.45867853474837383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4174,
      "step": 13398
    },
    {
      "epoch": 0.45871276959945223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3967,
      "step": 13399
    },
    {
      "epoch": 0.4587470044505306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4011,
      "step": 13400
    },
    {
      "epoch": 0.458781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4018,
      "step": 13401
    },
    {
      "epoch": 0.4588154741526874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.391,
      "step": 13402
    },
    {
      "epoch": 0.4588497090037658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4131,
      "step": 13403
    },
    {
      "epoch": 0.4588839438548442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4163,
      "step": 13404
    },
    {
      "epoch": 0.4589181787059226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4165,
      "step": 13405
    },
    {
      "epoch": 0.458952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4036,
      "step": 13406
    },
    {
      "epoch": 0.4589866484080794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4097,
      "step": 13407
    },
    {
      "epoch": 0.4590208832591578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4172,
      "step": 13408
    },
    {
      "epoch": 0.4590551181102362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3948,
      "step": 13409
    },
    {
      "epoch": 0.45908935296131465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4085,
      "step": 13410
    },
    {
      "epoch": 0.45912358781239304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4092,
      "step": 13411
    },
    {
      "epoch": 0.45915782266347144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4094,
      "step": 13412
    },
    {
      "epoch": 0.45919205751454983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3834,
      "step": 13413
    },
    {
      "epoch": 0.45922629236562823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4175,
      "step": 13414
    },
    {
      "epoch": 0.4592605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4064,
      "step": 13415
    },
    {
      "epoch": 0.459294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3994,
      "step": 13416
    },
    {
      "epoch": 0.4593289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4028,
      "step": 13417
    },
    {
      "epoch": 0.4593632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3958,
      "step": 13418
    },
    {
      "epoch": 0.4593974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4054,
      "step": 13419
    },
    {
      "epoch": 0.4594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4091,
      "step": 13420
    },
    {
      "epoch": 0.459465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4025,
      "step": 13421
    },
    {
      "epoch": 0.4595001711742554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4133,
      "step": 13422
    },
    {
      "epoch": 0.4595344060253338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4132,
      "step": 13423
    },
    {
      "epoch": 0.4595686408764122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4005,
      "step": 13424
    },
    {
      "epoch": 0.4596028757274906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3903,
      "step": 13425
    },
    {
      "epoch": 0.459637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4035,
      "step": 13426
    },
    {
      "epoch": 0.4596713454296474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4124,
      "step": 13427
    },
    {
      "epoch": 0.4597055802807258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4071,
      "step": 13428
    },
    {
      "epoch": 0.4597398151318042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4274,
      "step": 13429
    },
    {
      "epoch": 0.45977404998288257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4038,
      "step": 13430
    },
    {
      "epoch": 0.45980828483396097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3936,
      "step": 13431
    },
    {
      "epoch": 0.45984251968503936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4138,
      "step": 13432
    },
    {
      "epoch": 0.45987675453611776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4028,
      "step": 13433
    },
    {
      "epoch": 0.45991098938719616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.41,
      "step": 13434
    },
    {
      "epoch": 0.45994522423827455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4128,
      "step": 13435
    },
    {
      "epoch": 0.45997945908935295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4163,
      "step": 13436
    },
    {
      "epoch": 0.46001369394043135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.419,
      "step": 13437
    },
    {
      "epoch": 0.46004792879150974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4123,
      "step": 13438
    },
    {
      "epoch": 0.46008216364258814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4086,
      "step": 13439
    },
    {
      "epoch": 0.46011639849366653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4229,
      "step": 13440
    },
    {
      "epoch": 0.46015063334474493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4051,
      "step": 13441
    },
    {
      "epoch": 0.4601848681958233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.421,
      "step": 13442
    },
    {
      "epoch": 0.4602191030469017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4243,
      "step": 13443
    },
    {
      "epoch": 0.4602533378979801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4085,
      "step": 13444
    },
    {
      "epoch": 0.4602875727490585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4257,
      "step": 13445
    },
    {
      "epoch": 0.4603218076001369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3981,
      "step": 13446
    },
    {
      "epoch": 0.46035604245121536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4116,
      "step": 13447
    },
    {
      "epoch": 0.46039027730229376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4069,
      "step": 13448
    },
    {
      "epoch": 0.46042451215337216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4048,
      "step": 13449
    },
    {
      "epoch": 0.46045874700445055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4111,
      "step": 13450
    },
    {
      "epoch": 0.46049298185552895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4117,
      "step": 13451
    },
    {
      "epoch": 0.46052721670660735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4282,
      "step": 13452
    },
    {
      "epoch": 0.46056145155768574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4044,
      "step": 13453
    },
    {
      "epoch": 0.46059568640876414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4039,
      "step": 13454
    },
    {
      "epoch": 0.46062992125984253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3867,
      "step": 13455
    },
    {
      "epoch": 0.46066415611092093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.418,
      "step": 13456
    },
    {
      "epoch": 0.4606983909619993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3973,
      "step": 13457
    },
    {
      "epoch": 0.4607326258130777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4012,
      "step": 13458
    },
    {
      "epoch": 0.4607668606641561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4011,
      "step": 13459
    },
    {
      "epoch": 0.4608010955152345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3867,
      "step": 13460
    },
    {
      "epoch": 0.4608353303663129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4031,
      "step": 13461
    },
    {
      "epoch": 0.4608695652173913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3997,
      "step": 13462
    },
    {
      "epoch": 0.4609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3871,
      "step": 13463
    },
    {
      "epoch": 0.4609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4015,
      "step": 13464
    },
    {
      "epoch": 0.4609722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4265,
      "step": 13465
    },
    {
      "epoch": 0.4610065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3948,
      "step": 13466
    },
    {
      "epoch": 0.4610407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3947,
      "step": 13467
    },
    {
      "epoch": 0.4610749743238617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4075,
      "step": 13468
    },
    {
      "epoch": 0.4611092091749401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4222,
      "step": 13469
    },
    {
      "epoch": 0.4611434440260185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3928,
      "step": 13470
    },
    {
      "epoch": 0.4611776788770969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4022,
      "step": 13471
    },
    {
      "epoch": 0.46121191372817527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4158,
      "step": 13472
    },
    {
      "epoch": 0.46124614857925367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4061,
      "step": 13473
    },
    {
      "epoch": 0.46128038343033206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4247,
      "step": 13474
    },
    {
      "epoch": 0.46131461828141046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4007,
      "step": 13475
    },
    {
      "epoch": 0.46134885313248886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4094,
      "step": 13476
    },
    {
      "epoch": 0.46138308798356725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4075,
      "step": 13477
    },
    {
      "epoch": 0.46141732283464565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4244,
      "step": 13478
    },
    {
      "epoch": 0.46145155768572405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4209,
      "step": 13479
    },
    {
      "epoch": 0.46148579253680244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4059,
      "step": 13480
    },
    {
      "epoch": 0.46152002738788084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3933,
      "step": 13481
    },
    {
      "epoch": 0.46155426223895923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4033,
      "step": 13482
    },
    {
      "epoch": 0.4615884970900377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.404,
      "step": 13483
    },
    {
      "epoch": 0.4616227319411161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4139,
      "step": 13484
    },
    {
      "epoch": 0.4616569667921945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4036,
      "step": 13485
    },
    {
      "epoch": 0.4616912016432729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3997,
      "step": 13486
    },
    {
      "epoch": 0.46172543649435127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4014,
      "step": 13487
    },
    {
      "epoch": 0.46175967134542967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4197,
      "step": 13488
    },
    {
      "epoch": 0.46179390619650806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4044,
      "step": 13489
    },
    {
      "epoch": 0.46182814104758646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.404,
      "step": 13490
    },
    {
      "epoch": 0.46186237589866486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4182,
      "step": 13491
    },
    {
      "epoch": 0.46189661074974325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4199,
      "step": 13492
    },
    {
      "epoch": 0.46193084560082165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4193,
      "step": 13493
    },
    {
      "epoch": 0.46196508045190005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4154,
      "step": 13494
    },
    {
      "epoch": 0.46199931530297844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3926,
      "step": 13495
    },
    {
      "epoch": 0.46203355015405684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4226,
      "step": 13496
    },
    {
      "epoch": 0.46206778500513523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3829,
      "step": 13497
    },
    {
      "epoch": 0.46210201985621363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4059,
      "step": 13498
    },
    {
      "epoch": 0.462136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4094,
      "step": 13499
    },
    {
      "epoch": 0.4621704895583704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3941,
      "step": 13500
    },
    {
      "epoch": 0.4622047244094488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3836,
      "step": 13501
    },
    {
      "epoch": 0.4622389592605272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3904,
      "step": 13502
    },
    {
      "epoch": 0.4622731941116056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4073,
      "step": 13503
    },
    {
      "epoch": 0.462307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4043,
      "step": 13504
    },
    {
      "epoch": 0.4623416638137624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4101,
      "step": 13505
    },
    {
      "epoch": 0.4623758986648408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4079,
      "step": 13506
    },
    {
      "epoch": 0.4624101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4121,
      "step": 13507
    },
    {
      "epoch": 0.4624443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3955,
      "step": 13508
    },
    {
      "epoch": 0.462478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.375,
      "step": 13509
    },
    {
      "epoch": 0.4625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3922,
      "step": 13510
    },
    {
      "epoch": 0.4625470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4005,
      "step": 13511
    },
    {
      "epoch": 0.4625813077713112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.417,
      "step": 13512
    },
    {
      "epoch": 0.4626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4132,
      "step": 13513
    },
    {
      "epoch": 0.46264977747346797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3985,
      "step": 13514
    },
    {
      "epoch": 0.46268401232454637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4174,
      "step": 13515
    },
    {
      "epoch": 0.46271824717562476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.409,
      "step": 13516
    },
    {
      "epoch": 0.46275248202670316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4038,
      "step": 13517
    },
    {
      "epoch": 0.46278671687778156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3974,
      "step": 13518
    },
    {
      "epoch": 0.46282095172885995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4003,
      "step": 13519
    },
    {
      "epoch": 0.4628551865799384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4063,
      "step": 13520
    },
    {
      "epoch": 0.4628894214310168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3984,
      "step": 13521
    },
    {
      "epoch": 0.4629236562820952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4239,
      "step": 13522
    },
    {
      "epoch": 0.4629578911331736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.418,
      "step": 13523
    },
    {
      "epoch": 0.462992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4128,
      "step": 13524
    },
    {
      "epoch": 0.4630263608353304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.411,
      "step": 13525
    },
    {
      "epoch": 0.4630605956864088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4135,
      "step": 13526
    },
    {
      "epoch": 0.4630948305374872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4235,
      "step": 13527
    },
    {
      "epoch": 0.4631290653885656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4259,
      "step": 13528
    },
    {
      "epoch": 0.46316330023964397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4056,
      "step": 13529
    },
    {
      "epoch": 0.46319753509072237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4187,
      "step": 13530
    },
    {
      "epoch": 0.46323176994180076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4078,
      "step": 13531
    },
    {
      "epoch": 0.46326600479287916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4064,
      "step": 13532
    },
    {
      "epoch": 0.46330023964395756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3896,
      "step": 13533
    },
    {
      "epoch": 0.46333447449503595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3859,
      "step": 13534
    },
    {
      "epoch": 0.46336870934611435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4113,
      "step": 13535
    },
    {
      "epoch": 0.46340294419719275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3944,
      "step": 13536
    },
    {
      "epoch": 0.46343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4085,
      "step": 13537
    },
    {
      "epoch": 0.46347141389934954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4211,
      "step": 13538
    },
    {
      "epoch": 0.46350564875042793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4139,
      "step": 13539
    },
    {
      "epoch": 0.46353988360150633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3983,
      "step": 13540
    },
    {
      "epoch": 0.4635741184525847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.404,
      "step": 13541
    },
    {
      "epoch": 0.4636083533036631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4186,
      "step": 13542
    },
    {
      "epoch": 0.4636425881547415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4003,
      "step": 13543
    },
    {
      "epoch": 0.4636768230058199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3963,
      "step": 13544
    },
    {
      "epoch": 0.4637110578568983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.39,
      "step": 13545
    },
    {
      "epoch": 0.4637452927079767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3799,
      "step": 13546
    },
    {
      "epoch": 0.4637795275590551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3711,
      "step": 13547
    },
    {
      "epoch": 0.4638137624101335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3959,
      "step": 13548
    },
    {
      "epoch": 0.4638479972612119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4171,
      "step": 13549
    },
    {
      "epoch": 0.4638822321122903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4128,
      "step": 13550
    },
    {
      "epoch": 0.4639164669633687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.41,
      "step": 13551
    },
    {
      "epoch": 0.4639507018144471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3908,
      "step": 13552
    },
    {
      "epoch": 0.4639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4149,
      "step": 13553
    },
    {
      "epoch": 0.4640191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4086,
      "step": 13554
    },
    {
      "epoch": 0.4640534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3839,
      "step": 13555
    },
    {
      "epoch": 0.46408764121876067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4083,
      "step": 13556
    },
    {
      "epoch": 0.4641218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4186,
      "step": 13557
    },
    {
      "epoch": 0.4641561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4005,
      "step": 13558
    },
    {
      "epoch": 0.4641903457719959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3889,
      "step": 13559
    },
    {
      "epoch": 0.4642245806230743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4192,
      "step": 13560
    },
    {
      "epoch": 0.4642588154741527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3997,
      "step": 13561
    },
    {
      "epoch": 0.4642930503252311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.396,
      "step": 13562
    },
    {
      "epoch": 0.4643272851763095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4308,
      "step": 13563
    },
    {
      "epoch": 0.4643615200273879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4128,
      "step": 13564
    },
    {
      "epoch": 0.4643957548784663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4169,
      "step": 13565
    },
    {
      "epoch": 0.4644299897295447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.387,
      "step": 13566
    },
    {
      "epoch": 0.4644642245806231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3997,
      "step": 13567
    },
    {
      "epoch": 0.4644984594317015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4213,
      "step": 13568
    },
    {
      "epoch": 0.4645326942827799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4127,
      "step": 13569
    },
    {
      "epoch": 0.4645669291338583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3966,
      "step": 13570
    },
    {
      "epoch": 0.46460116398493667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4168,
      "step": 13571
    },
    {
      "epoch": 0.46463539883601507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4175,
      "step": 13572
    },
    {
      "epoch": 0.46466963368709346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4256,
      "step": 13573
    },
    {
      "epoch": 0.46470386853817186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3949,
      "step": 13574
    },
    {
      "epoch": 0.46473810338925026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4094,
      "step": 13575
    },
    {
      "epoch": 0.46477233824032865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4056,
      "step": 13576
    },
    {
      "epoch": 0.46480657309140705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3912,
      "step": 13577
    },
    {
      "epoch": 0.46484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.392,
      "step": 13578
    },
    {
      "epoch": 0.46487504279356384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3931,
      "step": 13579
    },
    {
      "epoch": 0.46490927764464224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4049,
      "step": 13580
    },
    {
      "epoch": 0.46494351249572063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4004,
      "step": 13581
    },
    {
      "epoch": 0.46497774734679903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4183,
      "step": 13582
    },
    {
      "epoch": 0.4650119821978774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3991,
      "step": 13583
    },
    {
      "epoch": 0.4650462170489558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3997,
      "step": 13584
    },
    {
      "epoch": 0.4650804519000342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4037,
      "step": 13585
    },
    {
      "epoch": 0.4651146867511126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4092,
      "step": 13586
    },
    {
      "epoch": 0.465148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4049,
      "step": 13587
    },
    {
      "epoch": 0.4651831564532694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4142,
      "step": 13588
    },
    {
      "epoch": 0.4652173913043478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4161,
      "step": 13589
    },
    {
      "epoch": 0.4652516261554262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3997,
      "step": 13590
    },
    {
      "epoch": 0.4652858610065046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3771,
      "step": 13591
    },
    {
      "epoch": 0.465320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4059,
      "step": 13592
    },
    {
      "epoch": 0.4653543307086614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4031,
      "step": 13593
    },
    {
      "epoch": 0.46538856555973984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3996,
      "step": 13594
    },
    {
      "epoch": 0.46542280041081824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.394,
      "step": 13595
    },
    {
      "epoch": 0.46545703526189663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4012,
      "step": 13596
    },
    {
      "epoch": 0.46549127011297503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4105,
      "step": 13597
    },
    {
      "epoch": 0.4655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3992,
      "step": 13598
    },
    {
      "epoch": 0.4655597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3997,
      "step": 13599
    },
    {
      "epoch": 0.4655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.408,
      "step": 13600
    },
    {
      "epoch": 0.4656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4162,
      "step": 13601
    },
    {
      "epoch": 0.465662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3998,
      "step": 13602
    },
    {
      "epoch": 0.4656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3967,
      "step": 13603
    },
    {
      "epoch": 0.4657309140705238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4129,
      "step": 13604
    },
    {
      "epoch": 0.4657651489216022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4032,
      "step": 13605
    },
    {
      "epoch": 0.4657993837726806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3975,
      "step": 13606
    },
    {
      "epoch": 0.465833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.386,
      "step": 13607
    },
    {
      "epoch": 0.4658678534748374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3914,
      "step": 13608
    },
    {
      "epoch": 0.4659020883259158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4059,
      "step": 13609
    },
    {
      "epoch": 0.4659363231769942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4038,
      "step": 13610
    },
    {
      "epoch": 0.4659705580280726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3972,
      "step": 13611
    },
    {
      "epoch": 0.466004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4004,
      "step": 13612
    },
    {
      "epoch": 0.46603902773022937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.416,
      "step": 13613
    },
    {
      "epoch": 0.46607326258130777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4099,
      "step": 13614
    },
    {
      "epoch": 0.46610749743238616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.396,
      "step": 13615
    },
    {
      "epoch": 0.46614173228346456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3856,
      "step": 13616
    },
    {
      "epoch": 0.46617596713454296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4003,
      "step": 13617
    },
    {
      "epoch": 0.46621020198562135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3923,
      "step": 13618
    },
    {
      "epoch": 0.46624443683669975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4023,
      "step": 13619
    },
    {
      "epoch": 0.46627867168777815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4295,
      "step": 13620
    },
    {
      "epoch": 0.46631290653885654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4064,
      "step": 13621
    },
    {
      "epoch": 0.46634714138993494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4039,
      "step": 13622
    },
    {
      "epoch": 0.46638137624101333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3966,
      "step": 13623
    },
    {
      "epoch": 0.46641561109209173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3974,
      "step": 13624
    },
    {
      "epoch": 0.4664498459431701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4181,
      "step": 13625
    },
    {
      "epoch": 0.4664840807942485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4083,
      "step": 13626
    },
    {
      "epoch": 0.4665183156453269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4079,
      "step": 13627
    },
    {
      "epoch": 0.4665525504964053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3939,
      "step": 13628
    },
    {
      "epoch": 0.4665867853474837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4241,
      "step": 13629
    },
    {
      "epoch": 0.4666210201985621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3936,
      "step": 13630
    },
    {
      "epoch": 0.46665525504964056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4139,
      "step": 13631
    },
    {
      "epoch": 0.46668948990071896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4,
      "step": 13632
    },
    {
      "epoch": 0.46672372475179735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.412,
      "step": 13633
    },
    {
      "epoch": 0.46675795960287575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4128,
      "step": 13634
    },
    {
      "epoch": 0.46679219445395415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3794,
      "step": 13635
    },
    {
      "epoch": 0.46682642930503254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.402,
      "step": 13636
    },
    {
      "epoch": 0.46686066415611094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4063,
      "step": 13637
    },
    {
      "epoch": 0.46689489900718933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4092,
      "step": 13638
    },
    {
      "epoch": 0.46692913385826773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3902,
      "step": 13639
    },
    {
      "epoch": 0.4669633687093461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4038,
      "step": 13640
    },
    {
      "epoch": 0.4669976035604245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4059,
      "step": 13641
    },
    {
      "epoch": 0.4670318384115029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3956,
      "step": 13642
    },
    {
      "epoch": 0.4670660732625813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3883,
      "step": 13643
    },
    {
      "epoch": 0.4671003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4132,
      "step": 13644
    },
    {
      "epoch": 0.4671345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.414,
      "step": 13645
    },
    {
      "epoch": 0.4671687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3939,
      "step": 13646
    },
    {
      "epoch": 0.4672030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3937,
      "step": 13647
    },
    {
      "epoch": 0.4672372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3959,
      "step": 13648
    },
    {
      "epoch": 0.4672714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4028,
      "step": 13649
    },
    {
      "epoch": 0.4673057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4025,
      "step": 13650
    },
    {
      "epoch": 0.4673399520712085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4297,
      "step": 13651
    },
    {
      "epoch": 0.4673741869222869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.418,
      "step": 13652
    },
    {
      "epoch": 0.4674084217733653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4027,
      "step": 13653
    },
    {
      "epoch": 0.4674426566244437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4053,
      "step": 13654
    },
    {
      "epoch": 0.46747689147552207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4071,
      "step": 13655
    },
    {
      "epoch": 0.46751112632660047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3892,
      "step": 13656
    },
    {
      "epoch": 0.46754536117767886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4088,
      "step": 13657
    },
    {
      "epoch": 0.46757959602875726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4219,
      "step": 13658
    },
    {
      "epoch": 0.46761383087983566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4124,
      "step": 13659
    },
    {
      "epoch": 0.46764806573091405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.382,
      "step": 13660
    },
    {
      "epoch": 0.46768230058199245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3877,
      "step": 13661
    },
    {
      "epoch": 0.46771653543307085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4027,
      "step": 13662
    },
    {
      "epoch": 0.46775077028414924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3962,
      "step": 13663
    },
    {
      "epoch": 0.46778500513522764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3886,
      "step": 13664
    },
    {
      "epoch": 0.46781923998630603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3964,
      "step": 13665
    },
    {
      "epoch": 0.46785347483738443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4207,
      "step": 13666
    },
    {
      "epoch": 0.4678877096884629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4177,
      "step": 13667
    },
    {
      "epoch": 0.4679219445395413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3955,
      "step": 13668
    },
    {
      "epoch": 0.4679561793906197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.41,
      "step": 13669
    },
    {
      "epoch": 0.46799041424169807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4203,
      "step": 13670
    },
    {
      "epoch": 0.46802464909277647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4026,
      "step": 13671
    },
    {
      "epoch": 0.46805888394385486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4133,
      "step": 13672
    },
    {
      "epoch": 0.46809311879493326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4124,
      "step": 13673
    },
    {
      "epoch": 0.46812735364601166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3866,
      "step": 13674
    },
    {
      "epoch": 0.46816158849709005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3917,
      "step": 13675
    },
    {
      "epoch": 0.46819582334816845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4008,
      "step": 13676
    },
    {
      "epoch": 0.46823005819924685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3801,
      "step": 13677
    },
    {
      "epoch": 0.46826429305032524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4088,
      "step": 13678
    },
    {
      "epoch": 0.46829852790140364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4041,
      "step": 13679
    },
    {
      "epoch": 0.46833276275248203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3983,
      "step": 13680
    },
    {
      "epoch": 0.46836699760356043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3897,
      "step": 13681
    },
    {
      "epoch": 0.4684012324546388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3951,
      "step": 13682
    },
    {
      "epoch": 0.4684354673057172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4044,
      "step": 13683
    },
    {
      "epoch": 0.4684697021567956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3953,
      "step": 13684
    },
    {
      "epoch": 0.468503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3986,
      "step": 13685
    },
    {
      "epoch": 0.4685381718589524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4099,
      "step": 13686
    },
    {
      "epoch": 0.4685724067100308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3818,
      "step": 13687
    },
    {
      "epoch": 0.4686066415611092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3956,
      "step": 13688
    },
    {
      "epoch": 0.4686408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4115,
      "step": 13689
    },
    {
      "epoch": 0.468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4102,
      "step": 13690
    },
    {
      "epoch": 0.4687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3964,
      "step": 13691
    },
    {
      "epoch": 0.4687435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4163,
      "step": 13692
    },
    {
      "epoch": 0.4687778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3824,
      "step": 13693
    },
    {
      "epoch": 0.4688120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3908,
      "step": 13694
    },
    {
      "epoch": 0.468846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4005,
      "step": 13695
    },
    {
      "epoch": 0.4688805203697364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4002,
      "step": 13696
    },
    {
      "epoch": 0.46891475522081477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.392,
      "step": 13697
    },
    {
      "epoch": 0.46894899007189317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4041,
      "step": 13698
    },
    {
      "epoch": 0.46898322492297156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4106,
      "step": 13699
    },
    {
      "epoch": 0.46901745977404996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3854,
      "step": 13700
    },
    {
      "epoch": 0.46905169462512836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3881,
      "step": 13701
    },
    {
      "epoch": 0.46908592947620675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4001,
      "step": 13702
    },
    {
      "epoch": 0.46912016432728515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.393,
      "step": 13703
    },
    {
      "epoch": 0.4691543991783636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4031,
      "step": 13704
    },
    {
      "epoch": 0.469188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4012,
      "step": 13705
    },
    {
      "epoch": 0.4692228688805204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4036,
      "step": 13706
    },
    {
      "epoch": 0.4692571037315988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3989,
      "step": 13707
    },
    {
      "epoch": 0.4692913385826772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4006,
      "step": 13708
    },
    {
      "epoch": 0.4693255734337556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3824,
      "step": 13709
    },
    {
      "epoch": 0.469359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4137,
      "step": 13710
    },
    {
      "epoch": 0.4693940431359124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3939,
      "step": 13711
    },
    {
      "epoch": 0.46942827798699077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.405,
      "step": 13712
    },
    {
      "epoch": 0.46946251283806917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3919,
      "step": 13713
    },
    {
      "epoch": 0.46949674768914756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.409,
      "step": 13714
    },
    {
      "epoch": 0.46953098254022596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4063,
      "step": 13715
    },
    {
      "epoch": 0.46956521739130436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.384,
      "step": 13716
    },
    {
      "epoch": 0.46959945224238275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4019,
      "step": 13717
    },
    {
      "epoch": 0.46963368709346115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4133,
      "step": 13718
    },
    {
      "epoch": 0.46966792194453955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3967,
      "step": 13719
    },
    {
      "epoch": 0.46970215679561794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3829,
      "step": 13720
    },
    {
      "epoch": 0.46973639164669634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3925,
      "step": 13721
    },
    {
      "epoch": 0.46977062649777473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4024,
      "step": 13722
    },
    {
      "epoch": 0.46980486134885313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4373,
      "step": 13723
    },
    {
      "epoch": 0.4698390961999315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4042,
      "step": 13724
    },
    {
      "epoch": 0.4698733310510099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3969,
      "step": 13725
    },
    {
      "epoch": 0.4699075659020883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3975,
      "step": 13726
    },
    {
      "epoch": 0.4699418007531667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4324,
      "step": 13727
    },
    {
      "epoch": 0.4699760356042451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4133,
      "step": 13728
    },
    {
      "epoch": 0.4700102704553235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4089,
      "step": 13729
    },
    {
      "epoch": 0.4700445053064019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4015,
      "step": 13730
    },
    {
      "epoch": 0.4700787401574803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3912,
      "step": 13731
    },
    {
      "epoch": 0.4701129750085587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4118,
      "step": 13732
    },
    {
      "epoch": 0.4701472098596371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3925,
      "step": 13733
    },
    {
      "epoch": 0.4701814447107155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4006,
      "step": 13734
    },
    {
      "epoch": 0.4702156795617939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3934,
      "step": 13735
    },
    {
      "epoch": 0.4702499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3918,
      "step": 13736
    },
    {
      "epoch": 0.4702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3934,
      "step": 13737
    },
    {
      "epoch": 0.4703183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3875,
      "step": 13738
    },
    {
      "epoch": 0.47035261896610747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4053,
      "step": 13739
    },
    {
      "epoch": 0.47038685381718587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3844,
      "step": 13740
    },
    {
      "epoch": 0.4704210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3997,
      "step": 13741
    },
    {
      "epoch": 0.4704553235193427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3914,
      "step": 13742
    },
    {
      "epoch": 0.4704895583704211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4157,
      "step": 13743
    },
    {
      "epoch": 0.4705237932214995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4012,
      "step": 13744
    },
    {
      "epoch": 0.4705580280725779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3948,
      "step": 13745
    },
    {
      "epoch": 0.4705922629236563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4003,
      "step": 13746
    },
    {
      "epoch": 0.4706264977747347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3906,
      "step": 13747
    },
    {
      "epoch": 0.4706607326258131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3991,
      "step": 13748
    },
    {
      "epoch": 0.4706949674768915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4063,
      "step": 13749
    },
    {
      "epoch": 0.4707292023279699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4091,
      "step": 13750
    },
    {
      "epoch": 0.4707634371790483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3931,
      "step": 13751
    },
    {
      "epoch": 0.4707976720301267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.397,
      "step": 13752
    },
    {
      "epoch": 0.4708319068812051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.385,
      "step": 13753
    },
    {
      "epoch": 0.47086614173228347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3944,
      "step": 13754
    },
    {
      "epoch": 0.47090037658336187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4221,
      "step": 13755
    },
    {
      "epoch": 0.47093461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3899,
      "step": 13756
    },
    {
      "epoch": 0.47096884628551866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4049,
      "step": 13757
    },
    {
      "epoch": 0.47100308113659706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3923,
      "step": 13758
    },
    {
      "epoch": 0.47103731598767545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3896,
      "step": 13759
    },
    {
      "epoch": 0.47107155083875385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4039,
      "step": 13760
    },
    {
      "epoch": 0.47110578568983225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3991,
      "step": 13761
    },
    {
      "epoch": 0.47114002054091064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.396,
      "step": 13762
    },
    {
      "epoch": 0.47117425539198904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4061,
      "step": 13763
    },
    {
      "epoch": 0.47120849024306743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3853,
      "step": 13764
    },
    {
      "epoch": 0.47124272509414583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4047,
      "step": 13765
    },
    {
      "epoch": 0.4712769599452242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4027,
      "step": 13766
    },
    {
      "epoch": 0.4713111947963026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3992,
      "step": 13767
    },
    {
      "epoch": 0.471345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3947,
      "step": 13768
    },
    {
      "epoch": 0.4713796644984594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3887,
      "step": 13769
    },
    {
      "epoch": 0.4714138993495378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3843,
      "step": 13770
    },
    {
      "epoch": 0.4714481342006162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3992,
      "step": 13771
    },
    {
      "epoch": 0.4714823690516946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4081,
      "step": 13772
    },
    {
      "epoch": 0.471516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4022,
      "step": 13773
    },
    {
      "epoch": 0.4715508387538514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4239,
      "step": 13774
    },
    {
      "epoch": 0.4715850736049298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.413,
      "step": 13775
    },
    {
      "epoch": 0.4716193084560082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3882,
      "step": 13776
    },
    {
      "epoch": 0.4716535433070866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4003,
      "step": 13777
    },
    {
      "epoch": 0.47168777815816504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.393,
      "step": 13778
    },
    {
      "epoch": 0.47172201300924343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4159,
      "step": 13779
    },
    {
      "epoch": 0.47175624786032183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4065,
      "step": 13780
    },
    {
      "epoch": 0.4717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.384,
      "step": 13781
    },
    {
      "epoch": 0.4718247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4003,
      "step": 13782
    },
    {
      "epoch": 0.471858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3819,
      "step": 13783
    },
    {
      "epoch": 0.4718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3912,
      "step": 13784
    },
    {
      "epoch": 0.4719274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3964,
      "step": 13785
    },
    {
      "epoch": 0.4719616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4166,
      "step": 13786
    },
    {
      "epoch": 0.4719958918178706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3985,
      "step": 13787
    },
    {
      "epoch": 0.472030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4226,
      "step": 13788
    },
    {
      "epoch": 0.4720643615200274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4106,
      "step": 13789
    },
    {
      "epoch": 0.4720985963711058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4001,
      "step": 13790
    },
    {
      "epoch": 0.4721328312221842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.385,
      "step": 13791
    },
    {
      "epoch": 0.4721670660732626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4023,
      "step": 13792
    },
    {
      "epoch": 0.472201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3926,
      "step": 13793
    },
    {
      "epoch": 0.4722355357754194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3934,
      "step": 13794
    },
    {
      "epoch": 0.4722697706264978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3886,
      "step": 13795
    },
    {
      "epoch": 0.47230400547757617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3979,
      "step": 13796
    },
    {
      "epoch": 0.47233824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3944,
      "step": 13797
    },
    {
      "epoch": 0.47237247517973296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.394,
      "step": 13798
    },
    {
      "epoch": 0.47240671003081136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3989,
      "step": 13799
    },
    {
      "epoch": 0.47244094488188976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4136,
      "step": 13800
    },
    {
      "epoch": 0.47247517973296815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4071,
      "step": 13801
    },
    {
      "epoch": 0.47250941458404655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3873,
      "step": 13802
    },
    {
      "epoch": 0.47254364943512495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4172,
      "step": 13803
    },
    {
      "epoch": 0.47257788428620334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3899,
      "step": 13804
    },
    {
      "epoch": 0.47261211913728174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3973,
      "step": 13805
    },
    {
      "epoch": 0.47264635398836014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4024,
      "step": 13806
    },
    {
      "epoch": 0.47268058883943853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3965,
      "step": 13807
    },
    {
      "epoch": 0.4727148236905169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4082,
      "step": 13808
    },
    {
      "epoch": 0.4727490585415953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4079,
      "step": 13809
    },
    {
      "epoch": 0.4727832933926737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4198,
      "step": 13810
    },
    {
      "epoch": 0.4728175282437521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3964,
      "step": 13811
    },
    {
      "epoch": 0.4728517630948305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3787,
      "step": 13812
    },
    {
      "epoch": 0.4728859979459089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3862,
      "step": 13813
    },
    {
      "epoch": 0.4729202327969873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4031,
      "step": 13814
    },
    {
      "epoch": 0.47295446764806576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3964,
      "step": 13815
    },
    {
      "epoch": 0.47298870249914415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4016,
      "step": 13816
    },
    {
      "epoch": 0.47302293735022255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4092,
      "step": 13817
    },
    {
      "epoch": 0.47305717220130095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4095,
      "step": 13818
    },
    {
      "epoch": 0.47309140705237934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4081,
      "step": 13819
    },
    {
      "epoch": 0.47312564190345774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3849,
      "step": 13820
    },
    {
      "epoch": 0.47315987675453614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3984,
      "step": 13821
    },
    {
      "epoch": 0.47319411160561453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.398,
      "step": 13822
    },
    {
      "epoch": 0.4732283464566929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4019,
      "step": 13823
    },
    {
      "epoch": 0.4732625813077713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.384,
      "step": 13824
    },
    {
      "epoch": 0.4732968161588497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3871,
      "step": 13825
    },
    {
      "epoch": 0.4733310510099281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4109,
      "step": 13826
    },
    {
      "epoch": 0.4733652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3999,
      "step": 13827
    },
    {
      "epoch": 0.4733995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3933,
      "step": 13828
    },
    {
      "epoch": 0.4734337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3642,
      "step": 13829
    },
    {
      "epoch": 0.4734679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4015,
      "step": 13830
    },
    {
      "epoch": 0.4735022252653201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3965,
      "step": 13831
    },
    {
      "epoch": 0.4735364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3998,
      "step": 13832
    },
    {
      "epoch": 0.4735706949674769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4147,
      "step": 13833
    },
    {
      "epoch": 0.4736049298185553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3843,
      "step": 13834
    },
    {
      "epoch": 0.4736391646696337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4157,
      "step": 13835
    },
    {
      "epoch": 0.4736733995207121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4108,
      "step": 13836
    },
    {
      "epoch": 0.4737076343717905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3759,
      "step": 13837
    },
    {
      "epoch": 0.47374186922286887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4068,
      "step": 13838
    },
    {
      "epoch": 0.47377610407394727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3604,
      "step": 13839
    },
    {
      "epoch": 0.47381033892502566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.392,
      "step": 13840
    },
    {
      "epoch": 0.47384457377610406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3982,
      "step": 13841
    },
    {
      "epoch": 0.47387880862718246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3894,
      "step": 13842
    },
    {
      "epoch": 0.47391304347826085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3828,
      "step": 13843
    },
    {
      "epoch": 0.47394727832933925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3866,
      "step": 13844
    },
    {
      "epoch": 0.47398151318041765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3842,
      "step": 13845
    },
    {
      "epoch": 0.47401574803149604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4131,
      "step": 13846
    },
    {
      "epoch": 0.47404998288257444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3913,
      "step": 13847
    },
    {
      "epoch": 0.47408421773365284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4065,
      "step": 13848
    },
    {
      "epoch": 0.47411845258473123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4095,
      "step": 13849
    },
    {
      "epoch": 0.4741526874358096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3917,
      "step": 13850
    },
    {
      "epoch": 0.4741869222868881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3806,
      "step": 13851
    },
    {
      "epoch": 0.4742211571379665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4008,
      "step": 13852
    },
    {
      "epoch": 0.47425539198904487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3842,
      "step": 13853
    },
    {
      "epoch": 0.47428962684012327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3874,
      "step": 13854
    },
    {
      "epoch": 0.47432386169120166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3835,
      "step": 13855
    },
    {
      "epoch": 0.47435809654228006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4163,
      "step": 13856
    },
    {
      "epoch": 0.47439233139335846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4016,
      "step": 13857
    },
    {
      "epoch": 0.47442656624443685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4077,
      "step": 13858
    },
    {
      "epoch": 0.47446080109551525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3899,
      "step": 13859
    },
    {
      "epoch": 0.47449503594659365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.414,
      "step": 13860
    },
    {
      "epoch": 0.47452927079767204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3922,
      "step": 13861
    },
    {
      "epoch": 0.47456350564875044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3892,
      "step": 13862
    },
    {
      "epoch": 0.47459774049982884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.405,
      "step": 13863
    },
    {
      "epoch": 0.47463197535090723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3861,
      "step": 13864
    },
    {
      "epoch": 0.4746662102019856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3933,
      "step": 13865
    },
    {
      "epoch": 0.474700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3893,
      "step": 13866
    },
    {
      "epoch": 0.4747346799041424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3904,
      "step": 13867
    },
    {
      "epoch": 0.4747689147552208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4143,
      "step": 13868
    },
    {
      "epoch": 0.4748031496062992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4117,
      "step": 13869
    },
    {
      "epoch": 0.4748373844573776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3776,
      "step": 13870
    },
    {
      "epoch": 0.474871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4083,
      "step": 13871
    },
    {
      "epoch": 0.4749058541595344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3915,
      "step": 13872
    },
    {
      "epoch": 0.4749400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3834,
      "step": 13873
    },
    {
      "epoch": 0.4749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.393,
      "step": 13874
    },
    {
      "epoch": 0.4750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3925,
      "step": 13875
    },
    {
      "epoch": 0.475042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3817,
      "step": 13876
    },
    {
      "epoch": 0.4750770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3922,
      "step": 13877
    },
    {
      "epoch": 0.4751112632660048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3928,
      "step": 13878
    },
    {
      "epoch": 0.4751454981170832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4108,
      "step": 13879
    },
    {
      "epoch": 0.4751797329681616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3972,
      "step": 13880
    },
    {
      "epoch": 0.47521396781923997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4089,
      "step": 13881
    },
    {
      "epoch": 0.47524820267031836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3947,
      "step": 13882
    },
    {
      "epoch": 0.47528243752139676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.373,
      "step": 13883
    },
    {
      "epoch": 0.47531667237247516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3817,
      "step": 13884
    },
    {
      "epoch": 0.47535090722355355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.402,
      "step": 13885
    },
    {
      "epoch": 0.47538514207463195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3892,
      "step": 13886
    },
    {
      "epoch": 0.47541937692571035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3908,
      "step": 13887
    },
    {
      "epoch": 0.4754536117767888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.388,
      "step": 13888
    },
    {
      "epoch": 0.4754878466278672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4055,
      "step": 13889
    },
    {
      "epoch": 0.4755220814789456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3961,
      "step": 13890
    },
    {
      "epoch": 0.475556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3914,
      "step": 13891
    },
    {
      "epoch": 0.4755905511811024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4189,
      "step": 13892
    },
    {
      "epoch": 0.4756247860321808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3915,
      "step": 13893
    },
    {
      "epoch": 0.4756590208832592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4042,
      "step": 13894
    },
    {
      "epoch": 0.4756932557343376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4013,
      "step": 13895
    },
    {
      "epoch": 0.47572749058541597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4072,
      "step": 13896
    },
    {
      "epoch": 0.47576172543649436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4016,
      "step": 13897
    },
    {
      "epoch": 0.47579596028757276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3847,
      "step": 13898
    },
    {
      "epoch": 0.47583019513865116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3958,
      "step": 13899
    },
    {
      "epoch": 0.47586442998972955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4055,
      "step": 13900
    },
    {
      "epoch": 0.47589866484080795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3759,
      "step": 13901
    },
    {
      "epoch": 0.47593289969188635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3823,
      "step": 13902
    },
    {
      "epoch": 0.47596713454296474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4118,
      "step": 13903
    },
    {
      "epoch": 0.47600136939404314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3997,
      "step": 13904
    },
    {
      "epoch": 0.47603560424512154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4023,
      "step": 13905
    },
    {
      "epoch": 0.47606983909619993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4052,
      "step": 13906
    },
    {
      "epoch": 0.47610407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3903,
      "step": 13907
    },
    {
      "epoch": 0.4761383087983567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4168,
      "step": 13908
    },
    {
      "epoch": 0.4761725436494351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3905,
      "step": 13909
    },
    {
      "epoch": 0.4762067785005135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3944,
      "step": 13910
    },
    {
      "epoch": 0.4762410133515919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3647,
      "step": 13911
    },
    {
      "epoch": 0.4762752482026703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3914,
      "step": 13912
    },
    {
      "epoch": 0.4763094830537487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3852,
      "step": 13913
    },
    {
      "epoch": 0.4763437179048271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4058,
      "step": 13914
    },
    {
      "epoch": 0.4763779527559055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3745,
      "step": 13915
    },
    {
      "epoch": 0.4764121876069839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3892,
      "step": 13916
    },
    {
      "epoch": 0.4764464224580623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4097,
      "step": 13917
    },
    {
      "epoch": 0.4764806573091407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4073,
      "step": 13918
    },
    {
      "epoch": 0.4765148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3962,
      "step": 13919
    },
    {
      "epoch": 0.4765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4034,
      "step": 13920
    },
    {
      "epoch": 0.4765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4138,
      "step": 13921
    },
    {
      "epoch": 0.4766175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3953,
      "step": 13922
    },
    {
      "epoch": 0.47665183156453267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3991,
      "step": 13923
    },
    {
      "epoch": 0.47668606641561106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3906,
      "step": 13924
    },
    {
      "epoch": 0.4767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3816,
      "step": 13925
    },
    {
      "epoch": 0.4767545361177679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4054,
      "step": 13926
    },
    {
      "epoch": 0.4767887709688463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3921,
      "step": 13927
    },
    {
      "epoch": 0.4768230058199247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3798,
      "step": 13928
    },
    {
      "epoch": 0.4768572406710031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3893,
      "step": 13929
    },
    {
      "epoch": 0.4768914755220815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3915,
      "step": 13930
    },
    {
      "epoch": 0.4769257103731599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3936,
      "step": 13931
    },
    {
      "epoch": 0.4769599452242383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.392,
      "step": 13932
    },
    {
      "epoch": 0.4769941800753167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3847,
      "step": 13933
    },
    {
      "epoch": 0.4770284149263951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4056,
      "step": 13934
    },
    {
      "epoch": 0.4770626497774735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3724,
      "step": 13935
    },
    {
      "epoch": 0.4770968846285519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4047,
      "step": 13936
    },
    {
      "epoch": 0.4771311194796303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3863,
      "step": 13937
    },
    {
      "epoch": 0.47716535433070867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4056,
      "step": 13938
    },
    {
      "epoch": 0.47719958918178706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3848,
      "step": 13939
    },
    {
      "epoch": 0.47723382403286546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3879,
      "step": 13940
    },
    {
      "epoch": 0.47726805888394386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3941,
      "step": 13941
    },
    {
      "epoch": 0.47730229373502225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4194,
      "step": 13942
    },
    {
      "epoch": 0.47733652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4037,
      "step": 13943
    },
    {
      "epoch": 0.47737076343717905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4039,
      "step": 13944
    },
    {
      "epoch": 0.47740499828825744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.393,
      "step": 13945
    },
    {
      "epoch": 0.47743923313933584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3888,
      "step": 13946
    },
    {
      "epoch": 0.47747346799041424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3861,
      "step": 13947
    },
    {
      "epoch": 0.47750770284149263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.393,
      "step": 13948
    },
    {
      "epoch": 0.47754193769257103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4111,
      "step": 13949
    },
    {
      "epoch": 0.4775761725436494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4055,
      "step": 13950
    },
    {
      "epoch": 0.4776104073947278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3775,
      "step": 13951
    },
    {
      "epoch": 0.4776446422458062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3957,
      "step": 13952
    },
    {
      "epoch": 0.4776788770968846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3929,
      "step": 13953
    },
    {
      "epoch": 0.477713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4034,
      "step": 13954
    },
    {
      "epoch": 0.4777473467990414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3868,
      "step": 13955
    },
    {
      "epoch": 0.4777815816501198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.39,
      "step": 13956
    },
    {
      "epoch": 0.4778158165011982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4072,
      "step": 13957
    },
    {
      "epoch": 0.4778500513522766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3924,
      "step": 13958
    },
    {
      "epoch": 0.477884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4127,
      "step": 13959
    },
    {
      "epoch": 0.4779185210544334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3856,
      "step": 13960
    },
    {
      "epoch": 0.4779527559055118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3781,
      "step": 13961
    },
    {
      "epoch": 0.47798699075659024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.387,
      "step": 13962
    },
    {
      "epoch": 0.47802122560766863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4001,
      "step": 13963
    },
    {
      "epoch": 0.47805546045874703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3969,
      "step": 13964
    },
    {
      "epoch": 0.4780896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3797,
      "step": 13965
    },
    {
      "epoch": 0.4781239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3878,
      "step": 13966
    },
    {
      "epoch": 0.4781581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3911,
      "step": 13967
    },
    {
      "epoch": 0.4781923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4143,
      "step": 13968
    },
    {
      "epoch": 0.478226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3739,
      "step": 13969
    },
    {
      "epoch": 0.4782608695652174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3928,
      "step": 13970
    },
    {
      "epoch": 0.4782951044162958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4212,
      "step": 13971
    },
    {
      "epoch": 0.4783293392673742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3975,
      "step": 13972
    },
    {
      "epoch": 0.4783635741184526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3906,
      "step": 13973
    },
    {
      "epoch": 0.478397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3953,
      "step": 13974
    },
    {
      "epoch": 0.4784320438206094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3921,
      "step": 13975
    },
    {
      "epoch": 0.4784662786716878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3935,
      "step": 13976
    },
    {
      "epoch": 0.4785005135227662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3832,
      "step": 13977
    },
    {
      "epoch": 0.4785347483738446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3864,
      "step": 13978
    },
    {
      "epoch": 0.478568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.373,
      "step": 13979
    },
    {
      "epoch": 0.47860321807600137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3855,
      "step": 13980
    },
    {
      "epoch": 0.47863745292707977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4056,
      "step": 13981
    },
    {
      "epoch": 0.47867168777815816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3922,
      "step": 13982
    },
    {
      "epoch": 0.47870592262923656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4027,
      "step": 13983
    },
    {
      "epoch": 0.47874015748031495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3858,
      "step": 13984
    },
    {
      "epoch": 0.47877439233139335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3804,
      "step": 13985
    },
    {
      "epoch": 0.47880862718247175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3842,
      "step": 13986
    },
    {
      "epoch": 0.47884286203355014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3914,
      "step": 13987
    },
    {
      "epoch": 0.47887709688462854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3877,
      "step": 13988
    },
    {
      "epoch": 0.47891133173570694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3991,
      "step": 13989
    },
    {
      "epoch": 0.47894556658678533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3867,
      "step": 13990
    },
    {
      "epoch": 0.47897980143786373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3878,
      "step": 13991
    },
    {
      "epoch": 0.4790140362889421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3915,
      "step": 13992
    },
    {
      "epoch": 0.4790482711400205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3984,
      "step": 13993
    },
    {
      "epoch": 0.4790825059910989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3755,
      "step": 13994
    },
    {
      "epoch": 0.4791167408421773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3875,
      "step": 13995
    },
    {
      "epoch": 0.4791509756932557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3906,
      "step": 13996
    },
    {
      "epoch": 0.4791852105443341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3839,
      "step": 13997
    },
    {
      "epoch": 0.47921944539541256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3778,
      "step": 13998
    },
    {
      "epoch": 0.47925368024649095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.406,
      "step": 13999
    },
    {
      "epoch": 0.47928791509756935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.396,
      "step": 14000
    },
    {
      "epoch": 0.47932214994864775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.413,
      "step": 14001
    },
    {
      "epoch": 0.47935638479972614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3999,
      "step": 14002
    },
    {
      "epoch": 0.47939061965080454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3943,
      "step": 14003
    },
    {
      "epoch": 0.47942485450188294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3979,
      "step": 14004
    },
    {
      "epoch": 0.47945908935296133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3978,
      "step": 14005
    },
    {
      "epoch": 0.47949332420403973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3807,
      "step": 14006
    },
    {
      "epoch": 0.4795275590551181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3731,
      "step": 14007
    },
    {
      "epoch": 0.4795617939061965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3972,
      "step": 14008
    },
    {
      "epoch": 0.4795960287572749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3897,
      "step": 14009
    },
    {
      "epoch": 0.4796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4017,
      "step": 14010
    },
    {
      "epoch": 0.4796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.384,
      "step": 14011
    },
    {
      "epoch": 0.4796987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3752,
      "step": 14012
    },
    {
      "epoch": 0.4797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4067,
      "step": 14013
    },
    {
      "epoch": 0.4797672030126669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3875,
      "step": 14014
    },
    {
      "epoch": 0.4798014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3798,
      "step": 14015
    },
    {
      "epoch": 0.4798356727148237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3814,
      "step": 14016
    },
    {
      "epoch": 0.4798699075659021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.38,
      "step": 14017
    },
    {
      "epoch": 0.4799041424169805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3945,
      "step": 14018
    },
    {
      "epoch": 0.4799383772680589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3743,
      "step": 14019
    },
    {
      "epoch": 0.4799726121191373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3977,
      "step": 14020
    },
    {
      "epoch": 0.4800068469702157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3829,
      "step": 14021
    },
    {
      "epoch": 0.48004108182129407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.407,
      "step": 14022
    },
    {
      "epoch": 0.48007531667237247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3969,
      "step": 14023
    },
    {
      "epoch": 0.48010955152345086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3757,
      "step": 14024
    },
    {
      "epoch": 0.48014378637452926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3819,
      "step": 14025
    },
    {
      "epoch": 0.48017802122560765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3976,
      "step": 14026
    },
    {
      "epoch": 0.48021225607668605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3709,
      "step": 14027
    },
    {
      "epoch": 0.48024649092776445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3707,
      "step": 14028
    },
    {
      "epoch": 0.48028072577884284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3746,
      "step": 14029
    },
    {
      "epoch": 0.48031496062992124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3997,
      "step": 14030
    },
    {
      "epoch": 0.48034919548099964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.403,
      "step": 14031
    },
    {
      "epoch": 0.48038343033207803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4002,
      "step": 14032
    },
    {
      "epoch": 0.48041766518315643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3922,
      "step": 14033
    },
    {
      "epoch": 0.4804519000342348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3814,
      "step": 14034
    },
    {
      "epoch": 0.4804861348853133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3979,
      "step": 14035
    },
    {
      "epoch": 0.4805203697363917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.399,
      "step": 14036
    },
    {
      "epoch": 0.48055460458747007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3864,
      "step": 14037
    },
    {
      "epoch": 0.48058883943854847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3801,
      "step": 14038
    },
    {
      "epoch": 0.48062307428962686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3807,
      "step": 14039
    },
    {
      "epoch": 0.48065730914070526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3937,
      "step": 14040
    },
    {
      "epoch": 0.48069154399178365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3892,
      "step": 14041
    },
    {
      "epoch": 0.48072577884286205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3802,
      "step": 14042
    },
    {
      "epoch": 0.48076001369394045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.389,
      "step": 14043
    },
    {
      "epoch": 0.48079424854501884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4111,
      "step": 14044
    },
    {
      "epoch": 0.48082848339609724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3745,
      "step": 14045
    },
    {
      "epoch": 0.48086271824717564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3844,
      "step": 14046
    },
    {
      "epoch": 0.48089695309825403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3845,
      "step": 14047
    },
    {
      "epoch": 0.48093118794933243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.389,
      "step": 14048
    },
    {
      "epoch": 0.4809654228004108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3987,
      "step": 14049
    },
    {
      "epoch": 0.4809996576514892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3859,
      "step": 14050
    },
    {
      "epoch": 0.4810338925025676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3688,
      "step": 14051
    },
    {
      "epoch": 0.481068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3995,
      "step": 14052
    },
    {
      "epoch": 0.4811023622047244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3781,
      "step": 14053
    },
    {
      "epoch": 0.4811365970558028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3876,
      "step": 14054
    },
    {
      "epoch": 0.4811708319068812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3901,
      "step": 14055
    },
    {
      "epoch": 0.4812050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3914,
      "step": 14056
    },
    {
      "epoch": 0.481239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3778,
      "step": 14057
    },
    {
      "epoch": 0.4812735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3678,
      "step": 14058
    },
    {
      "epoch": 0.4813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3893,
      "step": 14059
    },
    {
      "epoch": 0.4813420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4074,
      "step": 14060
    },
    {
      "epoch": 0.4813762410133516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3929,
      "step": 14061
    },
    {
      "epoch": 0.48141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3911,
      "step": 14062
    },
    {
      "epoch": 0.4814447107155084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3907,
      "step": 14063
    },
    {
      "epoch": 0.48147894556658677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3922,
      "step": 14064
    },
    {
      "epoch": 0.48151318041766517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3953,
      "step": 14065
    },
    {
      "epoch": 0.48154741526874356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3753,
      "step": 14066
    },
    {
      "epoch": 0.48158165011982196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.383,
      "step": 14067
    },
    {
      "epoch": 0.48161588497090035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4005,
      "step": 14068
    },
    {
      "epoch": 0.48165011982197875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3934,
      "step": 14069
    },
    {
      "epoch": 0.48168435467305715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3848,
      "step": 14070
    },
    {
      "epoch": 0.48171858952413554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3892,
      "step": 14071
    },
    {
      "epoch": 0.481752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3561,
      "step": 14072
    },
    {
      "epoch": 0.4817870592262924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4008,
      "step": 14073
    },
    {
      "epoch": 0.4818212940773708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3887,
      "step": 14074
    },
    {
      "epoch": 0.4818555289284492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3721,
      "step": 14075
    },
    {
      "epoch": 0.4818897637795276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3984,
      "step": 14076
    },
    {
      "epoch": 0.481923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3864,
      "step": 14077
    },
    {
      "epoch": 0.4819582334816844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3995,
      "step": 14078
    },
    {
      "epoch": 0.48199246833276277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3827,
      "step": 14079
    },
    {
      "epoch": 0.48202670318384117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.396,
      "step": 14080
    },
    {
      "epoch": 0.48206093803491956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3793,
      "step": 14081
    },
    {
      "epoch": 0.48209517288599796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3716,
      "step": 14082
    },
    {
      "epoch": 0.48212940773707635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3771,
      "step": 14083
    },
    {
      "epoch": 0.48216364258815475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4052,
      "step": 14084
    },
    {
      "epoch": 0.48219787743923315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.391,
      "step": 14085
    },
    {
      "epoch": 0.48223211229031154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3957,
      "step": 14086
    },
    {
      "epoch": 0.48226634714138994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4089,
      "step": 14087
    },
    {
      "epoch": 0.48230058199246834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3765,
      "step": 14088
    },
    {
      "epoch": 0.48233481684354673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3929,
      "step": 14089
    },
    {
      "epoch": 0.48236905169462513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3862,
      "step": 14090
    },
    {
      "epoch": 0.4824032865457035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4045,
      "step": 14091
    },
    {
      "epoch": 0.4824375213967819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3895,
      "step": 14092
    },
    {
      "epoch": 0.4824717562478603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4085,
      "step": 14093
    },
    {
      "epoch": 0.4825059910989387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3735,
      "step": 14094
    },
    {
      "epoch": 0.4825402259500171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4044,
      "step": 14095
    },
    {
      "epoch": 0.4825744608010955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3906,
      "step": 14096
    },
    {
      "epoch": 0.4826086956521739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3899,
      "step": 14097
    },
    {
      "epoch": 0.4826429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3891,
      "step": 14098
    },
    {
      "epoch": 0.4826771653543307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3955,
      "step": 14099
    },
    {
      "epoch": 0.4827114002054091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4011,
      "step": 14100
    },
    {
      "epoch": 0.4827456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3799,
      "step": 14101
    },
    {
      "epoch": 0.4827798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3858,
      "step": 14102
    },
    {
      "epoch": 0.4828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3853,
      "step": 14103
    },
    {
      "epoch": 0.4828483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3788,
      "step": 14104
    },
    {
      "epoch": 0.4828825744608011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3684,
      "step": 14105
    },
    {
      "epoch": 0.48291680931187947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3903,
      "step": 14106
    },
    {
      "epoch": 0.48295104416295787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4009,
      "step": 14107
    },
    {
      "epoch": 0.48298527901403626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3926,
      "step": 14108
    },
    {
      "epoch": 0.4830195138651147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3901,
      "step": 14109
    },
    {
      "epoch": 0.4830537487161931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3774,
      "step": 14110
    },
    {
      "epoch": 0.4830879835672715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4132,
      "step": 14111
    },
    {
      "epoch": 0.4831222184183499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4019,
      "step": 14112
    },
    {
      "epoch": 0.4831564532694283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3724,
      "step": 14113
    },
    {
      "epoch": 0.4831906881205067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3932,
      "step": 14114
    },
    {
      "epoch": 0.4832249229715851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3988,
      "step": 14115
    },
    {
      "epoch": 0.4832591578226635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3908,
      "step": 14116
    },
    {
      "epoch": 0.4832933926737419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.374,
      "step": 14117
    },
    {
      "epoch": 0.4833276275248203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3732,
      "step": 14118
    },
    {
      "epoch": 0.4833618623758987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3756,
      "step": 14119
    },
    {
      "epoch": 0.4833960972269771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3899,
      "step": 14120
    },
    {
      "epoch": 0.48343033207805547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.396,
      "step": 14121
    },
    {
      "epoch": 0.48346456692913387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3968,
      "step": 14122
    },
    {
      "epoch": 0.48349880178021226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3986,
      "step": 14123
    },
    {
      "epoch": 0.48353303663129066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3956,
      "step": 14124
    },
    {
      "epoch": 0.48356727148236905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3784,
      "step": 14125
    },
    {
      "epoch": 0.48360150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3892,
      "step": 14126
    },
    {
      "epoch": 0.48363574118452585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.368,
      "step": 14127
    },
    {
      "epoch": 0.48366997603560424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3881,
      "step": 14128
    },
    {
      "epoch": 0.48370421088668264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3888,
      "step": 14129
    },
    {
      "epoch": 0.48373844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3993,
      "step": 14130
    },
    {
      "epoch": 0.48377268058883943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3901,
      "step": 14131
    },
    {
      "epoch": 0.48380691543991783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4127,
      "step": 14132
    },
    {
      "epoch": 0.4838411502909962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3764,
      "step": 14133
    },
    {
      "epoch": 0.4838753851420746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3903,
      "step": 14134
    },
    {
      "epoch": 0.483909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3851,
      "step": 14135
    },
    {
      "epoch": 0.4839438548442314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3768,
      "step": 14136
    },
    {
      "epoch": 0.4839780896953098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3875,
      "step": 14137
    },
    {
      "epoch": 0.4840123245463882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.369,
      "step": 14138
    },
    {
      "epoch": 0.4840465593974666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3938,
      "step": 14139
    },
    {
      "epoch": 0.484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3903,
      "step": 14140
    },
    {
      "epoch": 0.4841150290996234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3837,
      "step": 14141
    },
    {
      "epoch": 0.4841492639507018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3756,
      "step": 14142
    },
    {
      "epoch": 0.4841834988017802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3794,
      "step": 14143
    },
    {
      "epoch": 0.4842177336528586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4039,
      "step": 14144
    },
    {
      "epoch": 0.484251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.39,
      "step": 14145
    },
    {
      "epoch": 0.48428620335501543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3858,
      "step": 14146
    },
    {
      "epoch": 0.48432043820609383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3922,
      "step": 14147
    },
    {
      "epoch": 0.4843546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3769,
      "step": 14148
    },
    {
      "epoch": 0.4843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.381,
      "step": 14149
    },
    {
      "epoch": 0.484423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.394,
      "step": 14150
    },
    {
      "epoch": 0.4844573776104074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3727,
      "step": 14151
    },
    {
      "epoch": 0.4844916124614858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3657,
      "step": 14152
    },
    {
      "epoch": 0.4845258473125642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3768,
      "step": 14153
    },
    {
      "epoch": 0.4845600821636426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3962,
      "step": 14154
    },
    {
      "epoch": 0.484594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3875,
      "step": 14155
    },
    {
      "epoch": 0.4846285518657994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3873,
      "step": 14156
    },
    {
      "epoch": 0.4846627867168778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3968,
      "step": 14157
    },
    {
      "epoch": 0.4846970215679562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3782,
      "step": 14158
    },
    {
      "epoch": 0.4847312564190346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3795,
      "step": 14159
    },
    {
      "epoch": 0.484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3889,
      "step": 14160
    },
    {
      "epoch": 0.4847997261211914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3935,
      "step": 14161
    },
    {
      "epoch": 0.4848339609722698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4122,
      "step": 14162
    },
    {
      "epoch": 0.48486819582334817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3967,
      "step": 14163
    },
    {
      "epoch": 0.48490243067442657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3839,
      "step": 14164
    },
    {
      "epoch": 0.48493666552550496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3685,
      "step": 14165
    },
    {
      "epoch": 0.48497090037658336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3952,
      "step": 14166
    },
    {
      "epoch": 0.48500513522766175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3792,
      "step": 14167
    },
    {
      "epoch": 0.48503937007874015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3743,
      "step": 14168
    },
    {
      "epoch": 0.48507360492981855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3782,
      "step": 14169
    },
    {
      "epoch": 0.48510783978089694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3945,
      "step": 14170
    },
    {
      "epoch": 0.48514207463197534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.366,
      "step": 14171
    },
    {
      "epoch": 0.48517630948305374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3695,
      "step": 14172
    },
    {
      "epoch": 0.48521054433413213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3843,
      "step": 14173
    },
    {
      "epoch": 0.48524477918521053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.387,
      "step": 14174
    },
    {
      "epoch": 0.4852790140362889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3773,
      "step": 14175
    },
    {
      "epoch": 0.4853132488873673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3764,
      "step": 14176
    },
    {
      "epoch": 0.4853474837384457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3903,
      "step": 14177
    },
    {
      "epoch": 0.4853817185895241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3693,
      "step": 14178
    },
    {
      "epoch": 0.4854159534406025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3878,
      "step": 14179
    },
    {
      "epoch": 0.4854501882916809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3737,
      "step": 14180
    },
    {
      "epoch": 0.4854844231427593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3882,
      "step": 14181
    },
    {
      "epoch": 0.48551865799383775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3999,
      "step": 14182
    },
    {
      "epoch": 0.48555289284491615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3933,
      "step": 14183
    },
    {
      "epoch": 0.48558712769599455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3898,
      "step": 14184
    },
    {
      "epoch": 0.48562136254707294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3837,
      "step": 14185
    },
    {
      "epoch": 0.48565559739815134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3758,
      "step": 14186
    },
    {
      "epoch": 0.48568983224922974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3882,
      "step": 14187
    },
    {
      "epoch": 0.48572406710030813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3897,
      "step": 14188
    },
    {
      "epoch": 0.48575830195138653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3951,
      "step": 14189
    },
    {
      "epoch": 0.4857925368024649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3748,
      "step": 14190
    },
    {
      "epoch": 0.4858267716535433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.385,
      "step": 14191
    },
    {
      "epoch": 0.4858610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3705,
      "step": 14192
    },
    {
      "epoch": 0.4858952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3819,
      "step": 14193
    },
    {
      "epoch": 0.4859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4007,
      "step": 14194
    },
    {
      "epoch": 0.4859637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3827,
      "step": 14195
    },
    {
      "epoch": 0.4859979459089353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3749,
      "step": 14196
    },
    {
      "epoch": 0.4860321807600137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3833,
      "step": 14197
    },
    {
      "epoch": 0.4860664156110921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3823,
      "step": 14198
    },
    {
      "epoch": 0.4861006504621705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3595,
      "step": 14199
    },
    {
      "epoch": 0.4861348853132489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4018,
      "step": 14200
    },
    {
      "epoch": 0.4861691201643273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3841,
      "step": 14201
    },
    {
      "epoch": 0.4862033550154057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3734,
      "step": 14202
    },
    {
      "epoch": 0.4862375898664841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3884,
      "step": 14203
    },
    {
      "epoch": 0.4862718247175625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3989,
      "step": 14204
    },
    {
      "epoch": 0.48630605956864087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3833,
      "step": 14205
    },
    {
      "epoch": 0.48634029441971927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4089,
      "step": 14206
    },
    {
      "epoch": 0.48637452927079766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3727,
      "step": 14207
    },
    {
      "epoch": 0.48640876412187606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3611,
      "step": 14208
    },
    {
      "epoch": 0.48644299897295445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3641,
      "step": 14209
    },
    {
      "epoch": 0.48647723382403285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3794,
      "step": 14210
    },
    {
      "epoch": 0.48651146867511125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4016,
      "step": 14211
    },
    {
      "epoch": 0.48654570352618964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3919,
      "step": 14212
    },
    {
      "epoch": 0.48657993837726804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3726,
      "step": 14213
    },
    {
      "epoch": 0.48661417322834644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3816,
      "step": 14214
    },
    {
      "epoch": 0.48664840807942483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3865,
      "step": 14215
    },
    {
      "epoch": 0.48668264293050323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3851,
      "step": 14216
    },
    {
      "epoch": 0.4867168777815816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3785,
      "step": 14217
    },
    {
      "epoch": 0.48675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3776,
      "step": 14218
    },
    {
      "epoch": 0.4867853474837385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3989,
      "step": 14219
    },
    {
      "epoch": 0.48681958233481687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.403,
      "step": 14220
    },
    {
      "epoch": 0.48685381718589527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3903,
      "step": 14221
    },
    {
      "epoch": 0.48688805203697366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3853,
      "step": 14222
    },
    {
      "epoch": 0.48692228688805206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3814,
      "step": 14223
    },
    {
      "epoch": 0.48695652173913045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3892,
      "step": 14224
    },
    {
      "epoch": 0.48699075659020885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3878,
      "step": 14225
    },
    {
      "epoch": 0.48702499144128725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3918,
      "step": 14226
    },
    {
      "epoch": 0.48705922629236564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3901,
      "step": 14227
    },
    {
      "epoch": 0.48709346114344404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3811,
      "step": 14228
    },
    {
      "epoch": 0.48712769599452244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3799,
      "step": 14229
    },
    {
      "epoch": 0.48716193084560083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3712,
      "step": 14230
    },
    {
      "epoch": 0.48719616569667923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3922,
      "step": 14231
    },
    {
      "epoch": 0.4872304005477576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3983,
      "step": 14232
    },
    {
      "epoch": 0.487264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3809,
      "step": 14233
    },
    {
      "epoch": 0.4872988702499144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3929,
      "step": 14234
    },
    {
      "epoch": 0.4873331051009928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3991,
      "step": 14235
    },
    {
      "epoch": 0.4873673399520712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4033,
      "step": 14236
    },
    {
      "epoch": 0.4874015748031496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3989,
      "step": 14237
    },
    {
      "epoch": 0.487435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3897,
      "step": 14238
    },
    {
      "epoch": 0.4874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.387,
      "step": 14239
    },
    {
      "epoch": 0.4875042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3793,
      "step": 14240
    },
    {
      "epoch": 0.4875385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3939,
      "step": 14241
    },
    {
      "epoch": 0.4875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3948,
      "step": 14242
    },
    {
      "epoch": 0.48760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.384,
      "step": 14243
    },
    {
      "epoch": 0.4876412187606984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3782,
      "step": 14244
    },
    {
      "epoch": 0.4876754536117768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3884,
      "step": 14245
    },
    {
      "epoch": 0.4877096884628552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3691,
      "step": 14246
    },
    {
      "epoch": 0.48774392331393357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3643,
      "step": 14247
    },
    {
      "epoch": 0.48777815816501197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3851,
      "step": 14248
    },
    {
      "epoch": 0.48781239301609036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3776,
      "step": 14249
    },
    {
      "epoch": 0.48784662786716876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3945,
      "step": 14250
    },
    {
      "epoch": 0.48788086271824715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3696,
      "step": 14251
    },
    {
      "epoch": 0.48791509756932555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3765,
      "step": 14252
    },
    {
      "epoch": 0.48794933242040395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3878,
      "step": 14253
    },
    {
      "epoch": 0.48798356727148234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3698,
      "step": 14254
    },
    {
      "epoch": 0.48801780212256074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3796,
      "step": 14255
    },
    {
      "epoch": 0.4880520369736392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3793,
      "step": 14256
    },
    {
      "epoch": 0.4880862718247176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3955,
      "step": 14257
    },
    {
      "epoch": 0.488120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3754,
      "step": 14258
    },
    {
      "epoch": 0.4881547415268744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3919,
      "step": 14259
    },
    {
      "epoch": 0.4881889763779528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3845,
      "step": 14260
    },
    {
      "epoch": 0.4882232112290312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3991,
      "step": 14261
    },
    {
      "epoch": 0.48825744608010957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3865,
      "step": 14262
    },
    {
      "epoch": 0.48829168093118797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3764,
      "step": 14263
    },
    {
      "epoch": 0.48832591578226636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3976,
      "step": 14264
    },
    {
      "epoch": 0.48836015063334476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3643,
      "step": 14265
    },
    {
      "epoch": 0.48839438548442315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.369,
      "step": 14266
    },
    {
      "epoch": 0.48842862033550155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3889,
      "step": 14267
    },
    {
      "epoch": 0.48846285518657995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3666,
      "step": 14268
    },
    {
      "epoch": 0.48849709003765834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3752,
      "step": 14269
    },
    {
      "epoch": 0.48853132488873674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.394,
      "step": 14270
    },
    {
      "epoch": 0.48856555973981514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3685,
      "step": 14271
    },
    {
      "epoch": 0.48859979459089353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3946,
      "step": 14272
    },
    {
      "epoch": 0.48863402944197193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.397,
      "step": 14273
    },
    {
      "epoch": 0.4886682642930503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3745,
      "step": 14274
    },
    {
      "epoch": 0.4887024991441287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3845,
      "step": 14275
    },
    {
      "epoch": 0.4887367339952071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.385,
      "step": 14276
    },
    {
      "epoch": 0.4887709688462855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3831,
      "step": 14277
    },
    {
      "epoch": 0.4888052036973639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3615,
      "step": 14278
    },
    {
      "epoch": 0.4888394385484423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3868,
      "step": 14279
    },
    {
      "epoch": 0.4888736733995207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3704,
      "step": 14280
    },
    {
      "epoch": 0.4889079082505991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3734,
      "step": 14281
    },
    {
      "epoch": 0.4889421431016775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3735,
      "step": 14282
    },
    {
      "epoch": 0.4889763779527559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3723,
      "step": 14283
    },
    {
      "epoch": 0.4890106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3688,
      "step": 14284
    },
    {
      "epoch": 0.4890448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3972,
      "step": 14285
    },
    {
      "epoch": 0.4890790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3721,
      "step": 14286
    },
    {
      "epoch": 0.4891133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3676,
      "step": 14287
    },
    {
      "epoch": 0.4891475522081479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3843,
      "step": 14288
    },
    {
      "epoch": 0.48918178705922627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3762,
      "step": 14289
    },
    {
      "epoch": 0.48921602191030467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4003,
      "step": 14290
    },
    {
      "epoch": 0.48925025676138306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3715,
      "step": 14291
    },
    {
      "epoch": 0.48928449161246146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.395,
      "step": 14292
    },
    {
      "epoch": 0.4893187264635399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3619,
      "step": 14293
    },
    {
      "epoch": 0.4893529613146183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3815,
      "step": 14294
    },
    {
      "epoch": 0.4893871961656967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3918,
      "step": 14295
    },
    {
      "epoch": 0.4894214310167751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3927,
      "step": 14296
    },
    {
      "epoch": 0.4894556658678535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3798,
      "step": 14297
    },
    {
      "epoch": 0.4894899007189319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3923,
      "step": 14298
    },
    {
      "epoch": 0.4895241355700103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3849,
      "step": 14299
    },
    {
      "epoch": 0.4895583704210887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3758,
      "step": 14300
    },
    {
      "epoch": 0.4895926052721671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3693,
      "step": 14301
    },
    {
      "epoch": 0.4896268401232455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3617,
      "step": 14302
    },
    {
      "epoch": 0.4896610749743239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3903,
      "step": 14303
    },
    {
      "epoch": 0.48969530982540227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4185,
      "step": 14304
    },
    {
      "epoch": 0.48972954467648067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3723,
      "step": 14305
    },
    {
      "epoch": 0.48976377952755906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3857,
      "step": 14306
    },
    {
      "epoch": 0.48979801437863746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3788,
      "step": 14307
    },
    {
      "epoch": 0.48983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3967,
      "step": 14308
    },
    {
      "epoch": 0.48986648408079425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3822,
      "step": 14309
    },
    {
      "epoch": 0.48990071893187265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3857,
      "step": 14310
    },
    {
      "epoch": 0.48993495378295104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.394,
      "step": 14311
    },
    {
      "epoch": 0.48996918863402944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3818,
      "step": 14312
    },
    {
      "epoch": 0.49000342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3892,
      "step": 14313
    },
    {
      "epoch": 0.49003765833618623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3989,
      "step": 14314
    },
    {
      "epoch": 0.49007189318726463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3798,
      "step": 14315
    },
    {
      "epoch": 0.490106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3879,
      "step": 14316
    },
    {
      "epoch": 0.4901403628894214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3679,
      "step": 14317
    },
    {
      "epoch": 0.4901745977404998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4092,
      "step": 14318
    },
    {
      "epoch": 0.4902088325915782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4207,
      "step": 14319
    },
    {
      "epoch": 0.4902430674426566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.385,
      "step": 14320
    },
    {
      "epoch": 0.490277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3825,
      "step": 14321
    },
    {
      "epoch": 0.4903115371448134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3825,
      "step": 14322
    },
    {
      "epoch": 0.4903457719958918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.394,
      "step": 14323
    },
    {
      "epoch": 0.4903800068469702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3881,
      "step": 14324
    },
    {
      "epoch": 0.4904142416980486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3533,
      "step": 14325
    },
    {
      "epoch": 0.490448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3945,
      "step": 14326
    },
    {
      "epoch": 0.4904827114002054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3991,
      "step": 14327
    },
    {
      "epoch": 0.4905169462512838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3807,
      "step": 14328
    },
    {
      "epoch": 0.4905511811023622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3799,
      "step": 14329
    },
    {
      "epoch": 0.49058541595344063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.38,
      "step": 14330
    },
    {
      "epoch": 0.490619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3929,
      "step": 14331
    },
    {
      "epoch": 0.4906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3932,
      "step": 14332
    },
    {
      "epoch": 0.4906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3818,
      "step": 14333
    },
    {
      "epoch": 0.4907223553577542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3819,
      "step": 14334
    },
    {
      "epoch": 0.4907565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.39,
      "step": 14335
    },
    {
      "epoch": 0.490790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3777,
      "step": 14336
    },
    {
      "epoch": 0.4908250599109894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3846,
      "step": 14337
    },
    {
      "epoch": 0.4908592947620678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3868,
      "step": 14338
    },
    {
      "epoch": 0.4908935296131462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3659,
      "step": 14339
    },
    {
      "epoch": 0.4909277644642246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3731,
      "step": 14340
    },
    {
      "epoch": 0.490961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3888,
      "step": 14341
    },
    {
      "epoch": 0.4909962341663814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3816,
      "step": 14342
    },
    {
      "epoch": 0.4910304690174598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3948,
      "step": 14343
    },
    {
      "epoch": 0.4910647038685382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.382,
      "step": 14344
    },
    {
      "epoch": 0.4910989387196166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3974,
      "step": 14345
    },
    {
      "epoch": 0.49113317357069497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3805,
      "step": 14346
    },
    {
      "epoch": 0.49116740842177337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3892,
      "step": 14347
    },
    {
      "epoch": 0.49120164327285176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3789,
      "step": 14348
    },
    {
      "epoch": 0.49123587812393016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3663,
      "step": 14349
    },
    {
      "epoch": 0.49127011297500855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.381,
      "step": 14350
    },
    {
      "epoch": 0.49130434782608695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3886,
      "step": 14351
    },
    {
      "epoch": 0.49133858267716535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3766,
      "step": 14352
    },
    {
      "epoch": 0.49137281752824374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3917,
      "step": 14353
    },
    {
      "epoch": 0.49140705237932214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3816,
      "step": 14354
    },
    {
      "epoch": 0.49144128723040054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3633,
      "step": 14355
    },
    {
      "epoch": 0.49147552208147893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3602,
      "step": 14356
    },
    {
      "epoch": 0.49150975693255733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3857,
      "step": 14357
    },
    {
      "epoch": 0.4915439917836357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3931,
      "step": 14358
    },
    {
      "epoch": 0.4915782266347141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3786,
      "step": 14359
    },
    {
      "epoch": 0.4916124614857925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.366,
      "step": 14360
    },
    {
      "epoch": 0.4916466963368709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3854,
      "step": 14361
    },
    {
      "epoch": 0.4916809311879493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3989,
      "step": 14362
    },
    {
      "epoch": 0.4917151660390277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3841,
      "step": 14363
    },
    {
      "epoch": 0.4917494008901061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3738,
      "step": 14364
    },
    {
      "epoch": 0.4917836357411845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3785,
      "step": 14365
    },
    {
      "epoch": 0.49181787059226295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3859,
      "step": 14366
    },
    {
      "epoch": 0.49185210544334135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3878,
      "step": 14367
    },
    {
      "epoch": 0.49188634029441974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3734,
      "step": 14368
    },
    {
      "epoch": 0.49192057514549814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3809,
      "step": 14369
    },
    {
      "epoch": 0.49195480999657654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3679,
      "step": 14370
    },
    {
      "epoch": 0.49198904484765493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3681,
      "step": 14371
    },
    {
      "epoch": 0.49202327969873333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3716,
      "step": 14372
    },
    {
      "epoch": 0.4920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3845,
      "step": 14373
    },
    {
      "epoch": 0.4920917494008901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3961,
      "step": 14374
    },
    {
      "epoch": 0.4921259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3842,
      "step": 14375
    },
    {
      "epoch": 0.4921602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3736,
      "step": 14376
    },
    {
      "epoch": 0.4921944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3792,
      "step": 14377
    },
    {
      "epoch": 0.4922286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3972,
      "step": 14378
    },
    {
      "epoch": 0.4922629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.393,
      "step": 14379
    },
    {
      "epoch": 0.4922971585073605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.372,
      "step": 14380
    },
    {
      "epoch": 0.4923313933584389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3803,
      "step": 14381
    },
    {
      "epoch": 0.4923656282095173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3989,
      "step": 14382
    },
    {
      "epoch": 0.4923998630605957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3593,
      "step": 14383
    },
    {
      "epoch": 0.4924340979116741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3785,
      "step": 14384
    },
    {
      "epoch": 0.4924683327627525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3793,
      "step": 14385
    },
    {
      "epoch": 0.4925025676138309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3816,
      "step": 14386
    },
    {
      "epoch": 0.4925368024649093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3694,
      "step": 14387
    },
    {
      "epoch": 0.49257103731598767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3665,
      "step": 14388
    },
    {
      "epoch": 0.49260527216706607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3908,
      "step": 14389
    },
    {
      "epoch": 0.49263950701814446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3695,
      "step": 14390
    },
    {
      "epoch": 0.49267374186922286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3871,
      "step": 14391
    },
    {
      "epoch": 0.49270797672030126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3757,
      "step": 14392
    },
    {
      "epoch": 0.49274221157137965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.384,
      "step": 14393
    },
    {
      "epoch": 0.49277644642245805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3614,
      "step": 14394
    },
    {
      "epoch": 0.49281068127353644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3845,
      "step": 14395
    },
    {
      "epoch": 0.49284491612461484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3792,
      "step": 14396
    },
    {
      "epoch": 0.49287915097569324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3419,
      "step": 14397
    },
    {
      "epoch": 0.49291338582677163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4006,
      "step": 14398
    },
    {
      "epoch": 0.49294762067785003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3773,
      "step": 14399
    },
    {
      "epoch": 0.4929818555289284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3759,
      "step": 14400
    },
    {
      "epoch": 0.4930160903800068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3795,
      "step": 14401
    },
    {
      "epoch": 0.4930503252310852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3995,
      "step": 14402
    },
    {
      "epoch": 0.49308456008216367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.393,
      "step": 14403
    },
    {
      "epoch": 0.49311879493324207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.38,
      "step": 14404
    },
    {
      "epoch": 0.49315302978432046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.371,
      "step": 14405
    },
    {
      "epoch": 0.49318726463539886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3796,
      "step": 14406
    },
    {
      "epoch": 0.49322149948647726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3695,
      "step": 14407
    },
    {
      "epoch": 0.49325573433755565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3745,
      "step": 14408
    },
    {
      "epoch": 0.49328996918863405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3776,
      "step": 14409
    },
    {
      "epoch": 0.49332420403971244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3762,
      "step": 14410
    },
    {
      "epoch": 0.49335843889079084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3633,
      "step": 14411
    },
    {
      "epoch": 0.49339267374186924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3699,
      "step": 14412
    },
    {
      "epoch": 0.49342690859294763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3772,
      "step": 14413
    },
    {
      "epoch": 0.49346114344402603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3903,
      "step": 14414
    },
    {
      "epoch": 0.4934953782951044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4021,
      "step": 14415
    },
    {
      "epoch": 0.4935296131461828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.367,
      "step": 14416
    },
    {
      "epoch": 0.4935638479972612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3688,
      "step": 14417
    },
    {
      "epoch": 0.4935980828483396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3615,
      "step": 14418
    },
    {
      "epoch": 0.493632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3752,
      "step": 14419
    },
    {
      "epoch": 0.4936665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.377,
      "step": 14420
    },
    {
      "epoch": 0.4937007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3908,
      "step": 14421
    },
    {
      "epoch": 0.4937350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3892,
      "step": 14422
    },
    {
      "epoch": 0.4937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3639,
      "step": 14423
    },
    {
      "epoch": 0.49380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3742,
      "step": 14424
    },
    {
      "epoch": 0.4938377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3671,
      "step": 14425
    },
    {
      "epoch": 0.4938719616569668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.38,
      "step": 14426
    },
    {
      "epoch": 0.4939061965080452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3677,
      "step": 14427
    },
    {
      "epoch": 0.4939404313591236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3673,
      "step": 14428
    },
    {
      "epoch": 0.493974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3545,
      "step": 14429
    },
    {
      "epoch": 0.49400890106128037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3886,
      "step": 14430
    },
    {
      "epoch": 0.49404313591235877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3749,
      "step": 14431
    },
    {
      "epoch": 0.49407737076343716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3736,
      "step": 14432
    },
    {
      "epoch": 0.49411160561451556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3909,
      "step": 14433
    },
    {
      "epoch": 0.49414584046559396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3695,
      "step": 14434
    },
    {
      "epoch": 0.49418007531667235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3934,
      "step": 14435
    },
    {
      "epoch": 0.49421431016775075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3824,
      "step": 14436
    },
    {
      "epoch": 0.49424854501882914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.386,
      "step": 14437
    },
    {
      "epoch": 0.49428277986990754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3734,
      "step": 14438
    },
    {
      "epoch": 0.49431701472098594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3851,
      "step": 14439
    },
    {
      "epoch": 0.4943512495720644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3704,
      "step": 14440
    },
    {
      "epoch": 0.4943854844231428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4047,
      "step": 14441
    },
    {
      "epoch": 0.4944197192742212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3637,
      "step": 14442
    },
    {
      "epoch": 0.4944539541252996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3952,
      "step": 14443
    },
    {
      "epoch": 0.494488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.387,
      "step": 14444
    },
    {
      "epoch": 0.49452242382745637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3732,
      "step": 14445
    },
    {
      "epoch": 0.49455665867853477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3843,
      "step": 14446
    },
    {
      "epoch": 0.49459089352961316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3957,
      "step": 14447
    },
    {
      "epoch": 0.49462512838069156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3886,
      "step": 14448
    },
    {
      "epoch": 0.49465936323176996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.37,
      "step": 14449
    },
    {
      "epoch": 0.49469359808284835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3987,
      "step": 14450
    },
    {
      "epoch": 0.49472783293392675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3948,
      "step": 14451
    },
    {
      "epoch": 0.49476206778500514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3777,
      "step": 14452
    },
    {
      "epoch": 0.49479630263608354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3782,
      "step": 14453
    },
    {
      "epoch": 0.49483053748716194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3765,
      "step": 14454
    },
    {
      "epoch": 0.49486477233824033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3715,
      "step": 14455
    },
    {
      "epoch": 0.49489900718931873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3765,
      "step": 14456
    },
    {
      "epoch": 0.4949332420403971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3765,
      "step": 14457
    },
    {
      "epoch": 0.4949674768914755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3738,
      "step": 14458
    },
    {
      "epoch": 0.4950017117425539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3758,
      "step": 14459
    },
    {
      "epoch": 0.4950359465936323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3787,
      "step": 14460
    },
    {
      "epoch": 0.4950701814447107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4058,
      "step": 14461
    },
    {
      "epoch": 0.4951044162957891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3781,
      "step": 14462
    },
    {
      "epoch": 0.4951386511468675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3735,
      "step": 14463
    },
    {
      "epoch": 0.4951728859979459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3765,
      "step": 14464
    },
    {
      "epoch": 0.4952071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3745,
      "step": 14465
    },
    {
      "epoch": 0.4952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3872,
      "step": 14466
    },
    {
      "epoch": 0.4952755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3704,
      "step": 14467
    },
    {
      "epoch": 0.4953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3889,
      "step": 14468
    },
    {
      "epoch": 0.4953440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3933,
      "step": 14469
    },
    {
      "epoch": 0.4953782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3916,
      "step": 14470
    },
    {
      "epoch": 0.4954125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3792,
      "step": 14471
    },
    {
      "epoch": 0.49544676480657307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3746,
      "step": 14472
    },
    {
      "epoch": 0.49548099965765147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3579,
      "step": 14473
    },
    {
      "epoch": 0.49551523450872986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.381,
      "step": 14474
    },
    {
      "epoch": 0.49554946935980826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.36,
      "step": 14475
    },
    {
      "epoch": 0.49558370421088666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3997,
      "step": 14476
    },
    {
      "epoch": 0.4956179390619651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3776,
      "step": 14477
    },
    {
      "epoch": 0.4956521739130435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3766,
      "step": 14478
    },
    {
      "epoch": 0.4956864087641219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3842,
      "step": 14479
    },
    {
      "epoch": 0.4957206436152003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.363,
      "step": 14480
    },
    {
      "epoch": 0.4957548784662787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3826,
      "step": 14481
    },
    {
      "epoch": 0.4957891133173571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3611,
      "step": 14482
    },
    {
      "epoch": 0.4958233481684355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3828,
      "step": 14483
    },
    {
      "epoch": 0.4958575830195139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3628,
      "step": 14484
    },
    {
      "epoch": 0.4958918178705923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3693,
      "step": 14485
    },
    {
      "epoch": 0.4959260527216707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.382,
      "step": 14486
    },
    {
      "epoch": 0.49596028757274907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3875,
      "step": 14487
    },
    {
      "epoch": 0.49599452242382747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3824,
      "step": 14488
    },
    {
      "epoch": 0.49602875727490586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.374,
      "step": 14489
    },
    {
      "epoch": 0.49606299212598426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3818,
      "step": 14490
    },
    {
      "epoch": 0.49609722697706266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3837,
      "step": 14491
    },
    {
      "epoch": 0.49613146182814105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3682,
      "step": 14492
    },
    {
      "epoch": 0.49616569667921945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3768,
      "step": 14493
    },
    {
      "epoch": 0.49619993153029784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3707,
      "step": 14494
    },
    {
      "epoch": 0.49623416638137624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.355,
      "step": 14495
    },
    {
      "epoch": 0.49626840123245464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3664,
      "step": 14496
    },
    {
      "epoch": 0.49630263608353303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3758,
      "step": 14497
    },
    {
      "epoch": 0.49633687093461143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3794,
      "step": 14498
    },
    {
      "epoch": 0.4963711057856898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3605,
      "step": 14499
    },
    {
      "epoch": 0.4964053406367682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3668,
      "step": 14500
    },
    {
      "epoch": 0.4964395754878466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3875,
      "step": 14501
    },
    {
      "epoch": 0.496473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3668,
      "step": 14502
    },
    {
      "epoch": 0.4965080451900034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3874,
      "step": 14503
    },
    {
      "epoch": 0.4965422800410818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3822,
      "step": 14504
    },
    {
      "epoch": 0.4965765148921602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4014,
      "step": 14505
    },
    {
      "epoch": 0.4966107497432386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3638,
      "step": 14506
    },
    {
      "epoch": 0.496644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3622,
      "step": 14507
    },
    {
      "epoch": 0.4966792194453954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4034,
      "step": 14508
    },
    {
      "epoch": 0.4967134542964738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3751,
      "step": 14509
    },
    {
      "epoch": 0.4967476891475522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3749,
      "step": 14510
    },
    {
      "epoch": 0.4967819239986306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3825,
      "step": 14511
    },
    {
      "epoch": 0.496816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3572,
      "step": 14512
    },
    {
      "epoch": 0.4968503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3699,
      "step": 14513
    },
    {
      "epoch": 0.4968846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3669,
      "step": 14514
    },
    {
      "epoch": 0.4969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3848,
      "step": 14515
    },
    {
      "epoch": 0.4969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3756,
      "step": 14516
    },
    {
      "epoch": 0.496987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3773,
      "step": 14517
    },
    {
      "epoch": 0.4970215679561794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3632,
      "step": 14518
    },
    {
      "epoch": 0.4970558028072578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3693,
      "step": 14519
    },
    {
      "epoch": 0.4970900376583362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3726,
      "step": 14520
    },
    {
      "epoch": 0.4971242725094146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.374,
      "step": 14521
    },
    {
      "epoch": 0.497158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3768,
      "step": 14522
    },
    {
      "epoch": 0.4971927422115714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.386,
      "step": 14523
    },
    {
      "epoch": 0.4972269770626498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3799,
      "step": 14524
    },
    {
      "epoch": 0.4972612119137282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3877,
      "step": 14525
    },
    {
      "epoch": 0.4972954467648066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3707,
      "step": 14526
    },
    {
      "epoch": 0.497329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3805,
      "step": 14527
    },
    {
      "epoch": 0.4973639164669634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.363,
      "step": 14528
    },
    {
      "epoch": 0.49739815131804177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3592,
      "step": 14529
    },
    {
      "epoch": 0.49743238616912017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3751,
      "step": 14530
    },
    {
      "epoch": 0.49746662102019856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3881,
      "step": 14531
    },
    {
      "epoch": 0.49750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3788,
      "step": 14532
    },
    {
      "epoch": 0.49753509072235536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3808,
      "step": 14533
    },
    {
      "epoch": 0.49756932557343375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3696,
      "step": 14534
    },
    {
      "epoch": 0.49760356042451215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3766,
      "step": 14535
    },
    {
      "epoch": 0.49763779527559054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3865,
      "step": 14536
    },
    {
      "epoch": 0.49767203012666894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3741,
      "step": 14537
    },
    {
      "epoch": 0.49770626497774734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.392,
      "step": 14538
    },
    {
      "epoch": 0.49774049982882573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4099,
      "step": 14539
    },
    {
      "epoch": 0.49777473467990413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3572,
      "step": 14540
    },
    {
      "epoch": 0.4978089695309825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3826,
      "step": 14541
    },
    {
      "epoch": 0.4978432043820609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3832,
      "step": 14542
    },
    {
      "epoch": 0.4978774392331393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3666,
      "step": 14543
    },
    {
      "epoch": 0.4979116740842177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3742,
      "step": 14544
    },
    {
      "epoch": 0.4979459089352961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3658,
      "step": 14545
    },
    {
      "epoch": 0.4979801437863745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3573,
      "step": 14546
    },
    {
      "epoch": 0.4980143786374529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3698,
      "step": 14547
    },
    {
      "epoch": 0.4980486134885313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3934,
      "step": 14548
    },
    {
      "epoch": 0.4980828483396097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3777,
      "step": 14549
    },
    {
      "epoch": 0.49811708319068815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.37,
      "step": 14550
    },
    {
      "epoch": 0.49815131804176654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3691,
      "step": 14551
    },
    {
      "epoch": 0.49818555289284494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3574,
      "step": 14552
    },
    {
      "epoch": 0.49821978774392334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3829,
      "step": 14553
    },
    {
      "epoch": 0.49825402259500173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3866,
      "step": 14554
    },
    {
      "epoch": 0.49828825744608013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3726,
      "step": 14555
    },
    {
      "epoch": 0.4983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3702,
      "step": 14556
    },
    {
      "epoch": 0.4983567271482369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3635,
      "step": 14557
    },
    {
      "epoch": 0.4983909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3641,
      "step": 14558
    },
    {
      "epoch": 0.4984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3741,
      "step": 14559
    },
    {
      "epoch": 0.4984594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.377,
      "step": 14560
    },
    {
      "epoch": 0.4984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3621,
      "step": 14561
    },
    {
      "epoch": 0.4985279014036289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3748,
      "step": 14562
    },
    {
      "epoch": 0.4985621362547073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3826,
      "step": 14563
    },
    {
      "epoch": 0.4985963711057857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3727,
      "step": 14564
    },
    {
      "epoch": 0.4986306059568641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3831,
      "step": 14565
    },
    {
      "epoch": 0.4986648408079425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.385,
      "step": 14566
    },
    {
      "epoch": 0.4986990756590209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3742,
      "step": 14567
    },
    {
      "epoch": 0.4987333105100993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3691,
      "step": 14568
    },
    {
      "epoch": 0.4987675453611777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3862,
      "step": 14569
    },
    {
      "epoch": 0.4988017802122561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3691,
      "step": 14570
    },
    {
      "epoch": 0.49883601506333447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3781,
      "step": 14571
    },
    {
      "epoch": 0.49887024991441287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3761,
      "step": 14572
    },
    {
      "epoch": 0.49890448476549126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3651,
      "step": 14573
    },
    {
      "epoch": 0.49893871961656966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3756,
      "step": 14574
    },
    {
      "epoch": 0.49897295446764806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3774,
      "step": 14575
    },
    {
      "epoch": 0.49900718931872645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3677,
      "step": 14576
    },
    {
      "epoch": 0.49904142416980485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3682,
      "step": 14577
    },
    {
      "epoch": 0.49907565902088324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3726,
      "step": 14578
    },
    {
      "epoch": 0.49910989387196164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3722,
      "step": 14579
    },
    {
      "epoch": 0.49914412872304004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.384,
      "step": 14580
    },
    {
      "epoch": 0.49917836357411843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3774,
      "step": 14581
    },
    {
      "epoch": 0.49921259842519683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3826,
      "step": 14582
    },
    {
      "epoch": 0.4992468332762752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3799,
      "step": 14583
    },
    {
      "epoch": 0.4992810681273536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3708,
      "step": 14584
    },
    {
      "epoch": 0.499315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3754,
      "step": 14585
    },
    {
      "epoch": 0.4993495378295104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3828,
      "step": 14586
    },
    {
      "epoch": 0.49938377268058887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3591,
      "step": 14587
    },
    {
      "epoch": 0.49941800753166726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3783,
      "step": 14588
    },
    {
      "epoch": 0.49945224238274566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3598,
      "step": 14589
    },
    {
      "epoch": 0.49948647723382406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3702,
      "step": 14590
    },
    {
      "epoch": 0.49952071208490245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3624,
      "step": 14591
    },
    {
      "epoch": 0.49955494693598085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3859,
      "step": 14592
    },
    {
      "epoch": 0.49958918178705924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3729,
      "step": 14593
    },
    {
      "epoch": 0.49962341663813764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3738,
      "step": 14594
    },
    {
      "epoch": 0.49965765148921604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3601,
      "step": 14595
    },
    {
      "epoch": 0.49969188634029443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3741,
      "step": 14596
    },
    {
      "epoch": 0.49972612119137283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3698,
      "step": 14597
    },
    {
      "epoch": 0.4997603560424512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3825,
      "step": 14598
    },
    {
      "epoch": 0.4997945908935296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3613,
      "step": 14599
    },
    {
      "epoch": 0.499828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3707,
      "step": 14600
    },
    {
      "epoch": 0.4998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3615,
      "step": 14601
    },
    {
      "epoch": 0.4998972954467648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3703,
      "step": 14602
    },
    {
      "epoch": 0.4999315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3734,
      "step": 14603
    },
    {
      "epoch": 0.4999657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.379,
      "step": 14604
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3788,
      "step": 14605
    },
    {
      "epoch": 0.5000342348510785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3605,
      "step": 14606
    },
    {
      "epoch": 0.5000684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3764,
      "step": 14607
    },
    {
      "epoch": 0.5001027045532352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3885,
      "step": 14608
    },
    {
      "epoch": 0.5001369394043136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3806,
      "step": 14609
    },
    {
      "epoch": 0.500171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3662,
      "step": 14610
    },
    {
      "epoch": 0.5002054091064704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3712,
      "step": 14611
    },
    {
      "epoch": 0.5002396439575488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3639,
      "step": 14612
    },
    {
      "epoch": 0.5002738788086272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3638,
      "step": 14613
    },
    {
      "epoch": 0.5003081136597056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3795,
      "step": 14614
    },
    {
      "epoch": 0.500342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3604,
      "step": 14615
    },
    {
      "epoch": 0.5003765833618624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3724,
      "step": 14616
    },
    {
      "epoch": 0.5004108182129408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3667,
      "step": 14617
    },
    {
      "epoch": 0.5004450530640192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3649,
      "step": 14618
    },
    {
      "epoch": 0.5004792879150975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3633,
      "step": 14619
    },
    {
      "epoch": 0.500513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.382,
      "step": 14620
    },
    {
      "epoch": 0.5005477576172543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3627,
      "step": 14621
    },
    {
      "epoch": 0.5005819924683328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3925,
      "step": 14622
    },
    {
      "epoch": 0.5006162273194111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3785,
      "step": 14623
    },
    {
      "epoch": 0.5006504621704896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3833,
      "step": 14624
    },
    {
      "epoch": 0.5006846970215679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3623,
      "step": 14625
    },
    {
      "epoch": 0.5007189318726464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3641,
      "step": 14626
    },
    {
      "epoch": 0.5007531667237247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3727,
      "step": 14627
    },
    {
      "epoch": 0.5007874015748032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3739,
      "step": 14628
    },
    {
      "epoch": 0.5008216364258815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.363,
      "step": 14629
    },
    {
      "epoch": 0.50085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4036,
      "step": 14630
    },
    {
      "epoch": 0.5008901061280383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3696,
      "step": 14631
    },
    {
      "epoch": 0.5009243409791168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3769,
      "step": 14632
    },
    {
      "epoch": 0.5009585758301951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3845,
      "step": 14633
    },
    {
      "epoch": 0.5009928106812735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3741,
      "step": 14634
    },
    {
      "epoch": 0.5010270455323519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3804,
      "step": 14635
    },
    {
      "epoch": 0.5010612803834303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3696,
      "step": 14636
    },
    {
      "epoch": 0.5010955152345087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3781,
      "step": 14637
    },
    {
      "epoch": 0.5011297500855871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3591,
      "step": 14638
    },
    {
      "epoch": 0.5011639849366655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3784,
      "step": 14639
    },
    {
      "epoch": 0.5011982197877439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3665,
      "step": 14640
    },
    {
      "epoch": 0.5012324546388223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3663,
      "step": 14641
    },
    {
      "epoch": 0.5012666894899007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4052,
      "step": 14642
    },
    {
      "epoch": 0.5013009243409792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.391,
      "step": 14643
    },
    {
      "epoch": 0.5013351591920575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.367,
      "step": 14644
    },
    {
      "epoch": 0.501369394043136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3785,
      "step": 14645
    },
    {
      "epoch": 0.5014036288942143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3677,
      "step": 14646
    },
    {
      "epoch": 0.5014378637452928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3622,
      "step": 14647
    },
    {
      "epoch": 0.5014720985963711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3593,
      "step": 14648
    },
    {
      "epoch": 0.5015063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3795,
      "step": 14649
    },
    {
      "epoch": 0.5015405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3683,
      "step": 14650
    },
    {
      "epoch": 0.5015748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3776,
      "step": 14651
    },
    {
      "epoch": 0.5016090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3616,
      "step": 14652
    },
    {
      "epoch": 0.5016432728517631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3728,
      "step": 14653
    },
    {
      "epoch": 0.5016775077028415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3535,
      "step": 14654
    },
    {
      "epoch": 0.5017117425539199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3706,
      "step": 14655
    },
    {
      "epoch": 0.5017459774049983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3674,
      "step": 14656
    },
    {
      "epoch": 0.5017802122560767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3676,
      "step": 14657
    },
    {
      "epoch": 0.5018144471071551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3669,
      "step": 14658
    },
    {
      "epoch": 0.5018486819582335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3832,
      "step": 14659
    },
    {
      "epoch": 0.5018829168093119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3805,
      "step": 14660
    },
    {
      "epoch": 0.5019171516603903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3758,
      "step": 14661
    },
    {
      "epoch": 0.5019513865114686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3712,
      "step": 14662
    },
    {
      "epoch": 0.5019856213625471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3848,
      "step": 14663
    },
    {
      "epoch": 0.5020198562136254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3849,
      "step": 14664
    },
    {
      "epoch": 0.5020540910647039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3624,
      "step": 14665
    },
    {
      "epoch": 0.5020883259157822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3671,
      "step": 14666
    },
    {
      "epoch": 0.5021225607668607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3918,
      "step": 14667
    },
    {
      "epoch": 0.502156795617939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3877,
      "step": 14668
    },
    {
      "epoch": 0.5021910304690175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3768,
      "step": 14669
    },
    {
      "epoch": 0.5022252653200958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3779,
      "step": 14670
    },
    {
      "epoch": 0.5022595001711743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3764,
      "step": 14671
    },
    {
      "epoch": 0.5022937350222526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3858,
      "step": 14672
    },
    {
      "epoch": 0.5023279698733311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3758,
      "step": 14673
    },
    {
      "epoch": 0.5023622047244094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3841,
      "step": 14674
    },
    {
      "epoch": 0.5023964395754879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3784,
      "step": 14675
    },
    {
      "epoch": 0.5024306744265662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3781,
      "step": 14676
    },
    {
      "epoch": 0.5024649092776446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3851,
      "step": 14677
    },
    {
      "epoch": 0.502499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3929,
      "step": 14678
    },
    {
      "epoch": 0.5025333789798014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3872,
      "step": 14679
    },
    {
      "epoch": 0.5025676138308799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3582,
      "step": 14680
    },
    {
      "epoch": 0.5026018486819582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3834,
      "step": 14681
    },
    {
      "epoch": 0.5026360835330367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3693,
      "step": 14682
    },
    {
      "epoch": 0.502670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3879,
      "step": 14683
    },
    {
      "epoch": 0.5027045532351935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3807,
      "step": 14684
    },
    {
      "epoch": 0.5027387880862718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.357,
      "step": 14685
    },
    {
      "epoch": 0.5027730229373503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3828,
      "step": 14686
    },
    {
      "epoch": 0.5028072577884286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3658,
      "step": 14687
    },
    {
      "epoch": 0.5028414926395071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3809,
      "step": 14688
    },
    {
      "epoch": 0.5028757274905854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3773,
      "step": 14689
    },
    {
      "epoch": 0.5029099623416639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3539,
      "step": 14690
    },
    {
      "epoch": 0.5029441971927422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3618,
      "step": 14691
    },
    {
      "epoch": 0.5029784320438206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3752,
      "step": 14692
    },
    {
      "epoch": 0.503012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3561,
      "step": 14693
    },
    {
      "epoch": 0.5030469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3718,
      "step": 14694
    },
    {
      "epoch": 0.5030811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3692,
      "step": 14695
    },
    {
      "epoch": 0.5031153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3914,
      "step": 14696
    },
    {
      "epoch": 0.5031496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3889,
      "step": 14697
    },
    {
      "epoch": 0.503183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3623,
      "step": 14698
    },
    {
      "epoch": 0.5032180760013694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3826,
      "step": 14699
    },
    {
      "epoch": 0.5032523108524478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3761,
      "step": 14700
    },
    {
      "epoch": 0.5032865457035262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3703,
      "step": 14701
    },
    {
      "epoch": 0.5033207805546046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3756,
      "step": 14702
    },
    {
      "epoch": 0.503355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3834,
      "step": 14703
    },
    {
      "epoch": 0.5033892502567614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3859,
      "step": 14704
    },
    {
      "epoch": 0.5034234851078397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3809,
      "step": 14705
    },
    {
      "epoch": 0.5034577199589182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3685,
      "step": 14706
    },
    {
      "epoch": 0.5034919548099965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3502,
      "step": 14707
    },
    {
      "epoch": 0.503526189661075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3701,
      "step": 14708
    },
    {
      "epoch": 0.5035604245121533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3593,
      "step": 14709
    },
    {
      "epoch": 0.5035946593632318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3835,
      "step": 14710
    },
    {
      "epoch": 0.5036288942143101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3624,
      "step": 14711
    },
    {
      "epoch": 0.5036631290653886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.381,
      "step": 14712
    },
    {
      "epoch": 0.5036973639164669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3694,
      "step": 14713
    },
    {
      "epoch": 0.5037315987675454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3806,
      "step": 14714
    },
    {
      "epoch": 0.5037658336186237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3871,
      "step": 14715
    },
    {
      "epoch": 0.5038000684697022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3707,
      "step": 14716
    },
    {
      "epoch": 0.5038343033207806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3814,
      "step": 14717
    },
    {
      "epoch": 0.503868538171859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3545,
      "step": 14718
    },
    {
      "epoch": 0.5039027730229374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3707,
      "step": 14719
    },
    {
      "epoch": 0.5039370078740157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3864,
      "step": 14720
    },
    {
      "epoch": 0.5039712427250942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3788,
      "step": 14721
    },
    {
      "epoch": 0.5040054775761725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3731,
      "step": 14722
    },
    {
      "epoch": 0.504039712427251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3603,
      "step": 14723
    },
    {
      "epoch": 0.5040739472783293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3996,
      "step": 14724
    },
    {
      "epoch": 0.5041081821294078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3664,
      "step": 14725
    },
    {
      "epoch": 0.5041424169804861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3804,
      "step": 14726
    },
    {
      "epoch": 0.5041766518315646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3715,
      "step": 14727
    },
    {
      "epoch": 0.5042108866826429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3616,
      "step": 14728
    },
    {
      "epoch": 0.5042451215337214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3817,
      "step": 14729
    },
    {
      "epoch": 0.5042793563847997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3728,
      "step": 14730
    },
    {
      "epoch": 0.5043135912358782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3666,
      "step": 14731
    },
    {
      "epoch": 0.5043478260869565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3748,
      "step": 14732
    },
    {
      "epoch": 0.504382060938035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.355,
      "step": 14733
    },
    {
      "epoch": 0.5044162957891133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3604,
      "step": 14734
    },
    {
      "epoch": 0.5044505306401917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3785,
      "step": 14735
    },
    {
      "epoch": 0.5044847654912701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3922,
      "step": 14736
    },
    {
      "epoch": 0.5045190003423485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3771,
      "step": 14737
    },
    {
      "epoch": 0.5045532351934269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.365,
      "step": 14738
    },
    {
      "epoch": 0.5045874700445053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3563,
      "step": 14739
    },
    {
      "epoch": 0.5046217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3698,
      "step": 14740
    },
    {
      "epoch": 0.5046559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3644,
      "step": 14741
    },
    {
      "epoch": 0.5046901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.377,
      "step": 14742
    },
    {
      "epoch": 0.5047244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3677,
      "step": 14743
    },
    {
      "epoch": 0.5047586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3606,
      "step": 14744
    },
    {
      "epoch": 0.5047928791509757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3825,
      "step": 14745
    },
    {
      "epoch": 0.504827114002054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3603,
      "step": 14746
    },
    {
      "epoch": 0.5048613488531325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3824,
      "step": 14747
    },
    {
      "epoch": 0.5048955837042108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3843,
      "step": 14748
    },
    {
      "epoch": 0.5049298185552893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3649,
      "step": 14749
    },
    {
      "epoch": 0.5049640534063676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3636,
      "step": 14750
    },
    {
      "epoch": 0.5049982882574461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3724,
      "step": 14751
    },
    {
      "epoch": 0.5050325231085244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3665,
      "step": 14752
    },
    {
      "epoch": 0.5050667579596029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3752,
      "step": 14753
    },
    {
      "epoch": 0.5051009928106813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3635,
      "step": 14754
    },
    {
      "epoch": 0.5051352276617597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3755,
      "step": 14755
    },
    {
      "epoch": 0.5051694625128381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3507,
      "step": 14756
    },
    {
      "epoch": 0.5052036973639165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.382,
      "step": 14757
    },
    {
      "epoch": 0.5052379322149949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3729,
      "step": 14758
    },
    {
      "epoch": 0.5052721670660733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3601,
      "step": 14759
    },
    {
      "epoch": 0.5053064019171517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3657,
      "step": 14760
    },
    {
      "epoch": 0.50534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3821,
      "step": 14761
    },
    {
      "epoch": 0.5053748716193085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3809,
      "step": 14762
    },
    {
      "epoch": 0.5054091064703868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3604,
      "step": 14763
    },
    {
      "epoch": 0.5054433413214653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.38,
      "step": 14764
    },
    {
      "epoch": 0.5054775761725436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3603,
      "step": 14765
    },
    {
      "epoch": 0.5055118110236221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3421,
      "step": 14766
    },
    {
      "epoch": 0.5055460458747004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.359,
      "step": 14767
    },
    {
      "epoch": 0.5055802807257789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3644,
      "step": 14768
    },
    {
      "epoch": 0.5056145155768572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3752,
      "step": 14769
    },
    {
      "epoch": 0.5056487504279357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3523,
      "step": 14770
    },
    {
      "epoch": 0.505682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3607,
      "step": 14771
    },
    {
      "epoch": 0.5057172201300925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3557,
      "step": 14772
    },
    {
      "epoch": 0.5057514549811708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3733,
      "step": 14773
    },
    {
      "epoch": 0.5057856898322493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3863,
      "step": 14774
    },
    {
      "epoch": 0.5058199246833276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.363,
      "step": 14775
    },
    {
      "epoch": 0.505854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3817,
      "step": 14776
    },
    {
      "epoch": 0.5058883943854844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3718,
      "step": 14777
    },
    {
      "epoch": 0.5059226292365628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3744,
      "step": 14778
    },
    {
      "epoch": 0.5059568640876412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3743,
      "step": 14779
    },
    {
      "epoch": 0.5059910989387196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3355,
      "step": 14780
    },
    {
      "epoch": 0.506025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3756,
      "step": 14781
    },
    {
      "epoch": 0.5060595686408764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.359,
      "step": 14782
    },
    {
      "epoch": 0.5060938034919548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3689,
      "step": 14783
    },
    {
      "epoch": 0.5061280383430332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3561,
      "step": 14784
    },
    {
      "epoch": 0.5061622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3506,
      "step": 14785
    },
    {
      "epoch": 0.50619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3709,
      "step": 14786
    },
    {
      "epoch": 0.5062307428962683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3507,
      "step": 14787
    },
    {
      "epoch": 0.5062649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3683,
      "step": 14788
    },
    {
      "epoch": 0.5062992125984253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3731,
      "step": 14789
    },
    {
      "epoch": 0.5063334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3633,
      "step": 14790
    },
    {
      "epoch": 0.506367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3599,
      "step": 14791
    },
    {
      "epoch": 0.5064019171516604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3682,
      "step": 14792
    },
    {
      "epoch": 0.5064361520027388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3882,
      "step": 14793
    },
    {
      "epoch": 0.5064703868538172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3652,
      "step": 14794
    },
    {
      "epoch": 0.5065046217048956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3696,
      "step": 14795
    },
    {
      "epoch": 0.506538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3784,
      "step": 14796
    },
    {
      "epoch": 0.5065730914070524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3603,
      "step": 14797
    },
    {
      "epoch": 0.5066073262581308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3692,
      "step": 14798
    },
    {
      "epoch": 0.5066415611092092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3557,
      "step": 14799
    },
    {
      "epoch": 0.5066757959602876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3652,
      "step": 14800
    },
    {
      "epoch": 0.506710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3763,
      "step": 14801
    },
    {
      "epoch": 0.5067442656624443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3738,
      "step": 14802
    },
    {
      "epoch": 0.5067785005135228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3695,
      "step": 14803
    },
    {
      "epoch": 0.5068127353646011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3663,
      "step": 14804
    },
    {
      "epoch": 0.5068469702156796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3837,
      "step": 14805
    },
    {
      "epoch": 0.5068812050667579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3741,
      "step": 14806
    },
    {
      "epoch": 0.5069154399178364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3723,
      "step": 14807
    },
    {
      "epoch": 0.5069496747689147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3815,
      "step": 14808
    },
    {
      "epoch": 0.5069839096199932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3681,
      "step": 14809
    },
    {
      "epoch": 0.5070181444710715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3762,
      "step": 14810
    },
    {
      "epoch": 0.50705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4,
      "step": 14811
    },
    {
      "epoch": 0.5070866141732283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3752,
      "step": 14812
    },
    {
      "epoch": 0.5071208490243068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3715,
      "step": 14813
    },
    {
      "epoch": 0.5071550838753851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3724,
      "step": 14814
    },
    {
      "epoch": 0.5071893187264636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3512,
      "step": 14815
    },
    {
      "epoch": 0.5072235535775419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3645,
      "step": 14816
    },
    {
      "epoch": 0.5072577884286203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3732,
      "step": 14817
    },
    {
      "epoch": 0.5072920232796987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3645,
      "step": 14818
    },
    {
      "epoch": 0.5073262581307771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.376,
      "step": 14819
    },
    {
      "epoch": 0.5073604929818555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3801,
      "step": 14820
    },
    {
      "epoch": 0.5073947278329339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3752,
      "step": 14821
    },
    {
      "epoch": 0.5074289626840123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3842,
      "step": 14822
    },
    {
      "epoch": 0.5074631975350907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3635,
      "step": 14823
    },
    {
      "epoch": 0.5074974323861691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3905,
      "step": 14824
    },
    {
      "epoch": 0.5075316672372475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3581,
      "step": 14825
    },
    {
      "epoch": 0.507565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3665,
      "step": 14826
    },
    {
      "epoch": 0.5076001369394043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3745,
      "step": 14827
    },
    {
      "epoch": 0.5076343717904828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3844,
      "step": 14828
    },
    {
      "epoch": 0.5076686066415611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3538,
      "step": 14829
    },
    {
      "epoch": 0.5077028414926396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3611,
      "step": 14830
    },
    {
      "epoch": 0.5077370763437179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3662,
      "step": 14831
    },
    {
      "epoch": 0.5077713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3464,
      "step": 14832
    },
    {
      "epoch": 0.5078055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3512,
      "step": 14833
    },
    {
      "epoch": 0.5078397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3599,
      "step": 14834
    },
    {
      "epoch": 0.5078740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3906,
      "step": 14835
    },
    {
      "epoch": 0.5079082505991099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.372,
      "step": 14836
    },
    {
      "epoch": 0.5079424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3745,
      "step": 14837
    },
    {
      "epoch": 0.5079767203012667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3612,
      "step": 14838
    },
    {
      "epoch": 0.5080109551523451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3564,
      "step": 14839
    },
    {
      "epoch": 0.5080451900034235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3699,
      "step": 14840
    },
    {
      "epoch": 0.5080794248545019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3549,
      "step": 14841
    },
    {
      "epoch": 0.5081136597055803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3759,
      "step": 14842
    },
    {
      "epoch": 0.5081478945566587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3821,
      "step": 14843
    },
    {
      "epoch": 0.5081821294077371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3651,
      "step": 14844
    },
    {
      "epoch": 0.5082163642588154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3626,
      "step": 14845
    },
    {
      "epoch": 0.5082505991098939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3492,
      "step": 14846
    },
    {
      "epoch": 0.5082848339609722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3644,
      "step": 14847
    },
    {
      "epoch": 0.5083190688120507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3842,
      "step": 14848
    },
    {
      "epoch": 0.508353303663129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3604,
      "step": 14849
    },
    {
      "epoch": 0.5083875385142075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3593,
      "step": 14850
    },
    {
      "epoch": 0.5084217733652858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3438,
      "step": 14851
    },
    {
      "epoch": 0.5084560082163643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3653,
      "step": 14852
    },
    {
      "epoch": 0.5084902430674426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3673,
      "step": 14853
    },
    {
      "epoch": 0.5085244779185211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3839,
      "step": 14854
    },
    {
      "epoch": 0.5085587127695994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3433,
      "step": 14855
    },
    {
      "epoch": 0.5085929476206779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3563,
      "step": 14856
    },
    {
      "epoch": 0.5086271824717562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3571,
      "step": 14857
    },
    {
      "epoch": 0.5086614173228347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3688,
      "step": 14858
    },
    {
      "epoch": 0.508695652173913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3558,
      "step": 14859
    },
    {
      "epoch": 0.5087298870249914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.378,
      "step": 14860
    },
    {
      "epoch": 0.5087641218760698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3807,
      "step": 14861
    },
    {
      "epoch": 0.5087983567271482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3917,
      "step": 14862
    },
    {
      "epoch": 0.5088325915782267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3452,
      "step": 14863
    },
    {
      "epoch": 0.508866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3504,
      "step": 14864
    },
    {
      "epoch": 0.5089010612803835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3668,
      "step": 14865
    },
    {
      "epoch": 0.5089352961314618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3612,
      "step": 14866
    },
    {
      "epoch": 0.5089695309825403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3555,
      "step": 14867
    },
    {
      "epoch": 0.5090037658336186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3816,
      "step": 14868
    },
    {
      "epoch": 0.5090380006846971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.4004,
      "step": 14869
    },
    {
      "epoch": 0.5090722355357754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3764,
      "step": 14870
    },
    {
      "epoch": 0.5091064703868539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3995,
      "step": 14871
    },
    {
      "epoch": 0.5091407052379322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3772,
      "step": 14872
    },
    {
      "epoch": 0.5091749400890107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.364,
      "step": 14873
    },
    {
      "epoch": 0.509209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3769,
      "step": 14874
    },
    {
      "epoch": 0.5092434097911674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3626,
      "step": 14875
    },
    {
      "epoch": 0.5092776446422458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3664,
      "step": 14876
    },
    {
      "epoch": 0.5093118794933242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3715,
      "step": 14877
    },
    {
      "epoch": 0.5093461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3718,
      "step": 14878
    },
    {
      "epoch": 0.509380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3803,
      "step": 14879
    },
    {
      "epoch": 0.5094145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3827,
      "step": 14880
    },
    {
      "epoch": 0.5094488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3779,
      "step": 14881
    },
    {
      "epoch": 0.5094830537487162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.347,
      "step": 14882
    },
    {
      "epoch": 0.5095172885997946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3827,
      "step": 14883
    },
    {
      "epoch": 0.509551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3637,
      "step": 14884
    },
    {
      "epoch": 0.5095857583019514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3546,
      "step": 14885
    },
    {
      "epoch": 0.5096199931530297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3827,
      "step": 14886
    },
    {
      "epoch": 0.5096542280041082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3771,
      "step": 14887
    },
    {
      "epoch": 0.5096884628551865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3685,
      "step": 14888
    },
    {
      "epoch": 0.509722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3469,
      "step": 14889
    },
    {
      "epoch": 0.5097569325573433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3564,
      "step": 14890
    },
    {
      "epoch": 0.5097911674084218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3777,
      "step": 14891
    },
    {
      "epoch": 0.5098254022595001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3771,
      "step": 14892
    },
    {
      "epoch": 0.5098596371105786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3754,
      "step": 14893
    },
    {
      "epoch": 0.5098938719616569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3622,
      "step": 14894
    },
    {
      "epoch": 0.5099281068127354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3745,
      "step": 14895
    },
    {
      "epoch": 0.5099623416638137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.373,
      "step": 14896
    },
    {
      "epoch": 0.5099965765148922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3652,
      "step": 14897
    },
    {
      "epoch": 0.5100308113659705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3562,
      "step": 14898
    },
    {
      "epoch": 0.510065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3798,
      "step": 14899
    },
    {
      "epoch": 0.5100992810681274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3604,
      "step": 14900
    },
    {
      "epoch": 0.5101335159192057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.373,
      "step": 14901
    },
    {
      "epoch": 0.5101677507702842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3623,
      "step": 14902
    },
    {
      "epoch": 0.5102019856213625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3712,
      "step": 14903
    },
    {
      "epoch": 0.510236220472441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3624,
      "step": 14904
    },
    {
      "epoch": 0.5102704553235193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3743,
      "step": 14905
    },
    {
      "epoch": 0.5103046901745978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3705,
      "step": 14906
    },
    {
      "epoch": 0.5103389250256761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3659,
      "step": 14907
    },
    {
      "epoch": 0.5103731598767546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3749,
      "step": 14908
    },
    {
      "epoch": 0.5104073947278329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3607,
      "step": 14909
    },
    {
      "epoch": 0.5104416295789114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3605,
      "step": 14910
    },
    {
      "epoch": 0.5104758644299897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3528,
      "step": 14911
    },
    {
      "epoch": 0.5105100992810682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3878,
      "step": 14912
    },
    {
      "epoch": 0.5105443341321465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3702,
      "step": 14913
    },
    {
      "epoch": 0.510578568983225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3637,
      "step": 14914
    },
    {
      "epoch": 0.5106128038343033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3787,
      "step": 14915
    },
    {
      "epoch": 0.5106470386853817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3519,
      "step": 14916
    },
    {
      "epoch": 0.5106812735364601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3524,
      "step": 14917
    },
    {
      "epoch": 0.5107155083875385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3687,
      "step": 14918
    },
    {
      "epoch": 0.5107497432386169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3824,
      "step": 14919
    },
    {
      "epoch": 0.5107839780896953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3762,
      "step": 14920
    },
    {
      "epoch": 0.5108182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3529,
      "step": 14921
    },
    {
      "epoch": 0.5108524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.361,
      "step": 14922
    },
    {
      "epoch": 0.5108866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3588,
      "step": 14923
    },
    {
      "epoch": 0.5109209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3679,
      "step": 14924
    },
    {
      "epoch": 0.5109551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.372,
      "step": 14925
    },
    {
      "epoch": 0.5109893871961657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3647,
      "step": 14926
    },
    {
      "epoch": 0.511023622047244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3691,
      "step": 14927
    },
    {
      "epoch": 0.5110578568983225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3626,
      "step": 14928
    },
    {
      "epoch": 0.5110920917494008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3626,
      "step": 14929
    },
    {
      "epoch": 0.5111263266004793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3688,
      "step": 14930
    },
    {
      "epoch": 0.5111605614515576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3796,
      "step": 14931
    },
    {
      "epoch": 0.5111947963026361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3513,
      "step": 14932
    },
    {
      "epoch": 0.5112290311537144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3687,
      "step": 14933
    },
    {
      "epoch": 0.5112632660047929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3771,
      "step": 14934
    },
    {
      "epoch": 0.5112975008558712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3517,
      "step": 14935
    },
    {
      "epoch": 0.5113317357069497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.361,
      "step": 14936
    },
    {
      "epoch": 0.5113659705580281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3732,
      "step": 14937
    },
    {
      "epoch": 0.5114002054091065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3673,
      "step": 14938
    },
    {
      "epoch": 0.5114344402601849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3679,
      "step": 14939
    },
    {
      "epoch": 0.5114686751112633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3751,
      "step": 14940
    },
    {
      "epoch": 0.5115029099623417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3594,
      "step": 14941
    },
    {
      "epoch": 0.51153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3564,
      "step": 14942
    },
    {
      "epoch": 0.5115713796644985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3554,
      "step": 14943
    },
    {
      "epoch": 0.5116056145155768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3679,
      "step": 14944
    },
    {
      "epoch": 0.5116398493666553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3715,
      "step": 14945
    },
    {
      "epoch": 0.5116740842177336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3619,
      "step": 14946
    },
    {
      "epoch": 0.5117083190688121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3525,
      "step": 14947
    },
    {
      "epoch": 0.5117425539198904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3703,
      "step": 14948
    },
    {
      "epoch": 0.5117767887709689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3643,
      "step": 14949
    },
    {
      "epoch": 0.5118110236220472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3657,
      "step": 14950
    },
    {
      "epoch": 0.5118452584731257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3693,
      "step": 14951
    },
    {
      "epoch": 0.511879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3769,
      "step": 14952
    },
    {
      "epoch": 0.5119137281752825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3616,
      "step": 14953
    },
    {
      "epoch": 0.5119479630263608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3635,
      "step": 14954
    },
    {
      "epoch": 0.5119821978774393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.373,
      "step": 14955
    },
    {
      "epoch": 0.5120164327285176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3626,
      "step": 14956
    },
    {
      "epoch": 0.512050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3829,
      "step": 14957
    },
    {
      "epoch": 0.5120849024306744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3661,
      "step": 14958
    },
    {
      "epoch": 0.5121191372817528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3484,
      "step": 14959
    },
    {
      "epoch": 0.5121533721328312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3797,
      "step": 14960
    },
    {
      "epoch": 0.5121876069839096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3778,
      "step": 14961
    },
    {
      "epoch": 0.512221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3546,
      "step": 14962
    },
    {
      "epoch": 0.5122560766860664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.363,
      "step": 14963
    },
    {
      "epoch": 0.5122903115371448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3788,
      "step": 14964
    },
    {
      "epoch": 0.5123245463882232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.377,
      "step": 14965
    },
    {
      "epoch": 0.5123587812393016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3796,
      "step": 14966
    },
    {
      "epoch": 0.51239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3737,
      "step": 14967
    },
    {
      "epoch": 0.5124272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3563,
      "step": 14968
    },
    {
      "epoch": 0.5124614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3557,
      "step": 14969
    },
    {
      "epoch": 0.5124957206436151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3735,
      "step": 14970
    },
    {
      "epoch": 0.5125299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3709,
      "step": 14971
    },
    {
      "epoch": 0.5125641903457719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3635,
      "step": 14972
    },
    {
      "epoch": 0.5125984251968504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3823,
      "step": 14973
    },
    {
      "epoch": 0.5126326600479288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3713,
      "step": 14974
    },
    {
      "epoch": 0.5126668948990072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3511,
      "step": 14975
    },
    {
      "epoch": 0.5127011297500856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3761,
      "step": 14976
    },
    {
      "epoch": 0.512735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3425,
      "step": 14977
    },
    {
      "epoch": 0.5127695994522424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3433,
      "step": 14978
    },
    {
      "epoch": 0.5128038343033208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3697,
      "step": 14979
    },
    {
      "epoch": 0.5128380691543992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3577,
      "step": 14980
    },
    {
      "epoch": 0.5128723040054776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3473,
      "step": 14981
    },
    {
      "epoch": 0.512906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3649,
      "step": 14982
    },
    {
      "epoch": 0.5129407737076344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3608,
      "step": 14983
    },
    {
      "epoch": 0.5129750085587128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3688,
      "step": 14984
    },
    {
      "epoch": 0.5130092434097911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3535,
      "step": 14985
    },
    {
      "epoch": 0.5130434782608696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3702,
      "step": 14986
    },
    {
      "epoch": 0.5130777131119479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3638,
      "step": 14987
    },
    {
      "epoch": 0.5131119479630264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.363,
      "step": 14988
    },
    {
      "epoch": 0.5131461828141047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3629,
      "step": 14989
    },
    {
      "epoch": 0.5131804176651832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3759,
      "step": 14990
    },
    {
      "epoch": 0.5132146525162615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3491,
      "step": 14991
    },
    {
      "epoch": 0.51324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3463,
      "step": 14992
    },
    {
      "epoch": 0.5132831222184183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3428,
      "step": 14993
    },
    {
      "epoch": 0.5133173570694968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3514,
      "step": 14994
    },
    {
      "epoch": 0.5133515919205751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3542,
      "step": 14995
    },
    {
      "epoch": 0.5133858267716536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3662,
      "step": 14996
    },
    {
      "epoch": 0.5134200616227319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3508,
      "step": 14997
    },
    {
      "epoch": 0.5134542964738104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3739,
      "step": 14998
    },
    {
      "epoch": 0.5134885313248887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3593,
      "step": 14999
    },
    {
      "epoch": 0.5135227661759671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3668,
      "step": 15000
    },
    {
      "epoch": 0.5135570010270455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3599,
      "step": 15001
    },
    {
      "epoch": 0.5135912358781239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.371,
      "step": 15002
    },
    {
      "epoch": 0.5136254707292023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3671,
      "step": 15003
    },
    {
      "epoch": 0.5136597055802807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.361,
      "step": 15004
    },
    {
      "epoch": 0.5136939404313591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3432,
      "step": 15005
    },
    {
      "epoch": 0.5137281752824375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3723,
      "step": 15006
    },
    {
      "epoch": 0.5137624101335159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3801,
      "step": 15007
    },
    {
      "epoch": 0.5137966449845943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.364,
      "step": 15008
    },
    {
      "epoch": 0.5138308798356727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3683,
      "step": 15009
    },
    {
      "epoch": 0.5138651146867511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3748,
      "step": 15010
    },
    {
      "epoch": 0.5138993495378296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3777,
      "step": 15011
    },
    {
      "epoch": 0.5139335843889079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3611,
      "step": 15012
    },
    {
      "epoch": 0.5139678192399864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3564,
      "step": 15013
    },
    {
      "epoch": 0.5140020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3546,
      "step": 15014
    },
    {
      "epoch": 0.5140362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3664,
      "step": 15015
    },
    {
      "epoch": 0.5140705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3489,
      "step": 15016
    },
    {
      "epoch": 0.5141047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3644,
      "step": 15017
    },
    {
      "epoch": 0.5141389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.363,
      "step": 15018
    },
    {
      "epoch": 0.5141732283464567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3493,
      "step": 15019
    },
    {
      "epoch": 0.5142074631975351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3727,
      "step": 15020
    },
    {
      "epoch": 0.5142416980486135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3609,
      "step": 15021
    },
    {
      "epoch": 0.5142759328996919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3662,
      "step": 15022
    },
    {
      "epoch": 0.5143101677507703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3574,
      "step": 15023
    },
    {
      "epoch": 0.5143444026018487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.374,
      "step": 15024
    },
    {
      "epoch": 0.5143786374529271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3643,
      "step": 15025
    },
    {
      "epoch": 0.5144128723040055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3616,
      "step": 15026
    },
    {
      "epoch": 0.5144471071550839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3505,
      "step": 15027
    },
    {
      "epoch": 0.5144813420061622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3571,
      "step": 15028
    },
    {
      "epoch": 0.5145155768572407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3661,
      "step": 15029
    },
    {
      "epoch": 0.514549811708319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3648,
      "step": 15030
    },
    {
      "epoch": 0.5145840465593975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3645,
      "step": 15031
    },
    {
      "epoch": 0.5146182814104758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3533,
      "step": 15032
    },
    {
      "epoch": 0.5146525162615543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3729,
      "step": 15033
    },
    {
      "epoch": 0.5146867511126326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3743,
      "step": 15034
    },
    {
      "epoch": 0.5147209859637111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3596,
      "step": 15035
    },
    {
      "epoch": 0.5147552208147894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.346,
      "step": 15036
    },
    {
      "epoch": 0.5147894556658679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3757,
      "step": 15037
    },
    {
      "epoch": 0.5148236905169462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3588,
      "step": 15038
    },
    {
      "epoch": 0.5148579253680247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3441,
      "step": 15039
    },
    {
      "epoch": 0.514892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3579,
      "step": 15040
    },
    {
      "epoch": 0.5149263950701815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3574,
      "step": 15041
    },
    {
      "epoch": 0.5149606299212598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3673,
      "step": 15042
    },
    {
      "epoch": 0.5149948647723382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3827,
      "step": 15043
    },
    {
      "epoch": 0.5150290996234166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3525,
      "step": 15044
    },
    {
      "epoch": 0.515063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3679,
      "step": 15045
    },
    {
      "epoch": 0.5150975693255734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3497,
      "step": 15046
    },
    {
      "epoch": 0.5151318041766518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3793,
      "step": 15047
    },
    {
      "epoch": 0.5151660390277303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3613,
      "step": 15048
    },
    {
      "epoch": 0.5152002738788086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.348,
      "step": 15049
    },
    {
      "epoch": 0.5152345087298871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3693,
      "step": 15050
    },
    {
      "epoch": 0.5152687435809654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3737,
      "step": 15051
    },
    {
      "epoch": 0.5153029784320439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3753,
      "step": 15052
    },
    {
      "epoch": 0.5153372132831222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3702,
      "step": 15053
    },
    {
      "epoch": 0.5153714481342007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.342,
      "step": 15054
    },
    {
      "epoch": 0.515405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3525,
      "step": 15055
    },
    {
      "epoch": 0.5154399178363575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3646,
      "step": 15056
    },
    {
      "epoch": 0.5154741526874358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3476,
      "step": 15057
    },
    {
      "epoch": 0.5155083875385142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3569,
      "step": 15058
    },
    {
      "epoch": 0.5155426223895926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3654,
      "step": 15059
    },
    {
      "epoch": 0.515576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3564,
      "step": 15060
    },
    {
      "epoch": 0.5156110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3681,
      "step": 15061
    },
    {
      "epoch": 0.5156453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3579,
      "step": 15062
    },
    {
      "epoch": 0.5156795617939062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3673,
      "step": 15063
    },
    {
      "epoch": 0.5157137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3701,
      "step": 15064
    },
    {
      "epoch": 0.515748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3694,
      "step": 15065
    },
    {
      "epoch": 0.5157822663471414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3817,
      "step": 15066
    },
    {
      "epoch": 0.5158165011982198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3772,
      "step": 15067
    },
    {
      "epoch": 0.5158507360492982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3583,
      "step": 15068
    },
    {
      "epoch": 0.5158849709003765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3814,
      "step": 15069
    },
    {
      "epoch": 0.515919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3513,
      "step": 15070
    },
    {
      "epoch": 0.5159534406025333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3742,
      "step": 15071
    },
    {
      "epoch": 0.5159876754536118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3491,
      "step": 15072
    },
    {
      "epoch": 0.5160219103046901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3541,
      "step": 15073
    },
    {
      "epoch": 0.5160561451557686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3638,
      "step": 15074
    },
    {
      "epoch": 0.5160903800068469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3736,
      "step": 15075
    },
    {
      "epoch": 0.5161246148579254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3577,
      "step": 15076
    },
    {
      "epoch": 0.5161588497090037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3538,
      "step": 15077
    },
    {
      "epoch": 0.5161930845600822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3441,
      "step": 15078
    },
    {
      "epoch": 0.5162273194111605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3591,
      "step": 15079
    },
    {
      "epoch": 0.516261554262239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3667,
      "step": 15080
    },
    {
      "epoch": 0.5162957891133173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3635,
      "step": 15081
    },
    {
      "epoch": 0.5163300239643958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3475,
      "step": 15082
    },
    {
      "epoch": 0.5163642588154741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3665,
      "step": 15083
    },
    {
      "epoch": 0.5163984936665525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3694,
      "step": 15084
    },
    {
      "epoch": 0.516432728517631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3652,
      "step": 15085
    },
    {
      "epoch": 0.5164669633687093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3306,
      "step": 15086
    },
    {
      "epoch": 0.5165011982197878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3741,
      "step": 15087
    },
    {
      "epoch": 0.5165354330708661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3568,
      "step": 15088
    },
    {
      "epoch": 0.5165696679219446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3561,
      "step": 15089
    },
    {
      "epoch": 0.5166039027730229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3477,
      "step": 15090
    },
    {
      "epoch": 0.5166381376241014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3725,
      "step": 15091
    },
    {
      "epoch": 0.5166723724751797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3682,
      "step": 15092
    },
    {
      "epoch": 0.5167066073262582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.353,
      "step": 15093
    },
    {
      "epoch": 0.5167408421773365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3636,
      "step": 15094
    },
    {
      "epoch": 0.516775077028415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3441,
      "step": 15095
    },
    {
      "epoch": 0.5168093118794933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3467,
      "step": 15096
    },
    {
      "epoch": 0.5168435467305718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3317,
      "step": 15097
    },
    {
      "epoch": 0.5168777815816501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3705,
      "step": 15098
    },
    {
      "epoch": 0.5169120164327285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3579,
      "step": 15099
    },
    {
      "epoch": 0.5169462512838069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3641,
      "step": 15100
    },
    {
      "epoch": 0.5169804861348853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3643,
      "step": 15101
    },
    {
      "epoch": 0.5170147209859637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3597,
      "step": 15102
    },
    {
      "epoch": 0.5170489558370421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3646,
      "step": 15103
    },
    {
      "epoch": 0.5170831906881205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3561,
      "step": 15104
    },
    {
      "epoch": 0.5171174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3695,
      "step": 15105
    },
    {
      "epoch": 0.5171516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3754,
      "step": 15106
    },
    {
      "epoch": 0.5171858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3698,
      "step": 15107
    },
    {
      "epoch": 0.5172201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3808,
      "step": 15108
    },
    {
      "epoch": 0.5172543649435125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3508,
      "step": 15109
    },
    {
      "epoch": 0.5172885997945909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3683,
      "step": 15110
    },
    {
      "epoch": 0.5173228346456693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3682,
      "step": 15111
    },
    {
      "epoch": 0.5173570694967476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3649,
      "step": 15112
    },
    {
      "epoch": 0.5173913043478261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3448,
      "step": 15113
    },
    {
      "epoch": 0.5174255391989044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3604,
      "step": 15114
    },
    {
      "epoch": 0.5174597740499829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3613,
      "step": 15115
    },
    {
      "epoch": 0.5174940089010612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3488,
      "step": 15116
    },
    {
      "epoch": 0.5175282437521397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3678,
      "step": 15117
    },
    {
      "epoch": 0.517562478603218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3666,
      "step": 15118
    },
    {
      "epoch": 0.5175967134542965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3552,
      "step": 15119
    },
    {
      "epoch": 0.5176309483053748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3683,
      "step": 15120
    },
    {
      "epoch": 0.5176651831564533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3732,
      "step": 15121
    },
    {
      "epoch": 0.5176994180075317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3654,
      "step": 15122
    },
    {
      "epoch": 0.5177336528586101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3571,
      "step": 15123
    },
    {
      "epoch": 0.5177678877096885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3597,
      "step": 15124
    },
    {
      "epoch": 0.5178021225607669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3616,
      "step": 15125
    },
    {
      "epoch": 0.5178363574118453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3418,
      "step": 15126
    },
    {
      "epoch": 0.5178705922629236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.359,
      "step": 15127
    },
    {
      "epoch": 0.5179048271140021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3502,
      "step": 15128
    },
    {
      "epoch": 0.5179390619650804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3597,
      "step": 15129
    },
    {
      "epoch": 0.5179732968161589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3657,
      "step": 15130
    },
    {
      "epoch": 0.5180075316672372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3579,
      "step": 15131
    },
    {
      "epoch": 0.5180417665183157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3536,
      "step": 15132
    },
    {
      "epoch": 0.518076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.35,
      "step": 15133
    },
    {
      "epoch": 0.5181102362204725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3458,
      "step": 15134
    },
    {
      "epoch": 0.5181444710715508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3436,
      "step": 15135
    },
    {
      "epoch": 0.5181787059226293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3502,
      "step": 15136
    },
    {
      "epoch": 0.5182129407737076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3577,
      "step": 15137
    },
    {
      "epoch": 0.5182471756247861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3613,
      "step": 15138
    },
    {
      "epoch": 0.5182814104758644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3702,
      "step": 15139
    },
    {
      "epoch": 0.5183156453269429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3638,
      "step": 15140
    },
    {
      "epoch": 0.5183498801780212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3664,
      "step": 15141
    },
    {
      "epoch": 0.5183841150290996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3499,
      "step": 15142
    },
    {
      "epoch": 0.518418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3461,
      "step": 15143
    },
    {
      "epoch": 0.5184525847312564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3699,
      "step": 15144
    },
    {
      "epoch": 0.5184868195823348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3492,
      "step": 15145
    },
    {
      "epoch": 0.5185210544334132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3511,
      "step": 15146
    },
    {
      "epoch": 0.5185552892844916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.365,
      "step": 15147
    },
    {
      "epoch": 0.51858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3674,
      "step": 15148
    },
    {
      "epoch": 0.5186237589866484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3621,
      "step": 15149
    },
    {
      "epoch": 0.5186579938377268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3599,
      "step": 15150
    },
    {
      "epoch": 0.5186922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3521,
      "step": 15151
    },
    {
      "epoch": 0.5187264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3666,
      "step": 15152
    },
    {
      "epoch": 0.518760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3449,
      "step": 15153
    },
    {
      "epoch": 0.5187949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3677,
      "step": 15154
    },
    {
      "epoch": 0.5188291680931187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3637,
      "step": 15155
    },
    {
      "epoch": 0.5188634029441972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.364,
      "step": 15156
    },
    {
      "epoch": 0.5188976377952756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3393,
      "step": 15157
    },
    {
      "epoch": 0.518931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3559,
      "step": 15158
    },
    {
      "epoch": 0.5189661074974324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3638,
      "step": 15159
    },
    {
      "epoch": 0.5190003423485108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3758,
      "step": 15160
    },
    {
      "epoch": 0.5190345771995892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3722,
      "step": 15161
    },
    {
      "epoch": 0.5190688120506676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3317,
      "step": 15162
    },
    {
      "epoch": 0.519103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3648,
      "step": 15163
    },
    {
      "epoch": 0.5191372817528244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.343,
      "step": 15164
    },
    {
      "epoch": 0.5191715166039028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.348,
      "step": 15165
    },
    {
      "epoch": 0.5192057514549812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3522,
      "step": 15166
    },
    {
      "epoch": 0.5192399863060596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3799,
      "step": 15167
    },
    {
      "epoch": 0.519274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3663,
      "step": 15168
    },
    {
      "epoch": 0.5193084560082164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3674,
      "step": 15169
    },
    {
      "epoch": 0.5193426908592947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3463,
      "step": 15170
    },
    {
      "epoch": 0.5193769257103732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.35,
      "step": 15171
    },
    {
      "epoch": 0.5194111605614515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3754,
      "step": 15172
    },
    {
      "epoch": 0.51944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3496,
      "step": 15173
    },
    {
      "epoch": 0.5194796302636083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3528,
      "step": 15174
    },
    {
      "epoch": 0.5195138651146868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3483,
      "step": 15175
    },
    {
      "epoch": 0.5195480999657651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3616,
      "step": 15176
    },
    {
      "epoch": 0.5195823348168436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3491,
      "step": 15177
    },
    {
      "epoch": 0.5196165696679219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3439,
      "step": 15178
    },
    {
      "epoch": 0.5196508045190004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3546,
      "step": 15179
    },
    {
      "epoch": 0.5196850393700787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3855,
      "step": 15180
    },
    {
      "epoch": 0.5197192742211572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3619,
      "step": 15181
    },
    {
      "epoch": 0.5197535090722355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3709,
      "step": 15182
    },
    {
      "epoch": 0.519787743923314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3478,
      "step": 15183
    },
    {
      "epoch": 0.5198219787743923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3459,
      "step": 15184
    },
    {
      "epoch": 0.5198562136254707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.353,
      "step": 15185
    },
    {
      "epoch": 0.5198904484765491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3609,
      "step": 15186
    },
    {
      "epoch": 0.5199246833276275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3273,
      "step": 15187
    },
    {
      "epoch": 0.5199589181787059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3485,
      "step": 15188
    },
    {
      "epoch": 0.5199931530297843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3506,
      "step": 15189
    },
    {
      "epoch": 0.5200273878808627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3543,
      "step": 15190
    },
    {
      "epoch": 0.5200616227319411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.353,
      "step": 15191
    },
    {
      "epoch": 0.5200958575830195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3431,
      "step": 15192
    },
    {
      "epoch": 0.5201300924340979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3481,
      "step": 15193
    },
    {
      "epoch": 0.5201643272851764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3637,
      "step": 15194
    },
    {
      "epoch": 0.5201985621362547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3523,
      "step": 15195
    },
    {
      "epoch": 0.5202327969873332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3582,
      "step": 15196
    },
    {
      "epoch": 0.5202670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3372,
      "step": 15197
    },
    {
      "epoch": 0.52030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3594,
      "step": 15198
    },
    {
      "epoch": 0.5203355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3557,
      "step": 15199
    },
    {
      "epoch": 0.5203697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3429,
      "step": 15200
    },
    {
      "epoch": 0.5204039712427251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3356,
      "step": 15201
    },
    {
      "epoch": 0.5204382060938035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3593,
      "step": 15202
    },
    {
      "epoch": 0.5204724409448819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3575,
      "step": 15203
    },
    {
      "epoch": 0.5205066757959603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3345,
      "step": 15204
    },
    {
      "epoch": 0.5205409106470387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3485,
      "step": 15205
    },
    {
      "epoch": 0.5205751454981171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3468,
      "step": 15206
    },
    {
      "epoch": 0.5206093803491955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3583,
      "step": 15207
    },
    {
      "epoch": 0.5206436152002739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3625,
      "step": 15208
    },
    {
      "epoch": 0.5206778500513523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.358,
      "step": 15209
    },
    {
      "epoch": 0.5207120849024307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3565,
      "step": 15210
    },
    {
      "epoch": 0.520746319753509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3605,
      "step": 15211
    },
    {
      "epoch": 0.5207805546045875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3574,
      "step": 15212
    },
    {
      "epoch": 0.5208147894556658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3501,
      "step": 15213
    },
    {
      "epoch": 0.5208490243067443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3402,
      "step": 15214
    },
    {
      "epoch": 0.5208832591578226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3695,
      "step": 15215
    },
    {
      "epoch": 0.5209174940089011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3453,
      "step": 15216
    },
    {
      "epoch": 0.5209517288599794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3507,
      "step": 15217
    },
    {
      "epoch": 0.5209859637110579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3672,
      "step": 15218
    },
    {
      "epoch": 0.5210201985621362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.352,
      "step": 15219
    },
    {
      "epoch": 0.5210544334132147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3494,
      "step": 15220
    },
    {
      "epoch": 0.521088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3767,
      "step": 15221
    },
    {
      "epoch": 0.5211229031153715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3535,
      "step": 15222
    },
    {
      "epoch": 0.5211571379664498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3525,
      "step": 15223
    },
    {
      "epoch": 0.5211913728175283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3577,
      "step": 15224
    },
    {
      "epoch": 0.5212256076686066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3821,
      "step": 15225
    },
    {
      "epoch": 0.521259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3774,
      "step": 15226
    },
    {
      "epoch": 0.5212940773707634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3575,
      "step": 15227
    },
    {
      "epoch": 0.5213283122218418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3585,
      "step": 15228
    },
    {
      "epoch": 0.5213625470729202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3755,
      "step": 15229
    },
    {
      "epoch": 0.5213967819239986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3474,
      "step": 15230
    },
    {
      "epoch": 0.5214310167750771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3717,
      "step": 15231
    },
    {
      "epoch": 0.5214652516261554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3541,
      "step": 15232
    },
    {
      "epoch": 0.5214994864772339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3341,
      "step": 15233
    },
    {
      "epoch": 0.5215337213283122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3696,
      "step": 15234
    },
    {
      "epoch": 0.5215679561793907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3321,
      "step": 15235
    },
    {
      "epoch": 0.521602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3781,
      "step": 15236
    },
    {
      "epoch": 0.5216364258815475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.347,
      "step": 15237
    },
    {
      "epoch": 0.5216706607326258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3875,
      "step": 15238
    },
    {
      "epoch": 0.5217048955837043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.366,
      "step": 15239
    },
    {
      "epoch": 0.5217391304347826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.369,
      "step": 15240
    },
    {
      "epoch": 0.521773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3597,
      "step": 15241
    },
    {
      "epoch": 0.5218076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3508,
      "step": 15242
    },
    {
      "epoch": 0.5218418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3476,
      "step": 15243
    },
    {
      "epoch": 0.5218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.343,
      "step": 15244
    },
    {
      "epoch": 0.5219103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3744,
      "step": 15245
    },
    {
      "epoch": 0.521944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3602,
      "step": 15246
    },
    {
      "epoch": 0.5219787743923314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3505,
      "step": 15247
    },
    {
      "epoch": 0.5220130092434098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3789,
      "step": 15248
    },
    {
      "epoch": 0.5220472440944882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3598,
      "step": 15249
    },
    {
      "epoch": 0.5220814789455666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3555,
      "step": 15250
    },
    {
      "epoch": 0.522115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3638,
      "step": 15251
    },
    {
      "epoch": 0.5221499486477233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3718,
      "step": 15252
    },
    {
      "epoch": 0.5221841834988018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3692,
      "step": 15253
    },
    {
      "epoch": 0.5222184183498801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3783,
      "step": 15254
    },
    {
      "epoch": 0.5222526532009586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3299,
      "step": 15255
    },
    {
      "epoch": 0.5222868880520369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3513,
      "step": 15256
    },
    {
      "epoch": 0.5223211229031154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3657,
      "step": 15257
    },
    {
      "epoch": 0.5223553577541937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.345,
      "step": 15258
    },
    {
      "epoch": 0.5223895926052722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3369,
      "step": 15259
    },
    {
      "epoch": 0.5224238274563505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3688,
      "step": 15260
    },
    {
      "epoch": 0.522458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3789,
      "step": 15261
    },
    {
      "epoch": 0.5224922971585073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3729,
      "step": 15262
    },
    {
      "epoch": 0.5225265320095858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3653,
      "step": 15263
    },
    {
      "epoch": 0.5225607668606641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3375,
      "step": 15264
    },
    {
      "epoch": 0.5225950017117426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3356,
      "step": 15265
    },
    {
      "epoch": 0.5226292365628209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3434,
      "step": 15266
    },
    {
      "epoch": 0.5226634714138993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3667,
      "step": 15267
    },
    {
      "epoch": 0.5226977062649778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3721,
      "step": 15268
    },
    {
      "epoch": 0.5227319411160561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3674,
      "step": 15269
    },
    {
      "epoch": 0.5227661759671346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3514,
      "step": 15270
    },
    {
      "epoch": 0.5228004108182129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.369,
      "step": 15271
    },
    {
      "epoch": 0.5228346456692914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3635,
      "step": 15272
    },
    {
      "epoch": 0.5228688805203697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3383,
      "step": 15273
    },
    {
      "epoch": 0.5229031153714482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3616,
      "step": 15274
    },
    {
      "epoch": 0.5229373502225265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3803,
      "step": 15275
    },
    {
      "epoch": 0.522971585073605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3478,
      "step": 15276
    },
    {
      "epoch": 0.5230058199246833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3422,
      "step": 15277
    },
    {
      "epoch": 0.5230400547757618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3488,
      "step": 15278
    },
    {
      "epoch": 0.5230742896268401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3569,
      "step": 15279
    },
    {
      "epoch": 0.5231085244779186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3537,
      "step": 15280
    },
    {
      "epoch": 0.5231427593289969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3493,
      "step": 15281
    },
    {
      "epoch": 0.5231769941800753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3658,
      "step": 15282
    },
    {
      "epoch": 0.5232112290311537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3425,
      "step": 15283
    },
    {
      "epoch": 0.5232454638822321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3784,
      "step": 15284
    },
    {
      "epoch": 0.5232796987333105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3638,
      "step": 15285
    },
    {
      "epoch": 0.5233139335843889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3315,
      "step": 15286
    },
    {
      "epoch": 0.5233481684354673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3481,
      "step": 15287
    },
    {
      "epoch": 0.5233824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3463,
      "step": 15288
    },
    {
      "epoch": 0.5234166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3571,
      "step": 15289
    },
    {
      "epoch": 0.5234508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3665,
      "step": 15290
    },
    {
      "epoch": 0.5234851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3639,
      "step": 15291
    },
    {
      "epoch": 0.5235193426908593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3438,
      "step": 15292
    },
    {
      "epoch": 0.5235535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.362,
      "step": 15293
    },
    {
      "epoch": 0.5235878123930161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3574,
      "step": 15294
    },
    {
      "epoch": 0.5236220472440944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3668,
      "step": 15295
    },
    {
      "epoch": 0.5236562820951729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3532,
      "step": 15296
    },
    {
      "epoch": 0.5236905169462512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.37,
      "step": 15297
    },
    {
      "epoch": 0.5237247517973297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3389,
      "step": 15298
    },
    {
      "epoch": 0.523758986648408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3627,
      "step": 15299
    },
    {
      "epoch": 0.5237932214994865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3539,
      "step": 15300
    },
    {
      "epoch": 0.5238274563505648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3538,
      "step": 15301
    },
    {
      "epoch": 0.5238616912016433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3715,
      "step": 15302
    },
    {
      "epoch": 0.5238959260527216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3522,
      "step": 15303
    },
    {
      "epoch": 0.5239301609038001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3477,
      "step": 15304
    },
    {
      "epoch": 0.5239643957548785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3644,
      "step": 15305
    },
    {
      "epoch": 0.5239986306059569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3716,
      "step": 15306
    },
    {
      "epoch": 0.5240328654570353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3579,
      "step": 15307
    },
    {
      "epoch": 0.5240671003081137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.36,
      "step": 15308
    },
    {
      "epoch": 0.5241013351591921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3624,
      "step": 15309
    },
    {
      "epoch": 0.5241355700102704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3708,
      "step": 15310
    },
    {
      "epoch": 0.5241698048613489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.361,
      "step": 15311
    },
    {
      "epoch": 0.5242040397124272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3641,
      "step": 15312
    },
    {
      "epoch": 0.5242382745635057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3665,
      "step": 15313
    },
    {
      "epoch": 0.524272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3583,
      "step": 15314
    },
    {
      "epoch": 0.5243067442656625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3582,
      "step": 15315
    },
    {
      "epoch": 0.5243409791167408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3472,
      "step": 15316
    },
    {
      "epoch": 0.5243752139678193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.345,
      "step": 15317
    },
    {
      "epoch": 0.5244094488188976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3434,
      "step": 15318
    },
    {
      "epoch": 0.5244436836699761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3577,
      "step": 15319
    },
    {
      "epoch": 0.5244779185210544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3669,
      "step": 15320
    },
    {
      "epoch": 0.5245121533721329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3545,
      "step": 15321
    },
    {
      "epoch": 0.5245463882232112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3588,
      "step": 15322
    },
    {
      "epoch": 0.5245806230742897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3756,
      "step": 15323
    },
    {
      "epoch": 0.524614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3662,
      "step": 15324
    },
    {
      "epoch": 0.5246490927764464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3362,
      "step": 15325
    },
    {
      "epoch": 0.5246833276275248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3438,
      "step": 15326
    },
    {
      "epoch": 0.5247175624786032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3339,
      "step": 15327
    },
    {
      "epoch": 0.5247517973296816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3472,
      "step": 15328
    },
    {
      "epoch": 0.52478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3649,
      "step": 15329
    },
    {
      "epoch": 0.5248202670318384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3502,
      "step": 15330
    },
    {
      "epoch": 0.5248545018829168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3571,
      "step": 15331
    },
    {
      "epoch": 0.5248887367339952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3603,
      "step": 15332
    },
    {
      "epoch": 0.5249229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3525,
      "step": 15333
    },
    {
      "epoch": 0.524957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3541,
      "step": 15334
    },
    {
      "epoch": 0.5249914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3591,
      "step": 15335
    },
    {
      "epoch": 0.5250256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.335,
      "step": 15336
    },
    {
      "epoch": 0.5250599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3665,
      "step": 15337
    },
    {
      "epoch": 0.5250941458404655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3516,
      "step": 15338
    },
    {
      "epoch": 0.525128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3716,
      "step": 15339
    },
    {
      "epoch": 0.5251626155426223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3588,
      "step": 15340
    },
    {
      "epoch": 0.5251968503937008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3259,
      "step": 15341
    },
    {
      "epoch": 0.5252310852447792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3556,
      "step": 15342
    },
    {
      "epoch": 0.5252653200958576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.352,
      "step": 15343
    },
    {
      "epoch": 0.525299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3556,
      "step": 15344
    },
    {
      "epoch": 0.5253337897980144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.349,
      "step": 15345
    },
    {
      "epoch": 0.5253680246490928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3405,
      "step": 15346
    },
    {
      "epoch": 0.5254022595001712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3547,
      "step": 15347
    },
    {
      "epoch": 0.5254364943512496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3436,
      "step": 15348
    },
    {
      "epoch": 0.525470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3511,
      "step": 15349
    },
    {
      "epoch": 0.5255049640534064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3539,
      "step": 15350
    },
    {
      "epoch": 0.5255391989044848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3438,
      "step": 15351
    },
    {
      "epoch": 0.5255734337555632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3434,
      "step": 15352
    },
    {
      "epoch": 0.5256076686066415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.348,
      "step": 15353
    },
    {
      "epoch": 0.52564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3438,
      "step": 15354
    },
    {
      "epoch": 0.5256761383087983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3698,
      "step": 15355
    },
    {
      "epoch": 0.5257103731598768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3488,
      "step": 15356
    },
    {
      "epoch": 0.5257446080109551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3637,
      "step": 15357
    },
    {
      "epoch": 0.5257788428620336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3557,
      "step": 15358
    },
    {
      "epoch": 0.5258130777131119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3564,
      "step": 15359
    },
    {
      "epoch": 0.5258473125641904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3508,
      "step": 15360
    },
    {
      "epoch": 0.5258815474152687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3375,
      "step": 15361
    },
    {
      "epoch": 0.5259157822663472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3287,
      "step": 15362
    },
    {
      "epoch": 0.5259500171174255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3663,
      "step": 15363
    },
    {
      "epoch": 0.525984251968504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3397,
      "step": 15364
    },
    {
      "epoch": 0.5260184868195823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3628,
      "step": 15365
    },
    {
      "epoch": 0.5260527216706608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3687,
      "step": 15366
    },
    {
      "epoch": 0.5260869565217391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3449,
      "step": 15367
    },
    {
      "epoch": 0.5261211913728175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3388,
      "step": 15368
    },
    {
      "epoch": 0.5261554262238959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3555,
      "step": 15369
    },
    {
      "epoch": 0.5261896610749743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3646,
      "step": 15370
    },
    {
      "epoch": 0.5262238959260527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3569,
      "step": 15371
    },
    {
      "epoch": 0.5262581307771311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3663,
      "step": 15372
    },
    {
      "epoch": 0.5262923656282095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3445,
      "step": 15373
    },
    {
      "epoch": 0.5263266004792879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3361,
      "step": 15374
    },
    {
      "epoch": 0.5263608353303663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3712,
      "step": 15375
    },
    {
      "epoch": 0.5263950701814447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3685,
      "step": 15376
    },
    {
      "epoch": 0.526429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3503,
      "step": 15377
    },
    {
      "epoch": 0.5264635398836015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3511,
      "step": 15378
    },
    {
      "epoch": 0.52649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3557,
      "step": 15379
    },
    {
      "epoch": 0.5265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.358,
      "step": 15380
    },
    {
      "epoch": 0.5265662444368368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3421,
      "step": 15381
    },
    {
      "epoch": 0.5266004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3402,
      "step": 15382
    },
    {
      "epoch": 0.5266347141389935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3567,
      "step": 15383
    },
    {
      "epoch": 0.5266689489900719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3654,
      "step": 15384
    },
    {
      "epoch": 0.5267031838411503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3583,
      "step": 15385
    },
    {
      "epoch": 0.5267374186922287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3567,
      "step": 15386
    },
    {
      "epoch": 0.5267716535433071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3483,
      "step": 15387
    },
    {
      "epoch": 0.5268058883943855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3511,
      "step": 15388
    },
    {
      "epoch": 0.5268401232454639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3425,
      "step": 15389
    },
    {
      "epoch": 0.5268743580965423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3231,
      "step": 15390
    },
    {
      "epoch": 0.5269085929476207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3474,
      "step": 15391
    },
    {
      "epoch": 0.526942827798699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3544,
      "step": 15392
    },
    {
      "epoch": 0.5269770626497775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3502,
      "step": 15393
    },
    {
      "epoch": 0.5270112975008558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3549,
      "step": 15394
    },
    {
      "epoch": 0.5270455323519343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3464,
      "step": 15395
    },
    {
      "epoch": 0.5270797672030126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3452,
      "step": 15396
    },
    {
      "epoch": 0.5271140020540911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3452,
      "step": 15397
    },
    {
      "epoch": 0.5271482369051694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3549,
      "step": 15398
    },
    {
      "epoch": 0.5271824717562479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3511,
      "step": 15399
    },
    {
      "epoch": 0.5272167066073262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3549,
      "step": 15400
    },
    {
      "epoch": 0.5272509414584047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3687,
      "step": 15401
    },
    {
      "epoch": 0.527285176309483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3336,
      "step": 15402
    },
    {
      "epoch": 0.5273194111605615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3386,
      "step": 15403
    },
    {
      "epoch": 0.5273536460116398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3529,
      "step": 15404
    },
    {
      "epoch": 0.5273878808627183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3485,
      "step": 15405
    },
    {
      "epoch": 0.5274221157137966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3414,
      "step": 15406
    },
    {
      "epoch": 0.527456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3561,
      "step": 15407
    },
    {
      "epoch": 0.5274905854159534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3509,
      "step": 15408
    },
    {
      "epoch": 0.5275248202670318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3626,
      "step": 15409
    },
    {
      "epoch": 0.5275590551181102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3485,
      "step": 15410
    },
    {
      "epoch": 0.5275932899691886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3331,
      "step": 15411
    },
    {
      "epoch": 0.527627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3589,
      "step": 15412
    },
    {
      "epoch": 0.5276617596713454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3611,
      "step": 15413
    },
    {
      "epoch": 0.5276959945224238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.351,
      "step": 15414
    },
    {
      "epoch": 0.5277302293735022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3527,
      "step": 15415
    },
    {
      "epoch": 0.5277644642245807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3618,
      "step": 15416
    },
    {
      "epoch": 0.527798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.344,
      "step": 15417
    },
    {
      "epoch": 0.5278329339267375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3428,
      "step": 15418
    },
    {
      "epoch": 0.5278671687778158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3654,
      "step": 15419
    },
    {
      "epoch": 0.5279014036288943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3695,
      "step": 15420
    },
    {
      "epoch": 0.5279356384799726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3501,
      "step": 15421
    },
    {
      "epoch": 0.527969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3368,
      "step": 15422
    },
    {
      "epoch": 0.5280041081821294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3495,
      "step": 15423
    },
    {
      "epoch": 0.5280383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3583,
      "step": 15424
    },
    {
      "epoch": 0.5280725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3557,
      "step": 15425
    },
    {
      "epoch": 0.5281068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3642,
      "step": 15426
    },
    {
      "epoch": 0.528141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3469,
      "step": 15427
    },
    {
      "epoch": 0.5281752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3379,
      "step": 15428
    },
    {
      "epoch": 0.5282095172885998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3535,
      "step": 15429
    },
    {
      "epoch": 0.5282437521396782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3453,
      "step": 15430
    },
    {
      "epoch": 0.5282779869907566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3532,
      "step": 15431
    },
    {
      "epoch": 0.528312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3404,
      "step": 15432
    },
    {
      "epoch": 0.5283464566929134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3602,
      "step": 15433
    },
    {
      "epoch": 0.5283806915439918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3547,
      "step": 15434
    },
    {
      "epoch": 0.5284149263950702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3372,
      "step": 15435
    },
    {
      "epoch": 0.5284491612461486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3532,
      "step": 15436
    },
    {
      "epoch": 0.5284833960972269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3398,
      "step": 15437
    },
    {
      "epoch": 0.5285176309483054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3461,
      "step": 15438
    },
    {
      "epoch": 0.5285518657993837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3472,
      "step": 15439
    },
    {
      "epoch": 0.5285861006504622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3594,
      "step": 15440
    },
    {
      "epoch": 0.5286203355015405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3716,
      "step": 15441
    },
    {
      "epoch": 0.528654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3502,
      "step": 15442
    },
    {
      "epoch": 0.5286888052036973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3496,
      "step": 15443
    },
    {
      "epoch": 0.5287230400547758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3596,
      "step": 15444
    },
    {
      "epoch": 0.5287572749058541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3469,
      "step": 15445
    },
    {
      "epoch": 0.5287915097569326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3532,
      "step": 15446
    },
    {
      "epoch": 0.5288257446080109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3569,
      "step": 15447
    },
    {
      "epoch": 0.5288599794590894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3519,
      "step": 15448
    },
    {
      "epoch": 0.5288942143101677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3253,
      "step": 15449
    },
    {
      "epoch": 0.5289284491612462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3505,
      "step": 15450
    },
    {
      "epoch": 0.5289626840123245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3491,
      "step": 15451
    },
    {
      "epoch": 0.5289969188634029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.348,
      "step": 15452
    },
    {
      "epoch": 0.5290311537144814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3601,
      "step": 15453
    },
    {
      "epoch": 0.5290653885655597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3638,
      "step": 15454
    },
    {
      "epoch": 0.5290996234166382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3376,
      "step": 15455
    },
    {
      "epoch": 0.5291338582677165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3444,
      "step": 15456
    },
    {
      "epoch": 0.529168093118795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3547,
      "step": 15457
    },
    {
      "epoch": 0.5292023279698733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3519,
      "step": 15458
    },
    {
      "epoch": 0.5292365628209518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3466,
      "step": 15459
    },
    {
      "epoch": 0.5292707976720301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.344,
      "step": 15460
    },
    {
      "epoch": 0.5293050325231086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3429,
      "step": 15461
    },
    {
      "epoch": 0.5293392673741869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.335,
      "step": 15462
    },
    {
      "epoch": 0.5293735022252654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3503,
      "step": 15463
    },
    {
      "epoch": 0.5294077370763437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3484,
      "step": 15464
    },
    {
      "epoch": 0.5294419719274222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3546,
      "step": 15465
    },
    {
      "epoch": 0.5294762067785005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3298,
      "step": 15466
    },
    {
      "epoch": 0.5295104416295789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3431,
      "step": 15467
    },
    {
      "epoch": 0.5295446764806573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3521,
      "step": 15468
    },
    {
      "epoch": 0.5295789113317357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3526,
      "step": 15469
    },
    {
      "epoch": 0.5296131461828141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3574,
      "step": 15470
    },
    {
      "epoch": 0.5296473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3603,
      "step": 15471
    },
    {
      "epoch": 0.5296816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.334,
      "step": 15472
    },
    {
      "epoch": 0.5297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3394,
      "step": 15473
    },
    {
      "epoch": 0.5297500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3498,
      "step": 15474
    },
    {
      "epoch": 0.5297843204382061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.34,
      "step": 15475
    },
    {
      "epoch": 0.5298185552892845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3351,
      "step": 15476
    },
    {
      "epoch": 0.5298527901403629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3363,
      "step": 15477
    },
    {
      "epoch": 0.5298870249914412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3352,
      "step": 15478
    },
    {
      "epoch": 0.5299212598425197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3361,
      "step": 15479
    },
    {
      "epoch": 0.529955494693598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3344,
      "step": 15480
    },
    {
      "epoch": 0.5299897295446765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3182,
      "step": 15481
    },
    {
      "epoch": 0.5300239643957548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3429,
      "step": 15482
    },
    {
      "epoch": 0.5300581992468333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3454,
      "step": 15483
    },
    {
      "epoch": 0.5300924340979116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3464,
      "step": 15484
    },
    {
      "epoch": 0.5301266689489901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3494,
      "step": 15485
    },
    {
      "epoch": 0.5301609038000684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.358,
      "step": 15486
    },
    {
      "epoch": 0.5301951386511469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3558,
      "step": 15487
    },
    {
      "epoch": 0.5302293735022253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3615,
      "step": 15488
    },
    {
      "epoch": 0.5302636083533037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3273,
      "step": 15489
    },
    {
      "epoch": 0.5302978432043821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3781,
      "step": 15490
    },
    {
      "epoch": 0.5303320780554605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3403,
      "step": 15491
    },
    {
      "epoch": 0.5303663129065389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3278,
      "step": 15492
    },
    {
      "epoch": 0.5304005477576172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3549,
      "step": 15493
    },
    {
      "epoch": 0.5304347826086957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3403,
      "step": 15494
    },
    {
      "epoch": 0.530469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3472,
      "step": 15495
    },
    {
      "epoch": 0.5305032523108525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3546,
      "step": 15496
    },
    {
      "epoch": 0.5305374871619308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3486,
      "step": 15497
    },
    {
      "epoch": 0.5305717220130093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3422,
      "step": 15498
    },
    {
      "epoch": 0.5306059568640876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3328,
      "step": 15499
    },
    {
      "epoch": 0.5306401917151661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3428,
      "step": 15500
    },
    {
      "epoch": 0.5306744265662444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3582,
      "step": 15501
    },
    {
      "epoch": 0.5307086614173229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3202,
      "step": 15502
    },
    {
      "epoch": 0.5307428962684012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3446,
      "step": 15503
    },
    {
      "epoch": 0.5307771311194797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3642,
      "step": 15504
    },
    {
      "epoch": 0.530811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3599,
      "step": 15505
    },
    {
      "epoch": 0.5308456008216365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3564,
      "step": 15506
    },
    {
      "epoch": 0.5308798356727148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3713,
      "step": 15507
    },
    {
      "epoch": 0.5309140705237932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3479,
      "step": 15508
    },
    {
      "epoch": 0.5309483053748716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3468,
      "step": 15509
    },
    {
      "epoch": 0.53098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3517,
      "step": 15510
    },
    {
      "epoch": 0.5310167750770284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3498,
      "step": 15511
    },
    {
      "epoch": 0.5310510099281068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.342,
      "step": 15512
    },
    {
      "epoch": 0.5310852447791852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3356,
      "step": 15513
    },
    {
      "epoch": 0.5311194796302636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3429,
      "step": 15514
    },
    {
      "epoch": 0.531153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3416,
      "step": 15515
    },
    {
      "epoch": 0.5311879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3387,
      "step": 15516
    },
    {
      "epoch": 0.5312221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3513,
      "step": 15517
    },
    {
      "epoch": 0.5312564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3279,
      "step": 15518
    },
    {
      "epoch": 0.5312906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3491,
      "step": 15519
    },
    {
      "epoch": 0.531324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3419,
      "step": 15520
    },
    {
      "epoch": 0.5313591235878123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3517,
      "step": 15521
    },
    {
      "epoch": 0.5313933584388908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.347,
      "step": 15522
    },
    {
      "epoch": 0.5314275932899691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3474,
      "step": 15523
    },
    {
      "epoch": 0.5314618281410476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3524,
      "step": 15524
    },
    {
      "epoch": 0.531496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3509,
      "step": 15525
    },
    {
      "epoch": 0.5315302978432044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3441,
      "step": 15526
    },
    {
      "epoch": 0.5315645326942828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3572,
      "step": 15527
    },
    {
      "epoch": 0.5315987675453612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3447,
      "step": 15528
    },
    {
      "epoch": 0.5316330023964396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3566,
      "step": 15529
    },
    {
      "epoch": 0.531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3391,
      "step": 15530
    },
    {
      "epoch": 0.5317014720985964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3303,
      "step": 15531
    },
    {
      "epoch": 0.5317357069496748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3521,
      "step": 15532
    },
    {
      "epoch": 0.5317699418007532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3582,
      "step": 15533
    },
    {
      "epoch": 0.5318041766518316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3581,
      "step": 15534
    },
    {
      "epoch": 0.53183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3406,
      "step": 15535
    },
    {
      "epoch": 0.5318726463539883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3627,
      "step": 15536
    },
    {
      "epoch": 0.5319068812050668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3384,
      "step": 15537
    },
    {
      "epoch": 0.5319411160561451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3443,
      "step": 15538
    },
    {
      "epoch": 0.5319753509072236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3552,
      "step": 15539
    },
    {
      "epoch": 0.5320095857583019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3453,
      "step": 15540
    },
    {
      "epoch": 0.5320438206093804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3334,
      "step": 15541
    },
    {
      "epoch": 0.5320780554604587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3353,
      "step": 15542
    },
    {
      "epoch": 0.5321122903115372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3641,
      "step": 15543
    },
    {
      "epoch": 0.5321465251626155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3557,
      "step": 15544
    },
    {
      "epoch": 0.532180760013694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3622,
      "step": 15545
    },
    {
      "epoch": 0.5322149948647723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3493,
      "step": 15546
    },
    {
      "epoch": 0.5322492297158508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3298,
      "step": 15547
    },
    {
      "epoch": 0.5322834645669291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3333,
      "step": 15548
    },
    {
      "epoch": 0.5323176994180076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3376,
      "step": 15549
    },
    {
      "epoch": 0.5323519342690859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3466,
      "step": 15550
    },
    {
      "epoch": 0.5323861691201643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3444,
      "step": 15551
    },
    {
      "epoch": 0.5324204039712427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3596,
      "step": 15552
    },
    {
      "epoch": 0.5324546388223211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3299,
      "step": 15553
    },
    {
      "epoch": 0.5324888736733995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3427,
      "step": 15554
    },
    {
      "epoch": 0.5325231085244779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3522,
      "step": 15555
    },
    {
      "epoch": 0.5325573433755563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3466,
      "step": 15556
    },
    {
      "epoch": 0.5325915782266347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3386,
      "step": 15557
    },
    {
      "epoch": 0.5326258130777131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3437,
      "step": 15558
    },
    {
      "epoch": 0.5326600479287915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3332,
      "step": 15559
    },
    {
      "epoch": 0.5326942827798699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3499,
      "step": 15560
    },
    {
      "epoch": 0.5327285176309483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.365,
      "step": 15561
    },
    {
      "epoch": 0.5327627524820268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3599,
      "step": 15562
    },
    {
      "epoch": 0.5327969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3496,
      "step": 15563
    },
    {
      "epoch": 0.5328312221841836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3659,
      "step": 15564
    },
    {
      "epoch": 0.5328654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3603,
      "step": 15565
    },
    {
      "epoch": 0.5328996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3416,
      "step": 15566
    },
    {
      "epoch": 0.5329339267374187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3472,
      "step": 15567
    },
    {
      "epoch": 0.5329681615884971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3362,
      "step": 15568
    },
    {
      "epoch": 0.5330023964395755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3442,
      "step": 15569
    },
    {
      "epoch": 0.5330366312906539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3565,
      "step": 15570
    },
    {
      "epoch": 0.5330708661417323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3687,
      "step": 15571
    },
    {
      "epoch": 0.5331051009928107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3624,
      "step": 15572
    },
    {
      "epoch": 0.5331393358438891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3462,
      "step": 15573
    },
    {
      "epoch": 0.5331735706949675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3487,
      "step": 15574
    },
    {
      "epoch": 0.5332078055460459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3485,
      "step": 15575
    },
    {
      "epoch": 0.5332420403971243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3439,
      "step": 15576
    },
    {
      "epoch": 0.5332762752482026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3466,
      "step": 15577
    },
    {
      "epoch": 0.5333105100992811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3511,
      "step": 15578
    },
    {
      "epoch": 0.5333447449503594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3484,
      "step": 15579
    },
    {
      "epoch": 0.5333789798014379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3385,
      "step": 15580
    },
    {
      "epoch": 0.5334132146525162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3318,
      "step": 15581
    },
    {
      "epoch": 0.5334474495035947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.333,
      "step": 15582
    },
    {
      "epoch": 0.533481684354673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3405,
      "step": 15583
    },
    {
      "epoch": 0.5335159192057515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3682,
      "step": 15584
    },
    {
      "epoch": 0.5335501540568298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3459,
      "step": 15585
    },
    {
      "epoch": 0.5335843889079083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3529,
      "step": 15586
    },
    {
      "epoch": 0.5336186237589866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3369,
      "step": 15587
    },
    {
      "epoch": 0.5336528586100651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3505,
      "step": 15588
    },
    {
      "epoch": 0.5336870934611434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3461,
      "step": 15589
    },
    {
      "epoch": 0.5337213283122219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3521,
      "step": 15590
    },
    {
      "epoch": 0.5337555631633002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3483,
      "step": 15591
    },
    {
      "epoch": 0.5337897980143786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3433,
      "step": 15592
    },
    {
      "epoch": 0.533824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3539,
      "step": 15593
    },
    {
      "epoch": 0.5338582677165354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3457,
      "step": 15594
    },
    {
      "epoch": 0.5338925025676138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3573,
      "step": 15595
    },
    {
      "epoch": 0.5339267374186922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3588,
      "step": 15596
    },
    {
      "epoch": 0.5339609722697706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3578,
      "step": 15597
    },
    {
      "epoch": 0.533995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3522,
      "step": 15598
    },
    {
      "epoch": 0.5340294419719275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3508,
      "step": 15599
    },
    {
      "epoch": 0.5340636768230058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3364,
      "step": 15600
    },
    {
      "epoch": 0.5340979116740843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3588,
      "step": 15601
    },
    {
      "epoch": 0.5341321465251626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3475,
      "step": 15602
    },
    {
      "epoch": 0.5341663813762411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3481,
      "step": 15603
    },
    {
      "epoch": 0.5342006162273194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3329,
      "step": 15604
    },
    {
      "epoch": 0.5342348510783979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3392,
      "step": 15605
    },
    {
      "epoch": 0.5342690859294762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3359,
      "step": 15606
    },
    {
      "epoch": 0.5343033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3677,
      "step": 15607
    },
    {
      "epoch": 0.534337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.345,
      "step": 15608
    },
    {
      "epoch": 0.5343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3497,
      "step": 15609
    },
    {
      "epoch": 0.5344060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3432,
      "step": 15610
    },
    {
      "epoch": 0.5344402601848682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3393,
      "step": 15611
    },
    {
      "epoch": 0.5344744950359466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3506,
      "step": 15612
    },
    {
      "epoch": 0.534508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3638,
      "step": 15613
    },
    {
      "epoch": 0.5345429647381034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3477,
      "step": 15614
    },
    {
      "epoch": 0.5345771995891818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3259,
      "step": 15615
    },
    {
      "epoch": 0.5346114344402602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3467,
      "step": 15616
    },
    {
      "epoch": 0.5346456692913386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3389,
      "step": 15617
    },
    {
      "epoch": 0.534679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3232,
      "step": 15618
    },
    {
      "epoch": 0.5347141389934954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3396,
      "step": 15619
    },
    {
      "epoch": 0.5347483738445737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3425,
      "step": 15620
    },
    {
      "epoch": 0.5347826086956522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3487,
      "step": 15621
    },
    {
      "epoch": 0.5348168435467305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3392,
      "step": 15622
    },
    {
      "epoch": 0.534851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3423,
      "step": 15623
    },
    {
      "epoch": 0.5348853132488873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3487,
      "step": 15624
    },
    {
      "epoch": 0.5349195480999658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.364,
      "step": 15625
    },
    {
      "epoch": 0.5349537829510441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3346,
      "step": 15626
    },
    {
      "epoch": 0.5349880178021226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3403,
      "step": 15627
    },
    {
      "epoch": 0.5350222526532009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3513,
      "step": 15628
    },
    {
      "epoch": 0.5350564875042794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3446,
      "step": 15629
    },
    {
      "epoch": 0.5350907223553577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3456,
      "step": 15630
    },
    {
      "epoch": 0.5351249572064362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3604,
      "step": 15631
    },
    {
      "epoch": 0.5351591920575145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3405,
      "step": 15632
    },
    {
      "epoch": 0.535193426908593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3482,
      "step": 15633
    },
    {
      "epoch": 0.5352276617596713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3401,
      "step": 15634
    },
    {
      "epoch": 0.5352618966107497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3508,
      "step": 15635
    },
    {
      "epoch": 0.5352961314618282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3425,
      "step": 15636
    },
    {
      "epoch": 0.5353303663129065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3492,
      "step": 15637
    },
    {
      "epoch": 0.535364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.371,
      "step": 15638
    },
    {
      "epoch": 0.5353988360150633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3624,
      "step": 15639
    },
    {
      "epoch": 0.5354330708661418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.347,
      "step": 15640
    },
    {
      "epoch": 0.5354673057172201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3594,
      "step": 15641
    },
    {
      "epoch": 0.5355015405682986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3488,
      "step": 15642
    },
    {
      "epoch": 0.5355357754193769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3251,
      "step": 15643
    },
    {
      "epoch": 0.5355700102704554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3593,
      "step": 15644
    },
    {
      "epoch": 0.5356042451215337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3345,
      "step": 15645
    },
    {
      "epoch": 0.5356384799726122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3406,
      "step": 15646
    },
    {
      "epoch": 0.5356727148236905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3329,
      "step": 15647
    },
    {
      "epoch": 0.535706949674769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3492,
      "step": 15648
    },
    {
      "epoch": 0.5357411845258473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3522,
      "step": 15649
    },
    {
      "epoch": 0.5357754193769257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3384,
      "step": 15650
    },
    {
      "epoch": 0.5358096542280041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3531,
      "step": 15651
    },
    {
      "epoch": 0.5358438890790825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3396,
      "step": 15652
    },
    {
      "epoch": 0.5358781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3411,
      "step": 15653
    },
    {
      "epoch": 0.5359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3433,
      "step": 15654
    },
    {
      "epoch": 0.5359465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3533,
      "step": 15655
    },
    {
      "epoch": 0.5359808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3504,
      "step": 15656
    },
    {
      "epoch": 0.5360150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3359,
      "step": 15657
    },
    {
      "epoch": 0.5360492981855529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3438,
      "step": 15658
    },
    {
      "epoch": 0.5360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3597,
      "step": 15659
    },
    {
      "epoch": 0.5361177678877097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3486,
      "step": 15660
    },
    {
      "epoch": 0.536152002738788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3469,
      "step": 15661
    },
    {
      "epoch": 0.5361862375898665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.348,
      "step": 15662
    },
    {
      "epoch": 0.5362204724409448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3363,
      "step": 15663
    },
    {
      "epoch": 0.5362547072920233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3517,
      "step": 15664
    },
    {
      "epoch": 0.5362889421431016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.341,
      "step": 15665
    },
    {
      "epoch": 0.5363231769941801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.343,
      "step": 15666
    },
    {
      "epoch": 0.5363574118452584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3395,
      "step": 15667
    },
    {
      "epoch": 0.5363916466963369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3516,
      "step": 15668
    },
    {
      "epoch": 0.5364258815474152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3494,
      "step": 15669
    },
    {
      "epoch": 0.5364601163984937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.333,
      "step": 15670
    },
    {
      "epoch": 0.536494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.35,
      "step": 15671
    },
    {
      "epoch": 0.5365285861006505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3381,
      "step": 15672
    },
    {
      "epoch": 0.5365628209517289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3524,
      "step": 15673
    },
    {
      "epoch": 0.5365970558028073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3511,
      "step": 15674
    },
    {
      "epoch": 0.5366312906538857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3572,
      "step": 15675
    },
    {
      "epoch": 0.536665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3469,
      "step": 15676
    },
    {
      "epoch": 0.5366997603560425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3463,
      "step": 15677
    },
    {
      "epoch": 0.5367339952071208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3297,
      "step": 15678
    },
    {
      "epoch": 0.5367682300581993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3458,
      "step": 15679
    },
    {
      "epoch": 0.5368024649092776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.331,
      "step": 15680
    },
    {
      "epoch": 0.5368366997603561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3345,
      "step": 15681
    },
    {
      "epoch": 0.5368709346114344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3433,
      "step": 15682
    },
    {
      "epoch": 0.5369051694625129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3483,
      "step": 15683
    },
    {
      "epoch": 0.5369394043135912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3455,
      "step": 15684
    },
    {
      "epoch": 0.5369736391646697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3482,
      "step": 15685
    },
    {
      "epoch": 0.537007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3561,
      "step": 15686
    },
    {
      "epoch": 0.5370421088668265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3373,
      "step": 15687
    },
    {
      "epoch": 0.5370763437179048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3413,
      "step": 15688
    },
    {
      "epoch": 0.5371105785689833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3259,
      "step": 15689
    },
    {
      "epoch": 0.5371448134200616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3262,
      "step": 15690
    },
    {
      "epoch": 0.53717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3304,
      "step": 15691
    },
    {
      "epoch": 0.5372132831222184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3524,
      "step": 15692
    },
    {
      "epoch": 0.5372475179732968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3628,
      "step": 15693
    },
    {
      "epoch": 0.5372817528243752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3433,
      "step": 15694
    },
    {
      "epoch": 0.5373159876754536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3499,
      "step": 15695
    },
    {
      "epoch": 0.537350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3334,
      "step": 15696
    },
    {
      "epoch": 0.5373844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3434,
      "step": 15697
    },
    {
      "epoch": 0.5374186922286888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3434,
      "step": 15698
    },
    {
      "epoch": 0.5374529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3485,
      "step": 15699
    },
    {
      "epoch": 0.5374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3389,
      "step": 15700
    },
    {
      "epoch": 0.537521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3248,
      "step": 15701
    },
    {
      "epoch": 0.5375556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.343,
      "step": 15702
    },
    {
      "epoch": 0.5375898664840808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3347,
      "step": 15703
    },
    {
      "epoch": 0.5376241013351591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3398,
      "step": 15704
    },
    {
      "epoch": 0.5376583361862376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3348,
      "step": 15705
    },
    {
      "epoch": 0.5376925710373159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3554,
      "step": 15706
    },
    {
      "epoch": 0.5377268058883944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3473,
      "step": 15707
    },
    {
      "epoch": 0.5377610407394727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3442,
      "step": 15708
    },
    {
      "epoch": 0.5377952755905512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3368,
      "step": 15709
    },
    {
      "epoch": 0.5378295104416296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3434,
      "step": 15710
    },
    {
      "epoch": 0.537863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3369,
      "step": 15711
    },
    {
      "epoch": 0.5378979801437864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3416,
      "step": 15712
    },
    {
      "epoch": 0.5379322149948648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3329,
      "step": 15713
    },
    {
      "epoch": 0.5379664498459432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3356,
      "step": 15714
    },
    {
      "epoch": 0.5380006846970216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3485,
      "step": 15715
    },
    {
      "epoch": 0.5380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3539,
      "step": 15716
    },
    {
      "epoch": 0.5380691543991784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3375,
      "step": 15717
    },
    {
      "epoch": 0.5381033892502568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3687,
      "step": 15718
    },
    {
      "epoch": 0.5381376241013351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3456,
      "step": 15719
    },
    {
      "epoch": 0.5381718589524136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3511,
      "step": 15720
    },
    {
      "epoch": 0.5382060938034919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.334,
      "step": 15721
    },
    {
      "epoch": 0.5382403286545704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3485,
      "step": 15722
    },
    {
      "epoch": 0.5382745635056487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3456,
      "step": 15723
    },
    {
      "epoch": 0.5383087983567272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.338,
      "step": 15724
    },
    {
      "epoch": 0.5383430332078055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3289,
      "step": 15725
    },
    {
      "epoch": 0.538377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3532,
      "step": 15726
    },
    {
      "epoch": 0.5384115029099623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3524,
      "step": 15727
    },
    {
      "epoch": 0.5384457377610408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3402,
      "step": 15728
    },
    {
      "epoch": 0.5384799726121191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3478,
      "step": 15729
    },
    {
      "epoch": 0.5385142074631976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3318,
      "step": 15730
    },
    {
      "epoch": 0.5385484423142759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3375,
      "step": 15731
    },
    {
      "epoch": 0.5385826771653544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3649,
      "step": 15732
    },
    {
      "epoch": 0.5386169120164327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3336,
      "step": 15733
    },
    {
      "epoch": 0.5386511468675111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3712,
      "step": 15734
    },
    {
      "epoch": 0.5386853817185895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.371,
      "step": 15735
    },
    {
      "epoch": 0.5387196165696679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3536,
      "step": 15736
    },
    {
      "epoch": 0.5387538514207463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.329,
      "step": 15737
    },
    {
      "epoch": 0.5387880862718247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3373,
      "step": 15738
    },
    {
      "epoch": 0.5388223211229031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3306,
      "step": 15739
    },
    {
      "epoch": 0.5388565559739815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3338,
      "step": 15740
    },
    {
      "epoch": 0.5388907908250599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3478,
      "step": 15741
    },
    {
      "epoch": 0.5389250256761383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3456,
      "step": 15742
    },
    {
      "epoch": 0.5389592605272167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3229,
      "step": 15743
    },
    {
      "epoch": 0.5389934953782951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3375,
      "step": 15744
    },
    {
      "epoch": 0.5390277302293734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3436,
      "step": 15745
    },
    {
      "epoch": 0.5390619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3558,
      "step": 15746
    },
    {
      "epoch": 0.5390961999315304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3521,
      "step": 15747
    },
    {
      "epoch": 0.5391304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3605,
      "step": 15748
    },
    {
      "epoch": 0.5391646696336871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3535,
      "step": 15749
    },
    {
      "epoch": 0.5391989044847655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3267,
      "step": 15750
    },
    {
      "epoch": 0.5392331393358439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3463,
      "step": 15751
    },
    {
      "epoch": 0.5392673741869223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3409,
      "step": 15752
    },
    {
      "epoch": 0.5393016090380007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.342,
      "step": 15753
    },
    {
      "epoch": 0.5393358438890791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3425,
      "step": 15754
    },
    {
      "epoch": 0.5393700787401575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3415,
      "step": 15755
    },
    {
      "epoch": 0.5394043135912359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.345,
      "step": 15756
    },
    {
      "epoch": 0.5394385484423143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3619,
      "step": 15757
    },
    {
      "epoch": 0.5394727832933927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3399,
      "step": 15758
    },
    {
      "epoch": 0.5395070181444711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3452,
      "step": 15759
    },
    {
      "epoch": 0.5395412529955494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3499,
      "step": 15760
    },
    {
      "epoch": 0.5395754878466279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3311,
      "step": 15761
    },
    {
      "epoch": 0.5396097226977062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3463,
      "step": 15762
    },
    {
      "epoch": 0.5396439575487847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3261,
      "step": 15763
    },
    {
      "epoch": 0.539678192399863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3427,
      "step": 15764
    },
    {
      "epoch": 0.5397124272509415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3409,
      "step": 15765
    },
    {
      "epoch": 0.5397466621020198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3326,
      "step": 15766
    },
    {
      "epoch": 0.5397808969530983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3304,
      "step": 15767
    },
    {
      "epoch": 0.5398151318041766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3269,
      "step": 15768
    },
    {
      "epoch": 0.5398493666552551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3325,
      "step": 15769
    },
    {
      "epoch": 0.5398836015063334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3385,
      "step": 15770
    },
    {
      "epoch": 0.5399178363574119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3121,
      "step": 15771
    },
    {
      "epoch": 0.5399520712084902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3454,
      "step": 15772
    },
    {
      "epoch": 0.5399863060595687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3468,
      "step": 15773
    },
    {
      "epoch": 0.540020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3358,
      "step": 15774
    },
    {
      "epoch": 0.5400547757617254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3267,
      "step": 15775
    },
    {
      "epoch": 0.5400890106128038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3277,
      "step": 15776
    },
    {
      "epoch": 0.5401232454638822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3314,
      "step": 15777
    },
    {
      "epoch": 0.5401574803149606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3385,
      "step": 15778
    },
    {
      "epoch": 0.540191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3533,
      "step": 15779
    },
    {
      "epoch": 0.5402259500171174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3314,
      "step": 15780
    },
    {
      "epoch": 0.5402601848681958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3415,
      "step": 15781
    },
    {
      "epoch": 0.5402944197192742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3406,
      "step": 15782
    },
    {
      "epoch": 0.5403286545703526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3492,
      "step": 15783
    },
    {
      "epoch": 0.5403628894214311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3388,
      "step": 15784
    },
    {
      "epoch": 0.5403971242725094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3361,
      "step": 15785
    },
    {
      "epoch": 0.5404313591235879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3646,
      "step": 15786
    },
    {
      "epoch": 0.5404655939746662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3306,
      "step": 15787
    },
    {
      "epoch": 0.5404998288257447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3548,
      "step": 15788
    },
    {
      "epoch": 0.540534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3328,
      "step": 15789
    },
    {
      "epoch": 0.5405682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3414,
      "step": 15790
    },
    {
      "epoch": 0.5406025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3464,
      "step": 15791
    },
    {
      "epoch": 0.5406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3481,
      "step": 15792
    },
    {
      "epoch": 0.5406710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3149,
      "step": 15793
    },
    {
      "epoch": 0.540705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3319,
      "step": 15794
    },
    {
      "epoch": 0.5407394727832934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.351,
      "step": 15795
    },
    {
      "epoch": 0.5407737076343718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3401,
      "step": 15796
    },
    {
      "epoch": 0.5408079424854502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3492,
      "step": 15797
    },
    {
      "epoch": 0.5408421773365286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3335,
      "step": 15798
    },
    {
      "epoch": 0.540876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3361,
      "step": 15799
    },
    {
      "epoch": 0.5409106470386854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3138,
      "step": 15800
    },
    {
      "epoch": 0.5409448818897638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3226,
      "step": 15801
    },
    {
      "epoch": 0.5409791167408422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3553,
      "step": 15802
    },
    {
      "epoch": 0.5410133515919205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3447,
      "step": 15803
    },
    {
      "epoch": 0.541047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.347,
      "step": 15804
    },
    {
      "epoch": 0.5410818212940773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3237,
      "step": 15805
    },
    {
      "epoch": 0.5411160561451558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.34,
      "step": 15806
    },
    {
      "epoch": 0.5411502909962341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.337,
      "step": 15807
    },
    {
      "epoch": 0.5411845258473126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3335,
      "step": 15808
    },
    {
      "epoch": 0.5412187606983909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3531,
      "step": 15809
    },
    {
      "epoch": 0.5412529955494694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3484,
      "step": 15810
    },
    {
      "epoch": 0.5412872304005477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.335,
      "step": 15811
    },
    {
      "epoch": 0.5413214652516262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3242,
      "step": 15812
    },
    {
      "epoch": 0.5413557001027045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.316,
      "step": 15813
    },
    {
      "epoch": 0.541389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3241,
      "step": 15814
    },
    {
      "epoch": 0.5414241698048613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3433,
      "step": 15815
    },
    {
      "epoch": 0.5414584046559398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3408,
      "step": 15816
    },
    {
      "epoch": 0.5414926395070181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3234,
      "step": 15817
    },
    {
      "epoch": 0.5415268743580965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.343,
      "step": 15818
    },
    {
      "epoch": 0.5415611092091749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3452,
      "step": 15819
    },
    {
      "epoch": 0.5415953440602533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3539,
      "step": 15820
    },
    {
      "epoch": 0.5416295789113318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3477,
      "step": 15821
    },
    {
      "epoch": 0.5416638137624101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3373,
      "step": 15822
    },
    {
      "epoch": 0.5416980486134886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3456,
      "step": 15823
    },
    {
      "epoch": 0.5417322834645669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3348,
      "step": 15824
    },
    {
      "epoch": 0.5417665183156454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3601,
      "step": 15825
    },
    {
      "epoch": 0.5418007531667237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3322,
      "step": 15826
    },
    {
      "epoch": 0.5418349880178022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3425,
      "step": 15827
    },
    {
      "epoch": 0.5418692228688805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.338,
      "step": 15828
    },
    {
      "epoch": 0.541903457719959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3287,
      "step": 15829
    },
    {
      "epoch": 0.5419376925710373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.35,
      "step": 15830
    },
    {
      "epoch": 0.5419719274221158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3477,
      "step": 15831
    },
    {
      "epoch": 0.5420061622731941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3383,
      "step": 15832
    },
    {
      "epoch": 0.5420403971242725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.334,
      "step": 15833
    },
    {
      "epoch": 0.5420746319753509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3436,
      "step": 15834
    },
    {
      "epoch": 0.5421088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3394,
      "step": 15835
    },
    {
      "epoch": 0.5421431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3277,
      "step": 15836
    },
    {
      "epoch": 0.5421773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3594,
      "step": 15837
    },
    {
      "epoch": 0.5422115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3535,
      "step": 15838
    },
    {
      "epoch": 0.5422458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3206,
      "step": 15839
    },
    {
      "epoch": 0.5422800410818213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3287,
      "step": 15840
    },
    {
      "epoch": 0.5423142759328997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3212,
      "step": 15841
    },
    {
      "epoch": 0.542348510783978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3444,
      "step": 15842
    },
    {
      "epoch": 0.5423827456350565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3383,
      "step": 15843
    },
    {
      "epoch": 0.5424169804861348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3607,
      "step": 15844
    },
    {
      "epoch": 0.5424512153372133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3412,
      "step": 15845
    },
    {
      "epoch": 0.5424854501882916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3284,
      "step": 15846
    },
    {
      "epoch": 0.5425196850393701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3517,
      "step": 15847
    },
    {
      "epoch": 0.5425539198904484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3359,
      "step": 15848
    },
    {
      "epoch": 0.5425881547415269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3346,
      "step": 15849
    },
    {
      "epoch": 0.5426223895926052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3412,
      "step": 15850
    },
    {
      "epoch": 0.5426566244436837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3302,
      "step": 15851
    },
    {
      "epoch": 0.542690859294762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3562,
      "step": 15852
    },
    {
      "epoch": 0.5427250941458405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3425,
      "step": 15853
    },
    {
      "epoch": 0.5427593289969188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3434,
      "step": 15854
    },
    {
      "epoch": 0.5427935638479973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3252,
      "step": 15855
    },
    {
      "epoch": 0.5428277986990757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3427,
      "step": 15856
    },
    {
      "epoch": 0.542862033550154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3385,
      "step": 15857
    },
    {
      "epoch": 0.5428962684012325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3295,
      "step": 15858
    },
    {
      "epoch": 0.5429305032523108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3637,
      "step": 15859
    },
    {
      "epoch": 0.5429647381033893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3501,
      "step": 15860
    },
    {
      "epoch": 0.5429989729544676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3286,
      "step": 15861
    },
    {
      "epoch": 0.5430332078055461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3617,
      "step": 15862
    },
    {
      "epoch": 0.5430674426566244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3336,
      "step": 15863
    },
    {
      "epoch": 0.5431016775077029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3417,
      "step": 15864
    },
    {
      "epoch": 0.5431359123587812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3501,
      "step": 15865
    },
    {
      "epoch": 0.5431701472098597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3502,
      "step": 15866
    },
    {
      "epoch": 0.543204382060938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3337,
      "step": 15867
    },
    {
      "epoch": 0.5432386169120165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3386,
      "step": 15868
    },
    {
      "epoch": 0.5432728517630948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3312,
      "step": 15869
    },
    {
      "epoch": 0.5433070866141733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3444,
      "step": 15870
    },
    {
      "epoch": 0.5433413214652516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3469,
      "step": 15871
    },
    {
      "epoch": 0.54337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3468,
      "step": 15872
    },
    {
      "epoch": 0.5434097911674084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3165,
      "step": 15873
    },
    {
      "epoch": 0.5434440260184868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3483,
      "step": 15874
    },
    {
      "epoch": 0.5434782608695652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.35,
      "step": 15875
    },
    {
      "epoch": 0.5435124957206436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3527,
      "step": 15876
    },
    {
      "epoch": 0.543546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3214,
      "step": 15877
    },
    {
      "epoch": 0.5435809654228004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3197,
      "step": 15878
    },
    {
      "epoch": 0.5436152002738788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3397,
      "step": 15879
    },
    {
      "epoch": 0.5436494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3437,
      "step": 15880
    },
    {
      "epoch": 0.5436836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3502,
      "step": 15881
    },
    {
      "epoch": 0.543717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3397,
      "step": 15882
    },
    {
      "epoch": 0.5437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3444,
      "step": 15883
    },
    {
      "epoch": 0.5437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3428,
      "step": 15884
    },
    {
      "epoch": 0.5438206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3481,
      "step": 15885
    },
    {
      "epoch": 0.5438548442314276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3497,
      "step": 15886
    },
    {
      "epoch": 0.5438890790825059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3441,
      "step": 15887
    },
    {
      "epoch": 0.5439233139335844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3386,
      "step": 15888
    },
    {
      "epoch": 0.5439575487846627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3311,
      "step": 15889
    },
    {
      "epoch": 0.5439917836357412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3423,
      "step": 15890
    },
    {
      "epoch": 0.5440260184868195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3427,
      "step": 15891
    },
    {
      "epoch": 0.544060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3402,
      "step": 15892
    },
    {
      "epoch": 0.5440944881889764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3349,
      "step": 15893
    },
    {
      "epoch": 0.5441287230400548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.341,
      "step": 15894
    },
    {
      "epoch": 0.5441629578911332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3317,
      "step": 15895
    },
    {
      "epoch": 0.5441971927422116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3383,
      "step": 15896
    },
    {
      "epoch": 0.54423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3544,
      "step": 15897
    },
    {
      "epoch": 0.5442656624443684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3281,
      "step": 15898
    },
    {
      "epoch": 0.5442998972954468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3468,
      "step": 15899
    },
    {
      "epoch": 0.5443341321465252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3394,
      "step": 15900
    },
    {
      "epoch": 0.5443683669976036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3525,
      "step": 15901
    },
    {
      "epoch": 0.5444026018486819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3406,
      "step": 15902
    },
    {
      "epoch": 0.5444368366997604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3382,
      "step": 15903
    },
    {
      "epoch": 0.5444710715508387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3328,
      "step": 15904
    },
    {
      "epoch": 0.5445053064019172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3416,
      "step": 15905
    },
    {
      "epoch": 0.5445395412529955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3375,
      "step": 15906
    },
    {
      "epoch": 0.544573776104074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.334,
      "step": 15907
    },
    {
      "epoch": 0.5446080109551523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3407,
      "step": 15908
    },
    {
      "epoch": 0.5446422458062308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3458,
      "step": 15909
    },
    {
      "epoch": 0.5446764806573091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3287,
      "step": 15910
    },
    {
      "epoch": 0.5447107155083876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3405,
      "step": 15911
    },
    {
      "epoch": 0.5447449503594659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3523,
      "step": 15912
    },
    {
      "epoch": 0.5447791852105444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3483,
      "step": 15913
    },
    {
      "epoch": 0.5448134200616227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3504,
      "step": 15914
    },
    {
      "epoch": 0.5448476549127012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3362,
      "step": 15915
    },
    {
      "epoch": 0.5448818897637795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3436,
      "step": 15916
    },
    {
      "epoch": 0.5449161246148579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3405,
      "step": 15917
    },
    {
      "epoch": 0.5449503594659363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3441,
      "step": 15918
    },
    {
      "epoch": 0.5449845943170147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3145,
      "step": 15919
    },
    {
      "epoch": 0.5450188291680931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3167,
      "step": 15920
    },
    {
      "epoch": 0.5450530640191715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3431,
      "step": 15921
    },
    {
      "epoch": 0.5450872988702499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3515,
      "step": 15922
    },
    {
      "epoch": 0.5451215337213283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3194,
      "step": 15923
    },
    {
      "epoch": 0.5451557685724067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3289,
      "step": 15924
    },
    {
      "epoch": 0.5451900034234851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3449,
      "step": 15925
    },
    {
      "epoch": 0.5452242382745635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3284,
      "step": 15926
    },
    {
      "epoch": 0.5452584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3416,
      "step": 15927
    },
    {
      "epoch": 0.5452927079767202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3232,
      "step": 15928
    },
    {
      "epoch": 0.5453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3488,
      "step": 15929
    },
    {
      "epoch": 0.5453611776788772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3698,
      "step": 15930
    },
    {
      "epoch": 0.5453954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3211,
      "step": 15931
    },
    {
      "epoch": 0.5454296473810339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3198,
      "step": 15932
    },
    {
      "epoch": 0.5454638822321123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.335,
      "step": 15933
    },
    {
      "epoch": 0.5454981170831907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3413,
      "step": 15934
    },
    {
      "epoch": 0.5455323519342691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3425,
      "step": 15935
    },
    {
      "epoch": 0.5455665867853475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3276,
      "step": 15936
    },
    {
      "epoch": 0.5456008216364259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3272,
      "step": 15937
    },
    {
      "epoch": 0.5456350564875043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3521,
      "step": 15938
    },
    {
      "epoch": 0.5456692913385827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3503,
      "step": 15939
    },
    {
      "epoch": 0.5457035261896611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.335,
      "step": 15940
    },
    {
      "epoch": 0.5457377610407395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3477,
      "step": 15941
    },
    {
      "epoch": 0.5457719958918179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3408,
      "step": 15942
    },
    {
      "epoch": 0.5458062307428962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3357,
      "step": 15943
    },
    {
      "epoch": 0.5458404655939747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3558,
      "step": 15944
    },
    {
      "epoch": 0.545874700445053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3373,
      "step": 15945
    },
    {
      "epoch": 0.5459089352961315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3304,
      "step": 15946
    },
    {
      "epoch": 0.5459431701472098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3397,
      "step": 15947
    },
    {
      "epoch": 0.5459774049982883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3088,
      "step": 15948
    },
    {
      "epoch": 0.5460116398493666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3383,
      "step": 15949
    },
    {
      "epoch": 0.5460458747004451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.355,
      "step": 15950
    },
    {
      "epoch": 0.5460801095515234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3263,
      "step": 15951
    },
    {
      "epoch": 0.5461143444026019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3388,
      "step": 15952
    },
    {
      "epoch": 0.5461485792536802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3434,
      "step": 15953
    },
    {
      "epoch": 0.5461828141047587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3303,
      "step": 15954
    },
    {
      "epoch": 0.546217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3331,
      "step": 15955
    },
    {
      "epoch": 0.5462512838069155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3304,
      "step": 15956
    },
    {
      "epoch": 0.5462855186579938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3414,
      "step": 15957
    },
    {
      "epoch": 0.5463197535090722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3274,
      "step": 15958
    },
    {
      "epoch": 0.5463539883601506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3378,
      "step": 15959
    },
    {
      "epoch": 0.546388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3281,
      "step": 15960
    },
    {
      "epoch": 0.5464224580623074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3362,
      "step": 15961
    },
    {
      "epoch": 0.5464566929133858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3436,
      "step": 15962
    },
    {
      "epoch": 0.5464909277644642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.323,
      "step": 15963
    },
    {
      "epoch": 0.5465251626155426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3243,
      "step": 15964
    },
    {
      "epoch": 0.546559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3376,
      "step": 15965
    },
    {
      "epoch": 0.5465936323176994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3544,
      "step": 15966
    },
    {
      "epoch": 0.5466278671687779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3339,
      "step": 15967
    },
    {
      "epoch": 0.5466621020198562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3541,
      "step": 15968
    },
    {
      "epoch": 0.5466963368709347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3273,
      "step": 15969
    },
    {
      "epoch": 0.546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3254,
      "step": 15970
    },
    {
      "epoch": 0.5467648065730915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3705,
      "step": 15971
    },
    {
      "epoch": 0.5467990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3382,
      "step": 15972
    },
    {
      "epoch": 0.5468332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3272,
      "step": 15973
    },
    {
      "epoch": 0.5468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3471,
      "step": 15974
    },
    {
      "epoch": 0.546901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3294,
      "step": 15975
    },
    {
      "epoch": 0.5469359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3422,
      "step": 15976
    },
    {
      "epoch": 0.5469702156795618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3308,
      "step": 15977
    },
    {
      "epoch": 0.5470044505306402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3399,
      "step": 15978
    },
    {
      "epoch": 0.5470386853817186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3236,
      "step": 15979
    },
    {
      "epoch": 0.547072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.324,
      "step": 15980
    },
    {
      "epoch": 0.5471071550838754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3301,
      "step": 15981
    },
    {
      "epoch": 0.5471413899349538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3474,
      "step": 15982
    },
    {
      "epoch": 0.5471756247860322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3253,
      "step": 15983
    },
    {
      "epoch": 0.5472098596371106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3353,
      "step": 15984
    },
    {
      "epoch": 0.547244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3178,
      "step": 15985
    },
    {
      "epoch": 0.5472783293392673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3239,
      "step": 15986
    },
    {
      "epoch": 0.5473125641903458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3202,
      "step": 15987
    },
    {
      "epoch": 0.5473467990414241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3533,
      "step": 15988
    },
    {
      "epoch": 0.5473810338925026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3351,
      "step": 15989
    },
    {
      "epoch": 0.5474152687435809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.344,
      "step": 15990
    },
    {
      "epoch": 0.5474495035946594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3318,
      "step": 15991
    },
    {
      "epoch": 0.5474837384457377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3359,
      "step": 15992
    },
    {
      "epoch": 0.5475179732968162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3182,
      "step": 15993
    },
    {
      "epoch": 0.5475522081478945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3359,
      "step": 15994
    },
    {
      "epoch": 0.547586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3417,
      "step": 15995
    },
    {
      "epoch": 0.5476206778500513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.335,
      "step": 15996
    },
    {
      "epoch": 0.5476549127011298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3336,
      "step": 15997
    },
    {
      "epoch": 0.5476891475522081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3468,
      "step": 15998
    },
    {
      "epoch": 0.5477233824032866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.33,
      "step": 15999
    },
    {
      "epoch": 0.5477576172543649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3324,
      "step": 16000
    },
    {
      "epoch": 0.5477918521054433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3401,
      "step": 16001
    },
    {
      "epoch": 0.5478260869565217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3336,
      "step": 16002
    },
    {
      "epoch": 0.5478603218076001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3335,
      "step": 16003
    },
    {
      "epoch": 0.5478945566586786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3469,
      "step": 16004
    },
    {
      "epoch": 0.5479287915097569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3266,
      "step": 16005
    },
    {
      "epoch": 0.5479630263608354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3553,
      "step": 16006
    },
    {
      "epoch": 0.5479972612119137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3334,
      "step": 16007
    },
    {
      "epoch": 0.5480314960629922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3238,
      "step": 16008
    },
    {
      "epoch": 0.5480657309140705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.345,
      "step": 16009
    },
    {
      "epoch": 0.548099965765149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3527,
      "step": 16010
    },
    {
      "epoch": 0.5481342006162273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3301,
      "step": 16011
    },
    {
      "epoch": 0.5481684354673058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3698,
      "step": 16012
    },
    {
      "epoch": 0.5482026703183841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3289,
      "step": 16013
    },
    {
      "epoch": 0.5482369051694626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3335,
      "step": 16014
    },
    {
      "epoch": 0.5482711400205409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3304,
      "step": 16015
    },
    {
      "epoch": 0.5483053748716193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3435,
      "step": 16016
    },
    {
      "epoch": 0.5483396097226977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3109,
      "step": 16017
    },
    {
      "epoch": 0.5483738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3539,
      "step": 16018
    },
    {
      "epoch": 0.5484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3355,
      "step": 16019
    },
    {
      "epoch": 0.5484423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3394,
      "step": 16020
    },
    {
      "epoch": 0.5484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3287,
      "step": 16021
    },
    {
      "epoch": 0.5485107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3374,
      "step": 16022
    },
    {
      "epoch": 0.5485450188291681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3308,
      "step": 16023
    },
    {
      "epoch": 0.5485792536802465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3417,
      "step": 16024
    },
    {
      "epoch": 0.5486134885313249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3438,
      "step": 16025
    },
    {
      "epoch": 0.5486477233824033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3047,
      "step": 16026
    },
    {
      "epoch": 0.5486819582334816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3364,
      "step": 16027
    },
    {
      "epoch": 0.5487161930845601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3261,
      "step": 16028
    },
    {
      "epoch": 0.5487504279356384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3306,
      "step": 16029
    },
    {
      "epoch": 0.5487846627867169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3326,
      "step": 16030
    },
    {
      "epoch": 0.5488188976377952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3328,
      "step": 16031
    },
    {
      "epoch": 0.5488531324888737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.328,
      "step": 16032
    },
    {
      "epoch": 0.548887367339952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.329,
      "step": 16033
    },
    {
      "epoch": 0.5489216021910305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3366,
      "step": 16034
    },
    {
      "epoch": 0.5489558370421088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3422,
      "step": 16035
    },
    {
      "epoch": 0.5489900718931873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3219,
      "step": 16036
    },
    {
      "epoch": 0.5490243067442656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3369,
      "step": 16037
    },
    {
      "epoch": 0.5490585415953441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3291,
      "step": 16038
    },
    {
      "epoch": 0.5490927764464224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3308,
      "step": 16039
    },
    {
      "epoch": 0.5491270112975009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3607,
      "step": 16040
    },
    {
      "epoch": 0.5491612461485793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3254,
      "step": 16041
    },
    {
      "epoch": 0.5491954809996576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3427,
      "step": 16042
    },
    {
      "epoch": 0.5492297158507361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3238,
      "step": 16043
    },
    {
      "epoch": 0.5492639507018144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3457,
      "step": 16044
    },
    {
      "epoch": 0.5492981855528929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3378,
      "step": 16045
    },
    {
      "epoch": 0.5493324204039712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3179,
      "step": 16046
    },
    {
      "epoch": 0.5493666552550497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3181,
      "step": 16047
    },
    {
      "epoch": 0.549400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3381,
      "step": 16048
    },
    {
      "epoch": 0.5494351249572065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3376,
      "step": 16049
    },
    {
      "epoch": 0.5494693598082848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3276,
      "step": 16050
    },
    {
      "epoch": 0.5495035946593633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.337,
      "step": 16051
    },
    {
      "epoch": 0.5495378295104416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3282,
      "step": 16052
    },
    {
      "epoch": 0.5495720643615201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3226,
      "step": 16053
    },
    {
      "epoch": 0.5496062992125984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3464,
      "step": 16054
    },
    {
      "epoch": 0.5496405340636769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3163,
      "step": 16055
    },
    {
      "epoch": 0.5496747689147552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3244,
      "step": 16056
    },
    {
      "epoch": 0.5497090037658336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3161,
      "step": 16057
    },
    {
      "epoch": 0.549743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3333,
      "step": 16058
    },
    {
      "epoch": 0.5497774734679904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3445,
      "step": 16059
    },
    {
      "epoch": 0.5498117083190688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.336,
      "step": 16060
    },
    {
      "epoch": 0.5498459431701472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3388,
      "step": 16061
    },
    {
      "epoch": 0.5498801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3392,
      "step": 16062
    },
    {
      "epoch": 0.549914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3345,
      "step": 16063
    },
    {
      "epoch": 0.5499486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3519,
      "step": 16064
    },
    {
      "epoch": 0.5499828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3316,
      "step": 16065
    },
    {
      "epoch": 0.5500171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3376,
      "step": 16066
    },
    {
      "epoch": 0.5500513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3312,
      "step": 16067
    },
    {
      "epoch": 0.550085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3403,
      "step": 16068
    },
    {
      "epoch": 0.5501198219787744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3442,
      "step": 16069
    },
    {
      "epoch": 0.5501540568298527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.322,
      "step": 16070
    },
    {
      "epoch": 0.5501882916809312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3452,
      "step": 16071
    },
    {
      "epoch": 0.5502225265320095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3315,
      "step": 16072
    },
    {
      "epoch": 0.550256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3353,
      "step": 16073
    },
    {
      "epoch": 0.5502909962341663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3279,
      "step": 16074
    },
    {
      "epoch": 0.5503252310852448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3358,
      "step": 16075
    },
    {
      "epoch": 0.5503594659363231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.345,
      "step": 16076
    },
    {
      "epoch": 0.5503937007874016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3424,
      "step": 16077
    },
    {
      "epoch": 0.55042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3386,
      "step": 16078
    },
    {
      "epoch": 0.5504621704895584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3408,
      "step": 16079
    },
    {
      "epoch": 0.5504964053406368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3275,
      "step": 16080
    },
    {
      "epoch": 0.5505306401917152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3262,
      "step": 16081
    },
    {
      "epoch": 0.5505648750427936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.33,
      "step": 16082
    },
    {
      "epoch": 0.550599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3292,
      "step": 16083
    },
    {
      "epoch": 0.5506333447449504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.314,
      "step": 16084
    },
    {
      "epoch": 0.5506675795960287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3437,
      "step": 16085
    },
    {
      "epoch": 0.5507018144471072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3289,
      "step": 16086
    },
    {
      "epoch": 0.5507360492981855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3171,
      "step": 16087
    },
    {
      "epoch": 0.550770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.335,
      "step": 16088
    },
    {
      "epoch": 0.5508045190003423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3314,
      "step": 16089
    },
    {
      "epoch": 0.5508387538514208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3241,
      "step": 16090
    },
    {
      "epoch": 0.5508729887024991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3406,
      "step": 16091
    },
    {
      "epoch": 0.5509072235535776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3478,
      "step": 16092
    },
    {
      "epoch": 0.5509414584046559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3376,
      "step": 16093
    },
    {
      "epoch": 0.5509756932557344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3298,
      "step": 16094
    },
    {
      "epoch": 0.5510099281068127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3232,
      "step": 16095
    },
    {
      "epoch": 0.5510441629578912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3361,
      "step": 16096
    },
    {
      "epoch": 0.5510783978089695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3105,
      "step": 16097
    },
    {
      "epoch": 0.551112632660048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3226,
      "step": 16098
    },
    {
      "epoch": 0.5511468675111263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3402,
      "step": 16099
    },
    {
      "epoch": 0.5511811023622047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3363,
      "step": 16100
    },
    {
      "epoch": 0.5512153372132831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3337,
      "step": 16101
    },
    {
      "epoch": 0.5512495720643615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3421,
      "step": 16102
    },
    {
      "epoch": 0.5512838069154399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3235,
      "step": 16103
    },
    {
      "epoch": 0.5513180417665183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3256,
      "step": 16104
    },
    {
      "epoch": 0.5513522766175967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3484,
      "step": 16105
    },
    {
      "epoch": 0.5513865114686751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3328,
      "step": 16106
    },
    {
      "epoch": 0.5514207463197535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3396,
      "step": 16107
    },
    {
      "epoch": 0.5514549811708319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3332,
      "step": 16108
    },
    {
      "epoch": 0.5514892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.314,
      "step": 16109
    },
    {
      "epoch": 0.5515234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3362,
      "step": 16110
    },
    {
      "epoch": 0.551557685724067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3316,
      "step": 16111
    },
    {
      "epoch": 0.5515919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3298,
      "step": 16112
    },
    {
      "epoch": 0.5516261554262238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3306,
      "step": 16113
    },
    {
      "epoch": 0.5516603902773023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3351,
      "step": 16114
    },
    {
      "epoch": 0.5516946251283807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3248,
      "step": 16115
    },
    {
      "epoch": 0.5517288599794591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.335,
      "step": 16116
    },
    {
      "epoch": 0.5517630948305375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3331,
      "step": 16117
    },
    {
      "epoch": 0.5517973296816159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.338,
      "step": 16118
    },
    {
      "epoch": 0.5518315645326943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3273,
      "step": 16119
    },
    {
      "epoch": 0.5518657993837727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3147,
      "step": 16120
    },
    {
      "epoch": 0.5519000342348511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3168,
      "step": 16121
    },
    {
      "epoch": 0.5519342690859295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3242,
      "step": 16122
    },
    {
      "epoch": 0.5519685039370079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3415,
      "step": 16123
    },
    {
      "epoch": 0.5520027387880863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3215,
      "step": 16124
    },
    {
      "epoch": 0.5520369736391647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3465,
      "step": 16125
    },
    {
      "epoch": 0.552071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3178,
      "step": 16126
    },
    {
      "epoch": 0.5521054433413215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3237,
      "step": 16127
    },
    {
      "epoch": 0.5521396781923998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3197,
      "step": 16128
    },
    {
      "epoch": 0.5521739130434783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3367,
      "step": 16129
    },
    {
      "epoch": 0.5522081478945566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3334,
      "step": 16130
    },
    {
      "epoch": 0.5522423827456351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3259,
      "step": 16131
    },
    {
      "epoch": 0.5522766175967134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.333,
      "step": 16132
    },
    {
      "epoch": 0.5523108524477919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3288,
      "step": 16133
    },
    {
      "epoch": 0.5523450872988702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3475,
      "step": 16134
    },
    {
      "epoch": 0.5523793221499487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3199,
      "step": 16135
    },
    {
      "epoch": 0.552413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3424,
      "step": 16136
    },
    {
      "epoch": 0.5524477918521055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3228,
      "step": 16137
    },
    {
      "epoch": 0.5524820267031838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3268,
      "step": 16138
    },
    {
      "epoch": 0.5525162615542623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3414,
      "step": 16139
    },
    {
      "epoch": 0.5525504964053406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3225,
      "step": 16140
    },
    {
      "epoch": 0.552584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3285,
      "step": 16141
    },
    {
      "epoch": 0.5526189661074974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3303,
      "step": 16142
    },
    {
      "epoch": 0.5526532009585758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3506,
      "step": 16143
    },
    {
      "epoch": 0.5526874358096542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3607,
      "step": 16144
    },
    {
      "epoch": 0.5527216706607326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3296,
      "step": 16145
    },
    {
      "epoch": 0.552755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3355,
      "step": 16146
    },
    {
      "epoch": 0.5527901403628894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3198,
      "step": 16147
    },
    {
      "epoch": 0.5528243752139678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3401,
      "step": 16148
    },
    {
      "epoch": 0.5528586100650462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3347,
      "step": 16149
    },
    {
      "epoch": 0.5528928449161246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3499,
      "step": 16150
    },
    {
      "epoch": 0.552927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3194,
      "step": 16151
    },
    {
      "epoch": 0.5529613146182815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3035,
      "step": 16152
    },
    {
      "epoch": 0.5529955494693598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.329,
      "step": 16153
    },
    {
      "epoch": 0.5530297843204383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3402,
      "step": 16154
    },
    {
      "epoch": 0.5530640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3076,
      "step": 16155
    },
    {
      "epoch": 0.553098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.311,
      "step": 16156
    },
    {
      "epoch": 0.5531324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3447,
      "step": 16157
    },
    {
      "epoch": 0.5531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3488,
      "step": 16158
    },
    {
      "epoch": 0.5532009585758302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3215,
      "step": 16159
    },
    {
      "epoch": 0.5532351934269086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3251,
      "step": 16160
    },
    {
      "epoch": 0.553269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3195,
      "step": 16161
    },
    {
      "epoch": 0.5533036631290654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3285,
      "step": 16162
    },
    {
      "epoch": 0.5533378979801438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3339,
      "step": 16163
    },
    {
      "epoch": 0.5533721328312222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3236,
      "step": 16164
    },
    {
      "epoch": 0.5534063676823006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3344,
      "step": 16165
    },
    {
      "epoch": 0.553440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3201,
      "step": 16166
    },
    {
      "epoch": 0.5534748373844574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.329,
      "step": 16167
    },
    {
      "epoch": 0.5535090722355358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3218,
      "step": 16168
    },
    {
      "epoch": 0.5535433070866141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3241,
      "step": 16169
    },
    {
      "epoch": 0.5535775419376926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3414,
      "step": 16170
    },
    {
      "epoch": 0.5536117767887709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3521,
      "step": 16171
    },
    {
      "epoch": 0.5536460116398494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3312,
      "step": 16172
    },
    {
      "epoch": 0.5536802464909277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.323,
      "step": 16173
    },
    {
      "epoch": 0.5537144813420062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3319,
      "step": 16174
    },
    {
      "epoch": 0.5537487161930845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3423,
      "step": 16175
    },
    {
      "epoch": 0.553782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3296,
      "step": 16176
    },
    {
      "epoch": 0.5538171858952413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3386,
      "step": 16177
    },
    {
      "epoch": 0.5538514207463198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3024,
      "step": 16178
    },
    {
      "epoch": 0.5538856555973981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3259,
      "step": 16179
    },
    {
      "epoch": 0.5539198904484766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3218,
      "step": 16180
    },
    {
      "epoch": 0.5539541252995549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3337,
      "step": 16181
    },
    {
      "epoch": 0.5539883601506334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3326,
      "step": 16182
    },
    {
      "epoch": 0.5540225950017117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3451,
      "step": 16183
    },
    {
      "epoch": 0.5540568298527901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3277,
      "step": 16184
    },
    {
      "epoch": 0.5540910647038685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3293,
      "step": 16185
    },
    {
      "epoch": 0.5541252995549469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3351,
      "step": 16186
    },
    {
      "epoch": 0.5541595344060254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3434,
      "step": 16187
    },
    {
      "epoch": 0.5541937692571037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.343,
      "step": 16188
    },
    {
      "epoch": 0.5542280041081822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3371,
      "step": 16189
    },
    {
      "epoch": 0.5542622389592605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3364,
      "step": 16190
    },
    {
      "epoch": 0.554296473810339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3413,
      "step": 16191
    },
    {
      "epoch": 0.5543307086614173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3242,
      "step": 16192
    },
    {
      "epoch": 0.5543649435124958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3292,
      "step": 16193
    },
    {
      "epoch": 0.5543991783635741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3228,
      "step": 16194
    },
    {
      "epoch": 0.5544334132146526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3239,
      "step": 16195
    },
    {
      "epoch": 0.5544676480657309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3511,
      "step": 16196
    },
    {
      "epoch": 0.5545018829168094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3449,
      "step": 16197
    },
    {
      "epoch": 0.5545361177678877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3253,
      "step": 16198
    },
    {
      "epoch": 0.5545703526189661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.334,
      "step": 16199
    },
    {
      "epoch": 0.5546045874700445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.35,
      "step": 16200
    },
    {
      "epoch": 0.5546388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3187,
      "step": 16201
    },
    {
      "epoch": 0.5546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3497,
      "step": 16202
    },
    {
      "epoch": 0.5547072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3244,
      "step": 16203
    },
    {
      "epoch": 0.5547415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3262,
      "step": 16204
    },
    {
      "epoch": 0.5547757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3235,
      "step": 16205
    },
    {
      "epoch": 0.5548099965765149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3423,
      "step": 16206
    },
    {
      "epoch": 0.5548442314275933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3322,
      "step": 16207
    },
    {
      "epoch": 0.5548784662786717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3184,
      "step": 16208
    },
    {
      "epoch": 0.5549127011297501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3212,
      "step": 16209
    },
    {
      "epoch": 0.5549469359808284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.314,
      "step": 16210
    },
    {
      "epoch": 0.5549811708319069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3386,
      "step": 16211
    },
    {
      "epoch": 0.5550154056829852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3397,
      "step": 16212
    },
    {
      "epoch": 0.5550496405340637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3293,
      "step": 16213
    },
    {
      "epoch": 0.555083875385142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3281,
      "step": 16214
    },
    {
      "epoch": 0.5551181102362205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3256,
      "step": 16215
    },
    {
      "epoch": 0.5551523450872988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3384,
      "step": 16216
    },
    {
      "epoch": 0.5551865799383773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3236,
      "step": 16217
    },
    {
      "epoch": 0.5552208147894556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3334,
      "step": 16218
    },
    {
      "epoch": 0.5552550496405341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3299,
      "step": 16219
    },
    {
      "epoch": 0.5552892844916124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3463,
      "step": 16220
    },
    {
      "epoch": 0.5553235193426909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3275,
      "step": 16221
    },
    {
      "epoch": 0.5553577541937692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3157,
      "step": 16222
    },
    {
      "epoch": 0.5553919890448477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3485,
      "step": 16223
    },
    {
      "epoch": 0.5554262238959261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3479,
      "step": 16224
    },
    {
      "epoch": 0.5554604587470044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3141,
      "step": 16225
    },
    {
      "epoch": 0.5554946935980829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3184,
      "step": 16226
    },
    {
      "epoch": 0.5555289284491612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3271,
      "step": 16227
    },
    {
      "epoch": 0.5555631633002397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3308,
      "step": 16228
    },
    {
      "epoch": 0.555597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3333,
      "step": 16229
    },
    {
      "epoch": 0.5556316330023965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3387,
      "step": 16230
    },
    {
      "epoch": 0.5556658678534748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3242,
      "step": 16231
    },
    {
      "epoch": 0.5557001027045533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3282,
      "step": 16232
    },
    {
      "epoch": 0.5557343375556316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3416,
      "step": 16233
    },
    {
      "epoch": 0.5557685724067101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.327,
      "step": 16234
    },
    {
      "epoch": 0.5558028072577884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3373,
      "step": 16235
    },
    {
      "epoch": 0.5558370421088669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3422,
      "step": 16236
    },
    {
      "epoch": 0.5558712769599452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3285,
      "step": 16237
    },
    {
      "epoch": 0.5559055118110237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3223,
      "step": 16238
    },
    {
      "epoch": 0.555939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3237,
      "step": 16239
    },
    {
      "epoch": 0.5559739815131804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3497,
      "step": 16240
    },
    {
      "epoch": 0.5560082163642588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.321,
      "step": 16241
    },
    {
      "epoch": 0.5560424512153372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3186,
      "step": 16242
    },
    {
      "epoch": 0.5560766860664156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.333,
      "step": 16243
    },
    {
      "epoch": 0.556110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2999,
      "step": 16244
    },
    {
      "epoch": 0.5561451557685724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3364,
      "step": 16245
    },
    {
      "epoch": 0.5561793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3319,
      "step": 16246
    },
    {
      "epoch": 0.5562136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3066,
      "step": 16247
    },
    {
      "epoch": 0.5562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.316,
      "step": 16248
    },
    {
      "epoch": 0.556282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3157,
      "step": 16249
    },
    {
      "epoch": 0.5563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3261,
      "step": 16250
    },
    {
      "epoch": 0.5563505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.319,
      "step": 16251
    },
    {
      "epoch": 0.5563847997261212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3317,
      "step": 16252
    },
    {
      "epoch": 0.5564190345771995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3494,
      "step": 16253
    },
    {
      "epoch": 0.556453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3408,
      "step": 16254
    },
    {
      "epoch": 0.5564875042793563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3377,
      "step": 16255
    },
    {
      "epoch": 0.5565217391304348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3157,
      "step": 16256
    },
    {
      "epoch": 0.5565559739815131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.325,
      "step": 16257
    },
    {
      "epoch": 0.5565902088325916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3308,
      "step": 16258
    },
    {
      "epoch": 0.5566244436836699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.319,
      "step": 16259
    },
    {
      "epoch": 0.5566586785347484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3483,
      "step": 16260
    },
    {
      "epoch": 0.5566929133858268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3229,
      "step": 16261
    },
    {
      "epoch": 0.5567271482369052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3252,
      "step": 16262
    },
    {
      "epoch": 0.5567613830879836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3246,
      "step": 16263
    },
    {
      "epoch": 0.556795617939062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.322,
      "step": 16264
    },
    {
      "epoch": 0.5568298527901404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3518,
      "step": 16265
    },
    {
      "epoch": 0.5568640876412188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3308,
      "step": 16266
    },
    {
      "epoch": 0.5568983224922972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3488,
      "step": 16267
    },
    {
      "epoch": 0.5569325573433755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3286,
      "step": 16268
    },
    {
      "epoch": 0.556966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3331,
      "step": 16269
    },
    {
      "epoch": 0.5570010270455323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3328,
      "step": 16270
    },
    {
      "epoch": 0.5570352618966108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3401,
      "step": 16271
    },
    {
      "epoch": 0.5570694967476891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3523,
      "step": 16272
    },
    {
      "epoch": 0.5571037315987676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3108,
      "step": 16273
    },
    {
      "epoch": 0.5571379664498459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3205,
      "step": 16274
    },
    {
      "epoch": 0.5571722013009244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3277,
      "step": 16275
    },
    {
      "epoch": 0.5572064361520027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3156,
      "step": 16276
    },
    {
      "epoch": 0.5572406710030812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.332,
      "step": 16277
    },
    {
      "epoch": 0.5572749058541595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3464,
      "step": 16278
    },
    {
      "epoch": 0.557309140705238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3272,
      "step": 16279
    },
    {
      "epoch": 0.5573433755563163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3248,
      "step": 16280
    },
    {
      "epoch": 0.5573776104073948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3378,
      "step": 16281
    },
    {
      "epoch": 0.5574118452584731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3254,
      "step": 16282
    },
    {
      "epoch": 0.5574460801095515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3225,
      "step": 16283
    },
    {
      "epoch": 0.5574803149606299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3246,
      "step": 16284
    },
    {
      "epoch": 0.5575145498117083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3083,
      "step": 16285
    },
    {
      "epoch": 0.5575487846627867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3121,
      "step": 16286
    },
    {
      "epoch": 0.5575830195138651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3416,
      "step": 16287
    },
    {
      "epoch": 0.5576172543649435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3391,
      "step": 16288
    },
    {
      "epoch": 0.5576514892160219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3341,
      "step": 16289
    },
    {
      "epoch": 0.5576857240671003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3282,
      "step": 16290
    },
    {
      "epoch": 0.5577199589181787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3232,
      "step": 16291
    },
    {
      "epoch": 0.5577541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3068,
      "step": 16292
    },
    {
      "epoch": 0.5577884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3185,
      "step": 16293
    },
    {
      "epoch": 0.5578226634714138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3146,
      "step": 16294
    },
    {
      "epoch": 0.5578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3432,
      "step": 16295
    },
    {
      "epoch": 0.5578911331735706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3187,
      "step": 16296
    },
    {
      "epoch": 0.5579253680246491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3286,
      "step": 16297
    },
    {
      "epoch": 0.5579596028757275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3223,
      "step": 16298
    },
    {
      "epoch": 0.5579938377268059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3248,
      "step": 16299
    },
    {
      "epoch": 0.5580280725778843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3243,
      "step": 16300
    },
    {
      "epoch": 0.5580623074289627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3188,
      "step": 16301
    },
    {
      "epoch": 0.5580965422800411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3423,
      "step": 16302
    },
    {
      "epoch": 0.5581307771311195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3176,
      "step": 16303
    },
    {
      "epoch": 0.5581650119821979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.32,
      "step": 16304
    },
    {
      "epoch": 0.5581992468332763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.324,
      "step": 16305
    },
    {
      "epoch": 0.5582334816843547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3502,
      "step": 16306
    },
    {
      "epoch": 0.5582677165354331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3167,
      "step": 16307
    },
    {
      "epoch": 0.5583019513865115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3258,
      "step": 16308
    },
    {
      "epoch": 0.5583361862375898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3294,
      "step": 16309
    },
    {
      "epoch": 0.5583704210886683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3199,
      "step": 16310
    },
    {
      "epoch": 0.5584046559397466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3356,
      "step": 16311
    },
    {
      "epoch": 0.5584388907908251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3246,
      "step": 16312
    },
    {
      "epoch": 0.5584731256419034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3494,
      "step": 16313
    },
    {
      "epoch": 0.5585073604929819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3306,
      "step": 16314
    },
    {
      "epoch": 0.5585415953440602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3085,
      "step": 16315
    },
    {
      "epoch": 0.5585758301951387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3266,
      "step": 16316
    },
    {
      "epoch": 0.558610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3367,
      "step": 16317
    },
    {
      "epoch": 0.5586442998972955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.338,
      "step": 16318
    },
    {
      "epoch": 0.5586785347483738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3259,
      "step": 16319
    },
    {
      "epoch": 0.5587127695994523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3181,
      "step": 16320
    },
    {
      "epoch": 0.5587470044505306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3346,
      "step": 16321
    },
    {
      "epoch": 0.5587812393016091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.299,
      "step": 16322
    },
    {
      "epoch": 0.5588154741526874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.318,
      "step": 16323
    },
    {
      "epoch": 0.5588497090037658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3416,
      "step": 16324
    },
    {
      "epoch": 0.5588839438548442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3325,
      "step": 16325
    },
    {
      "epoch": 0.5589181787059226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.332,
      "step": 16326
    },
    {
      "epoch": 0.558952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3291,
      "step": 16327
    },
    {
      "epoch": 0.5589866484080794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3184,
      "step": 16328
    },
    {
      "epoch": 0.5590208832591578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3282,
      "step": 16329
    },
    {
      "epoch": 0.5590551181102362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3251,
      "step": 16330
    },
    {
      "epoch": 0.5590893529613146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3239,
      "step": 16331
    },
    {
      "epoch": 0.559123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3052,
      "step": 16332
    },
    {
      "epoch": 0.5591578226634714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3279,
      "step": 16333
    },
    {
      "epoch": 0.5591920575145498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3288,
      "step": 16334
    },
    {
      "epoch": 0.5592262923656283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3402,
      "step": 16335
    },
    {
      "epoch": 0.5592605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3254,
      "step": 16336
    },
    {
      "epoch": 0.5592947620677851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3291,
      "step": 16337
    },
    {
      "epoch": 0.5593289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3167,
      "step": 16338
    },
    {
      "epoch": 0.5593632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3396,
      "step": 16339
    },
    {
      "epoch": 0.5593974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3164,
      "step": 16340
    },
    {
      "epoch": 0.5594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3301,
      "step": 16341
    },
    {
      "epoch": 0.559465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3334,
      "step": 16342
    },
    {
      "epoch": 0.5595001711742554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3137,
      "step": 16343
    },
    {
      "epoch": 0.5595344060253338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3182,
      "step": 16344
    },
    {
      "epoch": 0.5595686408764122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3477,
      "step": 16345
    },
    {
      "epoch": 0.5596028757274906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3311,
      "step": 16346
    },
    {
      "epoch": 0.559637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3277,
      "step": 16347
    },
    {
      "epoch": 0.5596713454296474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3203,
      "step": 16348
    },
    {
      "epoch": 0.5597055802807258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3286,
      "step": 16349
    },
    {
      "epoch": 0.5597398151318042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3278,
      "step": 16350
    },
    {
      "epoch": 0.5597740499828826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3262,
      "step": 16351
    },
    {
      "epoch": 0.559808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.337,
      "step": 16352
    },
    {
      "epoch": 0.5598425196850394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3226,
      "step": 16353
    },
    {
      "epoch": 0.5598767545361177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3262,
      "step": 16354
    },
    {
      "epoch": 0.5599109893871962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3095,
      "step": 16355
    },
    {
      "epoch": 0.5599452242382745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3097,
      "step": 16356
    },
    {
      "epoch": 0.559979459089353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3382,
      "step": 16357
    },
    {
      "epoch": 0.5600136939404313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3319,
      "step": 16358
    },
    {
      "epoch": 0.5600479287915098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3303,
      "step": 16359
    },
    {
      "epoch": 0.5600821636425881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3311,
      "step": 16360
    },
    {
      "epoch": 0.5601163984936666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.337,
      "step": 16361
    },
    {
      "epoch": 0.5601506333447449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3179,
      "step": 16362
    },
    {
      "epoch": 0.5601848681958234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3205,
      "step": 16363
    },
    {
      "epoch": 0.5602191030469017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3166,
      "step": 16364
    },
    {
      "epoch": 0.5602533378979802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3338,
      "step": 16365
    },
    {
      "epoch": 0.5602875727490585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3331,
      "step": 16366
    },
    {
      "epoch": 0.560321807600137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3098,
      "step": 16367
    },
    {
      "epoch": 0.5603560424512153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3433,
      "step": 16368
    },
    {
      "epoch": 0.5603902773022937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3123,
      "step": 16369
    },
    {
      "epoch": 0.5604245121533721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3413,
      "step": 16370
    },
    {
      "epoch": 0.5604587470044505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3278,
      "step": 16371
    },
    {
      "epoch": 0.560492981855529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3302,
      "step": 16372
    },
    {
      "epoch": 0.5605272167066073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3081,
      "step": 16373
    },
    {
      "epoch": 0.5605614515576858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3274,
      "step": 16374
    },
    {
      "epoch": 0.5605956864087641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3311,
      "step": 16375
    },
    {
      "epoch": 0.5606299212598426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3295,
      "step": 16376
    },
    {
      "epoch": 0.5606641561109209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3196,
      "step": 16377
    },
    {
      "epoch": 0.5606983909619994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3259,
      "step": 16378
    },
    {
      "epoch": 0.5607326258130777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3223,
      "step": 16379
    },
    {
      "epoch": 0.5607668606641562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3391,
      "step": 16380
    },
    {
      "epoch": 0.5608010955152345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3348,
      "step": 16381
    },
    {
      "epoch": 0.560835330366313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3134,
      "step": 16382
    },
    {
      "epoch": 0.5608695652173913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3282,
      "step": 16383
    },
    {
      "epoch": 0.5609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3058,
      "step": 16384
    },
    {
      "epoch": 0.5609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3124,
      "step": 16385
    },
    {
      "epoch": 0.5609722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3373,
      "step": 16386
    },
    {
      "epoch": 0.5610065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3268,
      "step": 16387
    },
    {
      "epoch": 0.5610407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.299,
      "step": 16388
    },
    {
      "epoch": 0.5610749743238617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3213,
      "step": 16389
    },
    {
      "epoch": 0.5611092091749401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3224,
      "step": 16390
    },
    {
      "epoch": 0.5611434440260185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3235,
      "step": 16391
    },
    {
      "epoch": 0.5611776788770969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.338,
      "step": 16392
    },
    {
      "epoch": 0.5612119137281752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.319,
      "step": 16393
    },
    {
      "epoch": 0.5612461485792537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3171,
      "step": 16394
    },
    {
      "epoch": 0.561280383430332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3132,
      "step": 16395
    },
    {
      "epoch": 0.5613146182814105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3135,
      "step": 16396
    },
    {
      "epoch": 0.5613488531324888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3304,
      "step": 16397
    },
    {
      "epoch": 0.5613830879835673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3237,
      "step": 16398
    },
    {
      "epoch": 0.5614173228346456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3133,
      "step": 16399
    },
    {
      "epoch": 0.5614515576857241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3243,
      "step": 16400
    },
    {
      "epoch": 0.5614857925368024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3503,
      "step": 16401
    },
    {
      "epoch": 0.5615200273878809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3406,
      "step": 16402
    },
    {
      "epoch": 0.5615542622389592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3398,
      "step": 16403
    },
    {
      "epoch": 0.5615884970900377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3251,
      "step": 16404
    },
    {
      "epoch": 0.561622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3079,
      "step": 16405
    },
    {
      "epoch": 0.5616569667921945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3164,
      "step": 16406
    },
    {
      "epoch": 0.5616912016432728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3101,
      "step": 16407
    },
    {
      "epoch": 0.5617254364943512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3127,
      "step": 16408
    },
    {
      "epoch": 0.5617596713454297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3063,
      "step": 16409
    },
    {
      "epoch": 0.561793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3239,
      "step": 16410
    },
    {
      "epoch": 0.5618281410475865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3436,
      "step": 16411
    },
    {
      "epoch": 0.5618623758986648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3273,
      "step": 16412
    },
    {
      "epoch": 0.5618966107497433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3296,
      "step": 16413
    },
    {
      "epoch": 0.5619308456008216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3395,
      "step": 16414
    },
    {
      "epoch": 0.5619650804519001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3252,
      "step": 16415
    },
    {
      "epoch": 0.5619993153029784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3208,
      "step": 16416
    },
    {
      "epoch": 0.5620335501540569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3148,
      "step": 16417
    },
    {
      "epoch": 0.5620677850051352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3029,
      "step": 16418
    },
    {
      "epoch": 0.5621020198562137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3289,
      "step": 16419
    },
    {
      "epoch": 0.562136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3243,
      "step": 16420
    },
    {
      "epoch": 0.5621704895583705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3245,
      "step": 16421
    },
    {
      "epoch": 0.5622047244094488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3182,
      "step": 16422
    },
    {
      "epoch": 0.5622389592605272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3223,
      "step": 16423
    },
    {
      "epoch": 0.5622731941116056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3035,
      "step": 16424
    },
    {
      "epoch": 0.562307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3174,
      "step": 16425
    },
    {
      "epoch": 0.5623416638137624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3221,
      "step": 16426
    },
    {
      "epoch": 0.5623758986648408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3185,
      "step": 16427
    },
    {
      "epoch": 0.5624101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3198,
      "step": 16428
    },
    {
      "epoch": 0.5624443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3135,
      "step": 16429
    },
    {
      "epoch": 0.562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.333,
      "step": 16430
    },
    {
      "epoch": 0.5625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3253,
      "step": 16431
    },
    {
      "epoch": 0.5625470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3434,
      "step": 16432
    },
    {
      "epoch": 0.5625813077713112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3047,
      "step": 16433
    },
    {
      "epoch": 0.5626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3185,
      "step": 16434
    },
    {
      "epoch": 0.562649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3321,
      "step": 16435
    },
    {
      "epoch": 0.5626840123245463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3231,
      "step": 16436
    },
    {
      "epoch": 0.5627182471756248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3245,
      "step": 16437
    },
    {
      "epoch": 0.5627524820267031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3474,
      "step": 16438
    },
    {
      "epoch": 0.5627867168777816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3418,
      "step": 16439
    },
    {
      "epoch": 0.5628209517288599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3117,
      "step": 16440
    },
    {
      "epoch": 0.5628551865799384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3242,
      "step": 16441
    },
    {
      "epoch": 0.5628894214310167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3206,
      "step": 16442
    },
    {
      "epoch": 0.5629236562820952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3451,
      "step": 16443
    },
    {
      "epoch": 0.5629578911331735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.321,
      "step": 16444
    },
    {
      "epoch": 0.562992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3116,
      "step": 16445
    },
    {
      "epoch": 0.5630263608353304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3381,
      "step": 16446
    },
    {
      "epoch": 0.5630605956864088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3262,
      "step": 16447
    },
    {
      "epoch": 0.5630948305374872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3351,
      "step": 16448
    },
    {
      "epoch": 0.5631290653885656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3136,
      "step": 16449
    },
    {
      "epoch": 0.563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3131,
      "step": 16450
    },
    {
      "epoch": 0.5631975350907223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3256,
      "step": 16451
    },
    {
      "epoch": 0.5632317699418008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3124,
      "step": 16452
    },
    {
      "epoch": 0.5632660047928791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3233,
      "step": 16453
    },
    {
      "epoch": 0.5633002396439576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3308,
      "step": 16454
    },
    {
      "epoch": 0.5633344744950359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3333,
      "step": 16455
    },
    {
      "epoch": 0.5633687093461144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3217,
      "step": 16456
    },
    {
      "epoch": 0.5634029441971927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3414,
      "step": 16457
    },
    {
      "epoch": 0.5634371790482712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3264,
      "step": 16458
    },
    {
      "epoch": 0.5634714138993495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3205,
      "step": 16459
    },
    {
      "epoch": 0.563505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3247,
      "step": 16460
    },
    {
      "epoch": 0.5635398836015063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3002,
      "step": 16461
    },
    {
      "epoch": 0.5635741184525848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3231,
      "step": 16462
    },
    {
      "epoch": 0.5636083533036631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3296,
      "step": 16463
    },
    {
      "epoch": 0.5636425881547416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3224,
      "step": 16464
    },
    {
      "epoch": 0.5636768230058199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3282,
      "step": 16465
    },
    {
      "epoch": 0.5637110578568983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3096,
      "step": 16466
    },
    {
      "epoch": 0.5637452927079767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3168,
      "step": 16467
    },
    {
      "epoch": 0.5637795275590551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3038,
      "step": 16468
    },
    {
      "epoch": 0.5638137624101335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3294,
      "step": 16469
    },
    {
      "epoch": 0.5638479972612119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3342,
      "step": 16470
    },
    {
      "epoch": 0.5638822321122903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3217,
      "step": 16471
    },
    {
      "epoch": 0.5639164669633687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3338,
      "step": 16472
    },
    {
      "epoch": 0.5639507018144471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3045,
      "step": 16473
    },
    {
      "epoch": 0.5639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3121,
      "step": 16474
    },
    {
      "epoch": 0.5640191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3303,
      "step": 16475
    },
    {
      "epoch": 0.5640534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3291,
      "step": 16476
    },
    {
      "epoch": 0.5640876412187606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3337,
      "step": 16477
    },
    {
      "epoch": 0.5641218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3312,
      "step": 16478
    },
    {
      "epoch": 0.5641561109209174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.317,
      "step": 16479
    },
    {
      "epoch": 0.5641903457719959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3146,
      "step": 16480
    },
    {
      "epoch": 0.5642245806230742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3298,
      "step": 16481
    },
    {
      "epoch": 0.5642588154741527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3373,
      "step": 16482
    },
    {
      "epoch": 0.5642930503252311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3176,
      "step": 16483
    },
    {
      "epoch": 0.5643272851763095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3204,
      "step": 16484
    },
    {
      "epoch": 0.5643615200273879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3196,
      "step": 16485
    },
    {
      "epoch": 0.5643957548784663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3257,
      "step": 16486
    },
    {
      "epoch": 0.5644299897295447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3096,
      "step": 16487
    },
    {
      "epoch": 0.5644642245806231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3164,
      "step": 16488
    },
    {
      "epoch": 0.5644984594317015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3322,
      "step": 16489
    },
    {
      "epoch": 0.5645326942827799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.316,
      "step": 16490
    },
    {
      "epoch": 0.5645669291338583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3328,
      "step": 16491
    },
    {
      "epoch": 0.5646011639849366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3307,
      "step": 16492
    },
    {
      "epoch": 0.5646353988360151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3137,
      "step": 16493
    },
    {
      "epoch": 0.5646696336870934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3018,
      "step": 16494
    },
    {
      "epoch": 0.5647038685381719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3227,
      "step": 16495
    },
    {
      "epoch": 0.5647381033892502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3138,
      "step": 16496
    },
    {
      "epoch": 0.5647723382403287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.334,
      "step": 16497
    },
    {
      "epoch": 0.564806573091407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3151,
      "step": 16498
    },
    {
      "epoch": 0.5648408079424855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3126,
      "step": 16499
    },
    {
      "epoch": 0.5648750427935638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3323,
      "step": 16500
    },
    {
      "epoch": 0.5649092776446423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3172,
      "step": 16501
    },
    {
      "epoch": 0.5649435124957206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.323,
      "step": 16502
    },
    {
      "epoch": 0.5649777473467991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3197,
      "step": 16503
    },
    {
      "epoch": 0.5650119821978774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3193,
      "step": 16504
    },
    {
      "epoch": 0.5650462170489559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3149,
      "step": 16505
    },
    {
      "epoch": 0.5650804519000342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3312,
      "step": 16506
    },
    {
      "epoch": 0.5651146867511126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3125,
      "step": 16507
    },
    {
      "epoch": 0.565148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3224,
      "step": 16508
    },
    {
      "epoch": 0.5651831564532694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3088,
      "step": 16509
    },
    {
      "epoch": 0.5652173913043478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3004,
      "step": 16510
    },
    {
      "epoch": 0.5652516261554262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3286,
      "step": 16511
    },
    {
      "epoch": 0.5652858610065046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3391,
      "step": 16512
    },
    {
      "epoch": 0.565320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3186,
      "step": 16513
    },
    {
      "epoch": 0.5653543307086614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3114,
      "step": 16514
    },
    {
      "epoch": 0.5653885655597398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3196,
      "step": 16515
    },
    {
      "epoch": 0.5654228004108182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3241,
      "step": 16516
    },
    {
      "epoch": 0.5654570352618966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3333,
      "step": 16517
    },
    {
      "epoch": 0.5654912701129751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3297,
      "step": 16518
    },
    {
      "epoch": 0.5655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3172,
      "step": 16519
    },
    {
      "epoch": 0.5655597398151319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3173,
      "step": 16520
    },
    {
      "epoch": 0.5655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3181,
      "step": 16521
    },
    {
      "epoch": 0.5656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3192,
      "step": 16522
    },
    {
      "epoch": 0.565662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3159,
      "step": 16523
    },
    {
      "epoch": 0.5656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3221,
      "step": 16524
    },
    {
      "epoch": 0.5657309140705238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3281,
      "step": 16525
    },
    {
      "epoch": 0.5657651489216022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3453,
      "step": 16526
    },
    {
      "epoch": 0.5657993837726806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3102,
      "step": 16527
    },
    {
      "epoch": 0.565833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3315,
      "step": 16528
    },
    {
      "epoch": 0.5658678534748374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3256,
      "step": 16529
    },
    {
      "epoch": 0.5659020883259158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.322,
      "step": 16530
    },
    {
      "epoch": 0.5659363231769942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.321,
      "step": 16531
    },
    {
      "epoch": 0.5659705580280726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3317,
      "step": 16532
    },
    {
      "epoch": 0.566004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.318,
      "step": 16533
    },
    {
      "epoch": 0.5660390277302294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.311,
      "step": 16534
    },
    {
      "epoch": 0.5660732625813077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3334,
      "step": 16535
    },
    {
      "epoch": 0.5661074974323862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.332,
      "step": 16536
    },
    {
      "epoch": 0.5661417322834645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3174,
      "step": 16537
    },
    {
      "epoch": 0.566175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.312,
      "step": 16538
    },
    {
      "epoch": 0.5662102019856213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3366,
      "step": 16539
    },
    {
      "epoch": 0.5662444368366998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.332,
      "step": 16540
    },
    {
      "epoch": 0.5662786716877781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3196,
      "step": 16541
    },
    {
      "epoch": 0.5663129065388566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2999,
      "step": 16542
    },
    {
      "epoch": 0.5663471413899349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3122,
      "step": 16543
    },
    {
      "epoch": 0.5663813762410134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3346,
      "step": 16544
    },
    {
      "epoch": 0.5664156110920917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3547,
      "step": 16545
    },
    {
      "epoch": 0.5664498459431702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3049,
      "step": 16546
    },
    {
      "epoch": 0.5664840807942485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3237,
      "step": 16547
    },
    {
      "epoch": 0.566518315645327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3174,
      "step": 16548
    },
    {
      "epoch": 0.5665525504964053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3214,
      "step": 16549
    },
    {
      "epoch": 0.5665867853474837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3256,
      "step": 16550
    },
    {
      "epoch": 0.5666210201985621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3176,
      "step": 16551
    },
    {
      "epoch": 0.5666552550496405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3364,
      "step": 16552
    },
    {
      "epoch": 0.5666894899007189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3422,
      "step": 16553
    },
    {
      "epoch": 0.5667237247517973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3444,
      "step": 16554
    },
    {
      "epoch": 0.5667579596028758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3215,
      "step": 16555
    },
    {
      "epoch": 0.5667921944539541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3239,
      "step": 16556
    },
    {
      "epoch": 0.5668264293050326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3249,
      "step": 16557
    },
    {
      "epoch": 0.5668606641561109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3206,
      "step": 16558
    },
    {
      "epoch": 0.5668948990071894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.319,
      "step": 16559
    },
    {
      "epoch": 0.5669291338582677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3253,
      "step": 16560
    },
    {
      "epoch": 0.5669633687093462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3255,
      "step": 16561
    },
    {
      "epoch": 0.5669976035604245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2945,
      "step": 16562
    },
    {
      "epoch": 0.567031838411503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3153,
      "step": 16563
    },
    {
      "epoch": 0.5670660732625813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.308,
      "step": 16564
    },
    {
      "epoch": 0.5671003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3167,
      "step": 16565
    },
    {
      "epoch": 0.5671345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3038,
      "step": 16566
    },
    {
      "epoch": 0.5671687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3267,
      "step": 16567
    },
    {
      "epoch": 0.5672030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3322,
      "step": 16568
    },
    {
      "epoch": 0.5672372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3363,
      "step": 16569
    },
    {
      "epoch": 0.5672714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3251,
      "step": 16570
    },
    {
      "epoch": 0.5673057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3328,
      "step": 16571
    },
    {
      "epoch": 0.5673399520712085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3173,
      "step": 16572
    },
    {
      "epoch": 0.5673741869222869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3006,
      "step": 16573
    },
    {
      "epoch": 0.5674084217733653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3338,
      "step": 16574
    },
    {
      "epoch": 0.5674426566244437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3353,
      "step": 16575
    },
    {
      "epoch": 0.567476891475522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3308,
      "step": 16576
    },
    {
      "epoch": 0.5675111263266005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3128,
      "step": 16577
    },
    {
      "epoch": 0.5675453611776788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3213,
      "step": 16578
    },
    {
      "epoch": 0.5675795960287573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3223,
      "step": 16579
    },
    {
      "epoch": 0.5676138308798356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3278,
      "step": 16580
    },
    {
      "epoch": 0.5676480657309141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3168,
      "step": 16581
    },
    {
      "epoch": 0.5676823005819924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3325,
      "step": 16582
    },
    {
      "epoch": 0.5677165354330709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3158,
      "step": 16583
    },
    {
      "epoch": 0.5677507702841492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3131,
      "step": 16584
    },
    {
      "epoch": 0.5677850051352277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.313,
      "step": 16585
    },
    {
      "epoch": 0.567819239986306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3319,
      "step": 16586
    },
    {
      "epoch": 0.5678534748373845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3071,
      "step": 16587
    },
    {
      "epoch": 0.5678877096884628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3268,
      "step": 16588
    },
    {
      "epoch": 0.5679219445395413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2978,
      "step": 16589
    },
    {
      "epoch": 0.5679561793906196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2982,
      "step": 16590
    },
    {
      "epoch": 0.567990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3289,
      "step": 16591
    },
    {
      "epoch": 0.5680246490927765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3101,
      "step": 16592
    },
    {
      "epoch": 0.5680588839438548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3149,
      "step": 16593
    },
    {
      "epoch": 0.5680931187949333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3249,
      "step": 16594
    },
    {
      "epoch": 0.5681273536460116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3134,
      "step": 16595
    },
    {
      "epoch": 0.5681615884970901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3087,
      "step": 16596
    },
    {
      "epoch": 0.5681958233481684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3171,
      "step": 16597
    },
    {
      "epoch": 0.5682300581992469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3052,
      "step": 16598
    },
    {
      "epoch": 0.5682642930503252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3196,
      "step": 16599
    },
    {
      "epoch": 0.5682985279014037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3217,
      "step": 16600
    },
    {
      "epoch": 0.568332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3066,
      "step": 16601
    },
    {
      "epoch": 0.5683669976035605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3098,
      "step": 16602
    },
    {
      "epoch": 0.5684012324546388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3199,
      "step": 16603
    },
    {
      "epoch": 0.5684354673057173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3251,
      "step": 16604
    },
    {
      "epoch": 0.5684697021567956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3357,
      "step": 16605
    },
    {
      "epoch": 0.568503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.318,
      "step": 16606
    },
    {
      "epoch": 0.5685381718589524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3285,
      "step": 16607
    },
    {
      "epoch": 0.5685724067100308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3242,
      "step": 16608
    },
    {
      "epoch": 0.5686066415611092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3164,
      "step": 16609
    },
    {
      "epoch": 0.5686408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3271,
      "step": 16610
    },
    {
      "epoch": 0.568675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3289,
      "step": 16611
    },
    {
      "epoch": 0.5687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3198,
      "step": 16612
    },
    {
      "epoch": 0.5687435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3006,
      "step": 16613
    },
    {
      "epoch": 0.5687778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3192,
      "step": 16614
    },
    {
      "epoch": 0.5688120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3085,
      "step": 16615
    },
    {
      "epoch": 0.568846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3224,
      "step": 16616
    },
    {
      "epoch": 0.5688805203697364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2873,
      "step": 16617
    },
    {
      "epoch": 0.5689147552208148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3062,
      "step": 16618
    },
    {
      "epoch": 0.5689489900718931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3357,
      "step": 16619
    },
    {
      "epoch": 0.5689832249229716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2958,
      "step": 16620
    },
    {
      "epoch": 0.5690174597740499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3065,
      "step": 16621
    },
    {
      "epoch": 0.5690516946251284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3182,
      "step": 16622
    },
    {
      "epoch": 0.5690859294762067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3213,
      "step": 16623
    },
    {
      "epoch": 0.5691201643272852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3134,
      "step": 16624
    },
    {
      "epoch": 0.5691543991783635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3223,
      "step": 16625
    },
    {
      "epoch": 0.569188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3185,
      "step": 16626
    },
    {
      "epoch": 0.5692228688805203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3217,
      "step": 16627
    },
    {
      "epoch": 0.5692571037315988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3019,
      "step": 16628
    },
    {
      "epoch": 0.5692913385826772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3171,
      "step": 16629
    },
    {
      "epoch": 0.5693255734337556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3304,
      "step": 16630
    },
    {
      "epoch": 0.569359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3081,
      "step": 16631
    },
    {
      "epoch": 0.5693940431359124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3165,
      "step": 16632
    },
    {
      "epoch": 0.5694282779869908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3441,
      "step": 16633
    },
    {
      "epoch": 0.5694625128380691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3111,
      "step": 16634
    },
    {
      "epoch": 0.5694967476891476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3148,
      "step": 16635
    },
    {
      "epoch": 0.5695309825402259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3213,
      "step": 16636
    },
    {
      "epoch": 0.5695652173913044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3256,
      "step": 16637
    },
    {
      "epoch": 0.5695994522423827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3345,
      "step": 16638
    },
    {
      "epoch": 0.5696336870934612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3319,
      "step": 16639
    },
    {
      "epoch": 0.5696679219445395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3237,
      "step": 16640
    },
    {
      "epoch": 0.569702156795618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.311,
      "step": 16641
    },
    {
      "epoch": 0.5697363916466963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3199,
      "step": 16642
    },
    {
      "epoch": 0.5697706264977748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3078,
      "step": 16643
    },
    {
      "epoch": 0.5698048613488531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3249,
      "step": 16644
    },
    {
      "epoch": 0.5698390961999316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3239,
      "step": 16645
    },
    {
      "epoch": 0.5698733310510099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3156,
      "step": 16646
    },
    {
      "epoch": 0.5699075659020884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3084,
      "step": 16647
    },
    {
      "epoch": 0.5699418007531667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3123,
      "step": 16648
    },
    {
      "epoch": 0.5699760356042451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3235,
      "step": 16649
    },
    {
      "epoch": 0.5700102704553235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.326,
      "step": 16650
    },
    {
      "epoch": 0.5700445053064019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3113,
      "step": 16651
    },
    {
      "epoch": 0.5700787401574803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3118,
      "step": 16652
    },
    {
      "epoch": 0.5701129750085587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3167,
      "step": 16653
    },
    {
      "epoch": 0.5701472098596371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3026,
      "step": 16654
    },
    {
      "epoch": 0.5701814447107155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.322,
      "step": 16655
    },
    {
      "epoch": 0.5702156795617939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3202,
      "step": 16656
    },
    {
      "epoch": 0.5702499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3105,
      "step": 16657
    },
    {
      "epoch": 0.5702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.299,
      "step": 16658
    },
    {
      "epoch": 0.5703183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3243,
      "step": 16659
    },
    {
      "epoch": 0.5703526189661074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3086,
      "step": 16660
    },
    {
      "epoch": 0.5703868538171859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3257,
      "step": 16661
    },
    {
      "epoch": 0.5704210886682642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3235,
      "step": 16662
    },
    {
      "epoch": 0.5704553235193427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3157,
      "step": 16663
    },
    {
      "epoch": 0.570489558370421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3108,
      "step": 16664
    },
    {
      "epoch": 0.5705237932214995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.299,
      "step": 16665
    },
    {
      "epoch": 0.5705580280725779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3221,
      "step": 16666
    },
    {
      "epoch": 0.5705922629236563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3022,
      "step": 16667
    },
    {
      "epoch": 0.5706264977747347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3038,
      "step": 16668
    },
    {
      "epoch": 0.5706607326258131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3292,
      "step": 16669
    },
    {
      "epoch": 0.5706949674768915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3276,
      "step": 16670
    },
    {
      "epoch": 0.5707292023279699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3225,
      "step": 16671
    },
    {
      "epoch": 0.5707634371790483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3061,
      "step": 16672
    },
    {
      "epoch": 0.5707976720301267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3125,
      "step": 16673
    },
    {
      "epoch": 0.5708319068812051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.314,
      "step": 16674
    },
    {
      "epoch": 0.5708661417322834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.332,
      "step": 16675
    },
    {
      "epoch": 0.5709003765833619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3222,
      "step": 16676
    },
    {
      "epoch": 0.5709346114344402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3123,
      "step": 16677
    },
    {
      "epoch": 0.5709688462855187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3015,
      "step": 16678
    },
    {
      "epoch": 0.571003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3186,
      "step": 16679
    },
    {
      "epoch": 0.5710373159876755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3314,
      "step": 16680
    },
    {
      "epoch": 0.5710715508387538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3298,
      "step": 16681
    },
    {
      "epoch": 0.5711057856898323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3041,
      "step": 16682
    },
    {
      "epoch": 0.5711400205409106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3151,
      "step": 16683
    },
    {
      "epoch": 0.5711742553919891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3102,
      "step": 16684
    },
    {
      "epoch": 0.5712084902430674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3171,
      "step": 16685
    },
    {
      "epoch": 0.5712427250941459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3147,
      "step": 16686
    },
    {
      "epoch": 0.5712769599452242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3163,
      "step": 16687
    },
    {
      "epoch": 0.5713111947963027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3275,
      "step": 16688
    },
    {
      "epoch": 0.571345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3187,
      "step": 16689
    },
    {
      "epoch": 0.5713796644984594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.32,
      "step": 16690
    },
    {
      "epoch": 0.5714138993495378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3243,
      "step": 16691
    },
    {
      "epoch": 0.5714481342006162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3138,
      "step": 16692
    },
    {
      "epoch": 0.5714823690516946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.332,
      "step": 16693
    },
    {
      "epoch": 0.571516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3086,
      "step": 16694
    },
    {
      "epoch": 0.5715508387538514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3243,
      "step": 16695
    },
    {
      "epoch": 0.5715850736049298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3278,
      "step": 16696
    },
    {
      "epoch": 0.5716193084560082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3088,
      "step": 16697
    },
    {
      "epoch": 0.5716535433070866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3377,
      "step": 16698
    },
    {
      "epoch": 0.571687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3187,
      "step": 16699
    },
    {
      "epoch": 0.5717220130092434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2969,
      "step": 16700
    },
    {
      "epoch": 0.5717562478603218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3148,
      "step": 16701
    },
    {
      "epoch": 0.5717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3214,
      "step": 16702
    },
    {
      "epoch": 0.5718247175624787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3147,
      "step": 16703
    },
    {
      "epoch": 0.571858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3145,
      "step": 16704
    },
    {
      "epoch": 0.5718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3134,
      "step": 16705
    },
    {
      "epoch": 0.5719274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.323,
      "step": 16706
    },
    {
      "epoch": 0.5719616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2961,
      "step": 16707
    },
    {
      "epoch": 0.5719958918178706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3273,
      "step": 16708
    },
    {
      "epoch": 0.572030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3052,
      "step": 16709
    },
    {
      "epoch": 0.5720643615200274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3055,
      "step": 16710
    },
    {
      "epoch": 0.5720985963711058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3001,
      "step": 16711
    },
    {
      "epoch": 0.5721328312221842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3088,
      "step": 16712
    },
    {
      "epoch": 0.5721670660732626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3058,
      "step": 16713
    },
    {
      "epoch": 0.572201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3064,
      "step": 16714
    },
    {
      "epoch": 0.5722355357754194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3268,
      "step": 16715
    },
    {
      "epoch": 0.5722697706264978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.302,
      "step": 16716
    },
    {
      "epoch": 0.5723040054775762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3144,
      "step": 16717
    },
    {
      "epoch": 0.5723382403286545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3086,
      "step": 16718
    },
    {
      "epoch": 0.572372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.331,
      "step": 16719
    },
    {
      "epoch": 0.5724067100308113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3118,
      "step": 16720
    },
    {
      "epoch": 0.5724409448818898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3207,
      "step": 16721
    },
    {
      "epoch": 0.5724751797329681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3184,
      "step": 16722
    },
    {
      "epoch": 0.5725094145840466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3096,
      "step": 16723
    },
    {
      "epoch": 0.5725436494351249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3055,
      "step": 16724
    },
    {
      "epoch": 0.5725778842862034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3101,
      "step": 16725
    },
    {
      "epoch": 0.5726121191372817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3268,
      "step": 16726
    },
    {
      "epoch": 0.5726463539883602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3177,
      "step": 16727
    },
    {
      "epoch": 0.5726805888394385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3376,
      "step": 16728
    },
    {
      "epoch": 0.572714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3159,
      "step": 16729
    },
    {
      "epoch": 0.5727490585415953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3022,
      "step": 16730
    },
    {
      "epoch": 0.5727832933926738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3328,
      "step": 16731
    },
    {
      "epoch": 0.5728175282437521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3129,
      "step": 16732
    },
    {
      "epoch": 0.5728517630948305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3231,
      "step": 16733
    },
    {
      "epoch": 0.5728859979459089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3274,
      "step": 16734
    },
    {
      "epoch": 0.5729202327969873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3145,
      "step": 16735
    },
    {
      "epoch": 0.5729544676480657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3115,
      "step": 16736
    },
    {
      "epoch": 0.5729887024991441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3178,
      "step": 16737
    },
    {
      "epoch": 0.5730229373502225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3234,
      "step": 16738
    },
    {
      "epoch": 0.5730571722013009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3049,
      "step": 16739
    },
    {
      "epoch": 0.5730914070523794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3173,
      "step": 16740
    },
    {
      "epoch": 0.5731256419034577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.312,
      "step": 16741
    },
    {
      "epoch": 0.5731598767545362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2967,
      "step": 16742
    },
    {
      "epoch": 0.5731941116056145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.334,
      "step": 16743
    },
    {
      "epoch": 0.573228346456693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3176,
      "step": 16744
    },
    {
      "epoch": 0.5732625813077713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3132,
      "step": 16745
    },
    {
      "epoch": 0.5732968161588498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3167,
      "step": 16746
    },
    {
      "epoch": 0.5733310510099281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3085,
      "step": 16747
    },
    {
      "epoch": 0.5733652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.319,
      "step": 16748
    },
    {
      "epoch": 0.5733995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.313,
      "step": 16749
    },
    {
      "epoch": 0.5734337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3088,
      "step": 16750
    },
    {
      "epoch": 0.5734679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3098,
      "step": 16751
    },
    {
      "epoch": 0.5735022252653201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3252,
      "step": 16752
    },
    {
      "epoch": 0.5735364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.314,
      "step": 16753
    },
    {
      "epoch": 0.5735706949674769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3172,
      "step": 16754
    },
    {
      "epoch": 0.5736049298185553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3068,
      "step": 16755
    },
    {
      "epoch": 0.5736391646696337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3215,
      "step": 16756
    },
    {
      "epoch": 0.5736733995207121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3062,
      "step": 16757
    },
    {
      "epoch": 0.5737076343717905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3244,
      "step": 16758
    },
    {
      "epoch": 0.5737418692228688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3349,
      "step": 16759
    },
    {
      "epoch": 0.5737761040739473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3127,
      "step": 16760
    },
    {
      "epoch": 0.5738103389250256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3134,
      "step": 16761
    },
    {
      "epoch": 0.5738445737761041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3179,
      "step": 16762
    },
    {
      "epoch": 0.5738788086271824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3094,
      "step": 16763
    },
    {
      "epoch": 0.5739130434782609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3015,
      "step": 16764
    },
    {
      "epoch": 0.5739472783293392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3193,
      "step": 16765
    },
    {
      "epoch": 0.5739815131804177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3217,
      "step": 16766
    },
    {
      "epoch": 0.574015748031496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3084,
      "step": 16767
    },
    {
      "epoch": 0.5740499828825745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3251,
      "step": 16768
    },
    {
      "epoch": 0.5740842177336528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.317,
      "step": 16769
    },
    {
      "epoch": 0.5741184525847313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2905,
      "step": 16770
    },
    {
      "epoch": 0.5741526874358096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3191,
      "step": 16771
    },
    {
      "epoch": 0.5741869222868881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3066,
      "step": 16772
    },
    {
      "epoch": 0.5742211571379664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3152,
      "step": 16773
    },
    {
      "epoch": 0.5742553919890448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3314,
      "step": 16774
    },
    {
      "epoch": 0.5742896268401232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3038,
      "step": 16775
    },
    {
      "epoch": 0.5743238616912016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3203,
      "step": 16776
    },
    {
      "epoch": 0.5743580965422801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3011,
      "step": 16777
    },
    {
      "epoch": 0.5743923313933584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3193,
      "step": 16778
    },
    {
      "epoch": 0.5744265662444369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3228,
      "step": 16779
    },
    {
      "epoch": 0.5744608010955152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3297,
      "step": 16780
    },
    {
      "epoch": 0.5744950359465937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3074,
      "step": 16781
    },
    {
      "epoch": 0.574529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3397,
      "step": 16782
    },
    {
      "epoch": 0.5745635056487505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3174,
      "step": 16783
    },
    {
      "epoch": 0.5745977404998288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3141,
      "step": 16784
    },
    {
      "epoch": 0.5746319753509073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3029,
      "step": 16785
    },
    {
      "epoch": 0.5746662102019856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.319,
      "step": 16786
    },
    {
      "epoch": 0.5747004450530641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3091,
      "step": 16787
    },
    {
      "epoch": 0.5747346799041424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3198,
      "step": 16788
    },
    {
      "epoch": 0.5747689147552208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3108,
      "step": 16789
    },
    {
      "epoch": 0.5748031496062992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3162,
      "step": 16790
    },
    {
      "epoch": 0.5748373844573776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3242,
      "step": 16791
    },
    {
      "epoch": 0.574871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2895,
      "step": 16792
    },
    {
      "epoch": 0.5749058541595344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3183,
      "step": 16793
    },
    {
      "epoch": 0.5749400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2928,
      "step": 16794
    },
    {
      "epoch": 0.5749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.318,
      "step": 16795
    },
    {
      "epoch": 0.5750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3087,
      "step": 16796
    },
    {
      "epoch": 0.575042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2981,
      "step": 16797
    },
    {
      "epoch": 0.5750770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3303,
      "step": 16798
    },
    {
      "epoch": 0.5751112632660048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3145,
      "step": 16799
    },
    {
      "epoch": 0.5751454981170832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3181,
      "step": 16800
    },
    {
      "epoch": 0.5751797329681616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3044,
      "step": 16801
    },
    {
      "epoch": 0.57521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3079,
      "step": 16802
    },
    {
      "epoch": 0.5752482026703184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2996,
      "step": 16803
    },
    {
      "epoch": 0.5752824375213967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3085,
      "step": 16804
    },
    {
      "epoch": 0.5753166723724752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3037,
      "step": 16805
    },
    {
      "epoch": 0.5753509072235535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3223,
      "step": 16806
    },
    {
      "epoch": 0.575385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3401,
      "step": 16807
    },
    {
      "epoch": 0.5754193769257103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3059,
      "step": 16808
    },
    {
      "epoch": 0.5754536117767888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3375,
      "step": 16809
    },
    {
      "epoch": 0.5754878466278671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3339,
      "step": 16810
    },
    {
      "epoch": 0.5755220814789456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3271,
      "step": 16811
    },
    {
      "epoch": 0.5755563163300239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3132,
      "step": 16812
    },
    {
      "epoch": 0.5755905511811024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2999,
      "step": 16813
    },
    {
      "epoch": 0.5756247860321808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3203,
      "step": 16814
    },
    {
      "epoch": 0.5756590208832592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3011,
      "step": 16815
    },
    {
      "epoch": 0.5756932557343376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3237,
      "step": 16816
    },
    {
      "epoch": 0.575727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3208,
      "step": 16817
    },
    {
      "epoch": 0.5757617254364944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3091,
      "step": 16818
    },
    {
      "epoch": 0.5757959602875727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2982,
      "step": 16819
    },
    {
      "epoch": 0.5758301951386512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3124,
      "step": 16820
    },
    {
      "epoch": 0.5758644299897295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3142,
      "step": 16821
    },
    {
      "epoch": 0.575898664840808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.325,
      "step": 16822
    },
    {
      "epoch": 0.5759328996918863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3107,
      "step": 16823
    },
    {
      "epoch": 0.5759671345429648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3241,
      "step": 16824
    },
    {
      "epoch": 0.5760013693940431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3207,
      "step": 16825
    },
    {
      "epoch": 0.5760356042451216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3073,
      "step": 16826
    },
    {
      "epoch": 0.5760698390961999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3113,
      "step": 16827
    },
    {
      "epoch": 0.5761040739472784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3043,
      "step": 16828
    },
    {
      "epoch": 0.5761383087983567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3266,
      "step": 16829
    },
    {
      "epoch": 0.5761725436494352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.308,
      "step": 16830
    },
    {
      "epoch": 0.5762067785005135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3095,
      "step": 16831
    },
    {
      "epoch": 0.576241013351592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3079,
      "step": 16832
    },
    {
      "epoch": 0.5762752482026703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3026,
      "step": 16833
    },
    {
      "epoch": 0.5763094830537487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2987,
      "step": 16834
    },
    {
      "epoch": 0.5763437179048271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2994,
      "step": 16835
    },
    {
      "epoch": 0.5763779527559055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3033,
      "step": 16836
    },
    {
      "epoch": 0.5764121876069839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3196,
      "step": 16837
    },
    {
      "epoch": 0.5764464224580623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3014,
      "step": 16838
    },
    {
      "epoch": 0.5764806573091407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3143,
      "step": 16839
    },
    {
      "epoch": 0.5765148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3212,
      "step": 16840
    },
    {
      "epoch": 0.5765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3128,
      "step": 16841
    },
    {
      "epoch": 0.5765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3157,
      "step": 16842
    },
    {
      "epoch": 0.5766175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3131,
      "step": 16843
    },
    {
      "epoch": 0.5766518315645327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3317,
      "step": 16844
    },
    {
      "epoch": 0.576686066415611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3073,
      "step": 16845
    },
    {
      "epoch": 0.5767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3136,
      "step": 16846
    },
    {
      "epoch": 0.5767545361177678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3101,
      "step": 16847
    },
    {
      "epoch": 0.5767887709688463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3137,
      "step": 16848
    },
    {
      "epoch": 0.5768230058199246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3175,
      "step": 16849
    },
    {
      "epoch": 0.5768572406710031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3091,
      "step": 16850
    },
    {
      "epoch": 0.5768914755220815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3074,
      "step": 16851
    },
    {
      "epoch": 0.5769257103731599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3248,
      "step": 16852
    },
    {
      "epoch": 0.5769599452242383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3356,
      "step": 16853
    },
    {
      "epoch": 0.5769941800753167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3104,
      "step": 16854
    },
    {
      "epoch": 0.5770284149263951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3035,
      "step": 16855
    },
    {
      "epoch": 0.5770626497774735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3095,
      "step": 16856
    },
    {
      "epoch": 0.5770968846285519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3152,
      "step": 16857
    },
    {
      "epoch": 0.5771311194796303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.329,
      "step": 16858
    },
    {
      "epoch": 0.5771653543307087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.308,
      "step": 16859
    },
    {
      "epoch": 0.577199589181787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3135,
      "step": 16860
    },
    {
      "epoch": 0.5772338240328655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3336,
      "step": 16861
    },
    {
      "epoch": 0.5772680588839438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3015,
      "step": 16862
    },
    {
      "epoch": 0.5773022937350223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3173,
      "step": 16863
    },
    {
      "epoch": 0.5773365285861006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3325,
      "step": 16864
    },
    {
      "epoch": 0.5773707634371791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3096,
      "step": 16865
    },
    {
      "epoch": 0.5774049982882574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3222,
      "step": 16866
    },
    {
      "epoch": 0.5774392331393359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.319,
      "step": 16867
    },
    {
      "epoch": 0.5774734679904142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3107,
      "step": 16868
    },
    {
      "epoch": 0.5775077028414927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3425,
      "step": 16869
    },
    {
      "epoch": 0.577541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3106,
      "step": 16870
    },
    {
      "epoch": 0.5775761725436495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2909,
      "step": 16871
    },
    {
      "epoch": 0.5776104073947278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2959,
      "step": 16872
    },
    {
      "epoch": 0.5776446422458063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3113,
      "step": 16873
    },
    {
      "epoch": 0.5776788770968846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3021,
      "step": 16874
    },
    {
      "epoch": 0.577713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.312,
      "step": 16875
    },
    {
      "epoch": 0.5777473467990414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3059,
      "step": 16876
    },
    {
      "epoch": 0.5777815816501198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3213,
      "step": 16877
    },
    {
      "epoch": 0.5778158165011982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2974,
      "step": 16878
    },
    {
      "epoch": 0.5778500513522766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3149,
      "step": 16879
    },
    {
      "epoch": 0.577884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2921,
      "step": 16880
    },
    {
      "epoch": 0.5779185210544334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3104,
      "step": 16881
    },
    {
      "epoch": 0.5779527559055118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3208,
      "step": 16882
    },
    {
      "epoch": 0.5779869907565902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3186,
      "step": 16883
    },
    {
      "epoch": 0.5780212256076686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3036,
      "step": 16884
    },
    {
      "epoch": 0.578055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3002,
      "step": 16885
    },
    {
      "epoch": 0.5780896953098255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3116,
      "step": 16886
    },
    {
      "epoch": 0.5781239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3254,
      "step": 16887
    },
    {
      "epoch": 0.5781581650119823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2962,
      "step": 16888
    },
    {
      "epoch": 0.5781923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2839,
      "step": 16889
    },
    {
      "epoch": 0.578226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3106,
      "step": 16890
    },
    {
      "epoch": 0.5782608695652174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.283,
      "step": 16891
    },
    {
      "epoch": 0.5782951044162958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3099,
      "step": 16892
    },
    {
      "epoch": 0.5783293392673742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3143,
      "step": 16893
    },
    {
      "epoch": 0.5783635741184526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.306,
      "step": 16894
    },
    {
      "epoch": 0.578397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2988,
      "step": 16895
    },
    {
      "epoch": 0.5784320438206094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.311,
      "step": 16896
    },
    {
      "epoch": 0.5784662786716878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.314,
      "step": 16897
    },
    {
      "epoch": 0.5785005135227662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3308,
      "step": 16898
    },
    {
      "epoch": 0.5785347483738446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.305,
      "step": 16899
    },
    {
      "epoch": 0.578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3028,
      "step": 16900
    },
    {
      "epoch": 0.5786032180760013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3054,
      "step": 16901
    },
    {
      "epoch": 0.5786374529270798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3074,
      "step": 16902
    },
    {
      "epoch": 0.5786716877781581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2933,
      "step": 16903
    },
    {
      "epoch": 0.5787059226292366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3136,
      "step": 16904
    },
    {
      "epoch": 0.5787401574803149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3135,
      "step": 16905
    },
    {
      "epoch": 0.5787743923313934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3125,
      "step": 16906
    },
    {
      "epoch": 0.5788086271824717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3012,
      "step": 16907
    },
    {
      "epoch": 0.5788428620335502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3348,
      "step": 16908
    },
    {
      "epoch": 0.5788770968846285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3198,
      "step": 16909
    },
    {
      "epoch": 0.578911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3175,
      "step": 16910
    },
    {
      "epoch": 0.5789455665867853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3144,
      "step": 16911
    },
    {
      "epoch": 0.5789798014378638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2921,
      "step": 16912
    },
    {
      "epoch": 0.5790140362889421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3107,
      "step": 16913
    },
    {
      "epoch": 0.5790482711400206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3012,
      "step": 16914
    },
    {
      "epoch": 0.5790825059910989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2894,
      "step": 16915
    },
    {
      "epoch": 0.5791167408421773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3187,
      "step": 16916
    },
    {
      "epoch": 0.5791509756932557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2994,
      "step": 16917
    },
    {
      "epoch": 0.5791852105443341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2889,
      "step": 16918
    },
    {
      "epoch": 0.5792194453954125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3124,
      "step": 16919
    },
    {
      "epoch": 0.5792536802464909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2789,
      "step": 16920
    },
    {
      "epoch": 0.5792879150975693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3256,
      "step": 16921
    },
    {
      "epoch": 0.5793221499486477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3179,
      "step": 16922
    },
    {
      "epoch": 0.5793563847997262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3188,
      "step": 16923
    },
    {
      "epoch": 0.5793906196508045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3102,
      "step": 16924
    },
    {
      "epoch": 0.579424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3136,
      "step": 16925
    },
    {
      "epoch": 0.5794590893529613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3008,
      "step": 16926
    },
    {
      "epoch": 0.5794933242040398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3111,
      "step": 16927
    },
    {
      "epoch": 0.5795275590551181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2932,
      "step": 16928
    },
    {
      "epoch": 0.5795617939061966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3195,
      "step": 16929
    },
    {
      "epoch": 0.5795960287572749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3037,
      "step": 16930
    },
    {
      "epoch": 0.5796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3111,
      "step": 16931
    },
    {
      "epoch": 0.5796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2975,
      "step": 16932
    },
    {
      "epoch": 0.5796987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3005,
      "step": 16933
    },
    {
      "epoch": 0.5797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3076,
      "step": 16934
    },
    {
      "epoch": 0.5797672030126669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3172,
      "step": 16935
    },
    {
      "epoch": 0.5798014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3097,
      "step": 16936
    },
    {
      "epoch": 0.5798356727148237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3188,
      "step": 16937
    },
    {
      "epoch": 0.5798699075659021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3127,
      "step": 16938
    },
    {
      "epoch": 0.5799041424169805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3181,
      "step": 16939
    },
    {
      "epoch": 0.5799383772680589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3221,
      "step": 16940
    },
    {
      "epoch": 0.5799726121191373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3109,
      "step": 16941
    },
    {
      "epoch": 0.5800068469702157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3139,
      "step": 16942
    },
    {
      "epoch": 0.5800410818212941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3115,
      "step": 16943
    },
    {
      "epoch": 0.5800753166723724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.285,
      "step": 16944
    },
    {
      "epoch": 0.5801095515234509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3074,
      "step": 16945
    },
    {
      "epoch": 0.5801437863745292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3056,
      "step": 16946
    },
    {
      "epoch": 0.5801780212256077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3408,
      "step": 16947
    },
    {
      "epoch": 0.580212256076686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3152,
      "step": 16948
    },
    {
      "epoch": 0.5802464909277645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2966,
      "step": 16949
    },
    {
      "epoch": 0.5802807257788428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3143,
      "step": 16950
    },
    {
      "epoch": 0.5803149606299213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3141,
      "step": 16951
    },
    {
      "epoch": 0.5803491954809996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3132,
      "step": 16952
    },
    {
      "epoch": 0.5803834303320781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3133,
      "step": 16953
    },
    {
      "epoch": 0.5804176651831564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3231,
      "step": 16954
    },
    {
      "epoch": 0.5804519000342349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3223,
      "step": 16955
    },
    {
      "epoch": 0.5804861348853132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3249,
      "step": 16956
    },
    {
      "epoch": 0.5805203697363917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2981,
      "step": 16957
    },
    {
      "epoch": 0.58055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3061,
      "step": 16958
    },
    {
      "epoch": 0.5805888394385484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3088,
      "step": 16959
    },
    {
      "epoch": 0.5806230742896269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3176,
      "step": 16960
    },
    {
      "epoch": 0.5806573091407052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2897,
      "step": 16961
    },
    {
      "epoch": 0.5806915439917837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3058,
      "step": 16962
    },
    {
      "epoch": 0.580725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3031,
      "step": 16963
    },
    {
      "epoch": 0.5807600136939405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3302,
      "step": 16964
    },
    {
      "epoch": 0.5807942485450188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3094,
      "step": 16965
    },
    {
      "epoch": 0.5808284833960973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3002,
      "step": 16966
    },
    {
      "epoch": 0.5808627182471756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3214,
      "step": 16967
    },
    {
      "epoch": 0.5808969530982541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2998,
      "step": 16968
    },
    {
      "epoch": 0.5809311879493324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3003,
      "step": 16969
    },
    {
      "epoch": 0.5809654228004109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3146,
      "step": 16970
    },
    {
      "epoch": 0.5809996576514892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2951,
      "step": 16971
    },
    {
      "epoch": 0.5810338925025677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3104,
      "step": 16972
    },
    {
      "epoch": 0.581068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2831,
      "step": 16973
    },
    {
      "epoch": 0.5811023622047244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3024,
      "step": 16974
    },
    {
      "epoch": 0.5811365970558028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.319,
      "step": 16975
    },
    {
      "epoch": 0.5811708319068812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3157,
      "step": 16976
    },
    {
      "epoch": 0.5812050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3218,
      "step": 16977
    },
    {
      "epoch": 0.581239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2921,
      "step": 16978
    },
    {
      "epoch": 0.5812735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3039,
      "step": 16979
    },
    {
      "epoch": 0.5813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3,
      "step": 16980
    },
    {
      "epoch": 0.5813420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3135,
      "step": 16981
    },
    {
      "epoch": 0.5813762410133516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3159,
      "step": 16982
    },
    {
      "epoch": 0.58141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2983,
      "step": 16983
    },
    {
      "epoch": 0.5814447107155084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3154,
      "step": 16984
    },
    {
      "epoch": 0.5814789455665867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3307,
      "step": 16985
    },
    {
      "epoch": 0.5815131804176652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3142,
      "step": 16986
    },
    {
      "epoch": 0.5815474152687435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3068,
      "step": 16987
    },
    {
      "epoch": 0.581581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3162,
      "step": 16988
    },
    {
      "epoch": 0.5816158849709003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.298,
      "step": 16989
    },
    {
      "epoch": 0.5816501198219788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.298,
      "step": 16990
    },
    {
      "epoch": 0.5816843546730571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3109,
      "step": 16991
    },
    {
      "epoch": 0.5817185895241356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.287,
      "step": 16992
    },
    {
      "epoch": 0.5817528243752139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3227,
      "step": 16993
    },
    {
      "epoch": 0.5817870592262924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2933,
      "step": 16994
    },
    {
      "epoch": 0.5818212940773707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3119,
      "step": 16995
    },
    {
      "epoch": 0.5818555289284492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3015,
      "step": 16996
    },
    {
      "epoch": 0.5818897637795276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3073,
      "step": 16997
    },
    {
      "epoch": 0.581923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3191,
      "step": 16998
    },
    {
      "epoch": 0.5819582334816844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3062,
      "step": 16999
    },
    {
      "epoch": 0.5819924683327627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.29,
      "step": 17000
    },
    {
      "epoch": 0.5820267031838412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2961,
      "step": 17001
    },
    {
      "epoch": 0.5820609380349195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3186,
      "step": 17002
    },
    {
      "epoch": 0.582095172885998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3055,
      "step": 17003
    },
    {
      "epoch": 0.5821294077370763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2968,
      "step": 17004
    },
    {
      "epoch": 0.5821636425881548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3264,
      "step": 17005
    },
    {
      "epoch": 0.5821978774392331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.32,
      "step": 17006
    },
    {
      "epoch": 0.5822321122903116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.302,
      "step": 17007
    },
    {
      "epoch": 0.5822663471413899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3214,
      "step": 17008
    },
    {
      "epoch": 0.5823005819924684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3035,
      "step": 17009
    },
    {
      "epoch": 0.5823348168435467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.296,
      "step": 17010
    },
    {
      "epoch": 0.5823690516946252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2958,
      "step": 17011
    },
    {
      "epoch": 0.5824032865457035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2917,
      "step": 17012
    },
    {
      "epoch": 0.582437521396782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3111,
      "step": 17013
    },
    {
      "epoch": 0.5824717562478603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2955,
      "step": 17014
    },
    {
      "epoch": 0.5825059910989387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3207,
      "step": 17015
    },
    {
      "epoch": 0.5825402259500171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2819,
      "step": 17016
    },
    {
      "epoch": 0.5825744608010955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3136,
      "step": 17017
    },
    {
      "epoch": 0.5826086956521739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.301,
      "step": 17018
    },
    {
      "epoch": 0.5826429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.313,
      "step": 17019
    },
    {
      "epoch": 0.5826771653543307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.325,
      "step": 17020
    },
    {
      "epoch": 0.5827114002054091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.308,
      "step": 17021
    },
    {
      "epoch": 0.5827456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3015,
      "step": 17022
    },
    {
      "epoch": 0.5827798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.315,
      "step": 17023
    },
    {
      "epoch": 0.5828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.317,
      "step": 17024
    },
    {
      "epoch": 0.5828483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3242,
      "step": 17025
    },
    {
      "epoch": 0.582882574460801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3109,
      "step": 17026
    },
    {
      "epoch": 0.5829168093118795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3079,
      "step": 17027
    },
    {
      "epoch": 0.5829510441629578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3143,
      "step": 17028
    },
    {
      "epoch": 0.5829852790140363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2981,
      "step": 17029
    },
    {
      "epoch": 0.5830195138651146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3144,
      "step": 17030
    },
    {
      "epoch": 0.5830537487161931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3087,
      "step": 17031
    },
    {
      "epoch": 0.5830879835672714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3098,
      "step": 17032
    },
    {
      "epoch": 0.5831222184183499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3027,
      "step": 17033
    },
    {
      "epoch": 0.5831564532694283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3072,
      "step": 17034
    },
    {
      "epoch": 0.5831906881205067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3087,
      "step": 17035
    },
    {
      "epoch": 0.5832249229715851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3111,
      "step": 17036
    },
    {
      "epoch": 0.5832591578226635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3207,
      "step": 17037
    },
    {
      "epoch": 0.5832933926737419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3066,
      "step": 17038
    },
    {
      "epoch": 0.5833276275248203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3059,
      "step": 17039
    },
    {
      "epoch": 0.5833618623758987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3142,
      "step": 17040
    },
    {
      "epoch": 0.583396097226977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3048,
      "step": 17041
    },
    {
      "epoch": 0.5834303320780555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3181,
      "step": 17042
    },
    {
      "epoch": 0.5834645669291338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3059,
      "step": 17043
    },
    {
      "epoch": 0.5834988017802123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3135,
      "step": 17044
    },
    {
      "epoch": 0.5835330366312906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3124,
      "step": 17045
    },
    {
      "epoch": 0.5835672714823691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2949,
      "step": 17046
    },
    {
      "epoch": 0.5836015063334474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3199,
      "step": 17047
    },
    {
      "epoch": 0.5836357411845259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.303,
      "step": 17048
    },
    {
      "epoch": 0.5836699760356042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3042,
      "step": 17049
    },
    {
      "epoch": 0.5837042108866827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3092,
      "step": 17050
    },
    {
      "epoch": 0.583738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3109,
      "step": 17051
    },
    {
      "epoch": 0.5837726805888395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3209,
      "step": 17052
    },
    {
      "epoch": 0.5838069154399178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3103,
      "step": 17053
    },
    {
      "epoch": 0.5838411502909963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.303,
      "step": 17054
    },
    {
      "epoch": 0.5838753851420746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3049,
      "step": 17055
    },
    {
      "epoch": 0.583909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2966,
      "step": 17056
    },
    {
      "epoch": 0.5839438548442314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3098,
      "step": 17057
    },
    {
      "epoch": 0.5839780896953098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3165,
      "step": 17058
    },
    {
      "epoch": 0.5840123245463882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3137,
      "step": 17059
    },
    {
      "epoch": 0.5840465593974666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3064,
      "step": 17060
    },
    {
      "epoch": 0.584080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3246,
      "step": 17061
    },
    {
      "epoch": 0.5841150290996234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3052,
      "step": 17062
    },
    {
      "epoch": 0.5841492639507018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2933,
      "step": 17063
    },
    {
      "epoch": 0.5841834988017802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3146,
      "step": 17064
    },
    {
      "epoch": 0.5842177336528586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3126,
      "step": 17065
    },
    {
      "epoch": 0.584251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3232,
      "step": 17066
    },
    {
      "epoch": 0.5842862033550154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3242,
      "step": 17067
    },
    {
      "epoch": 0.5843204382060938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3208,
      "step": 17068
    },
    {
      "epoch": 0.5843546730571721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3188,
      "step": 17069
    },
    {
      "epoch": 0.5843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3157,
      "step": 17070
    },
    {
      "epoch": 0.584423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3132,
      "step": 17071
    },
    {
      "epoch": 0.5844573776104074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3262,
      "step": 17072
    },
    {
      "epoch": 0.5844916124614858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3104,
      "step": 17073
    },
    {
      "epoch": 0.5845258473125642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3033,
      "step": 17074
    },
    {
      "epoch": 0.5845600821636426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3287,
      "step": 17075
    },
    {
      "epoch": 0.584594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.292,
      "step": 17076
    },
    {
      "epoch": 0.5846285518657994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.304,
      "step": 17077
    },
    {
      "epoch": 0.5846627867168778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3177,
      "step": 17078
    },
    {
      "epoch": 0.5846970215679562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3253,
      "step": 17079
    },
    {
      "epoch": 0.5847312564190346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2933,
      "step": 17080
    },
    {
      "epoch": 0.584765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3208,
      "step": 17081
    },
    {
      "epoch": 0.5847997261211914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2994,
      "step": 17082
    },
    {
      "epoch": 0.5848339609722698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2917,
      "step": 17083
    },
    {
      "epoch": 0.5848681958233481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2951,
      "step": 17084
    },
    {
      "epoch": 0.5849024306744266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3102,
      "step": 17085
    },
    {
      "epoch": 0.5849366655255049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3148,
      "step": 17086
    },
    {
      "epoch": 0.5849709003765834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3178,
      "step": 17087
    },
    {
      "epoch": 0.5850051352276617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2914,
      "step": 17088
    },
    {
      "epoch": 0.5850393700787402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3021,
      "step": 17089
    },
    {
      "epoch": 0.5850736049298185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2949,
      "step": 17090
    },
    {
      "epoch": 0.585107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3043,
      "step": 17091
    },
    {
      "epoch": 0.5851420746319753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.308,
      "step": 17092
    },
    {
      "epoch": 0.5851763094830538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3122,
      "step": 17093
    },
    {
      "epoch": 0.5852105443341321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3148,
      "step": 17094
    },
    {
      "epoch": 0.5852447791852106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2848,
      "step": 17095
    },
    {
      "epoch": 0.5852790140362889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2969,
      "step": 17096
    },
    {
      "epoch": 0.5853132488873674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2987,
      "step": 17097
    },
    {
      "epoch": 0.5853474837384457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3069,
      "step": 17098
    },
    {
      "epoch": 0.5853817185895241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2949,
      "step": 17099
    },
    {
      "epoch": 0.5854159534406025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3042,
      "step": 17100
    },
    {
      "epoch": 0.5854501882916809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3162,
      "step": 17101
    },
    {
      "epoch": 0.5854844231427593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2852,
      "step": 17102
    },
    {
      "epoch": 0.5855186579938377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3038,
      "step": 17103
    },
    {
      "epoch": 0.5855528928449161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.302,
      "step": 17104
    },
    {
      "epoch": 0.5855871276959945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.296,
      "step": 17105
    },
    {
      "epoch": 0.5856213625470729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2938,
      "step": 17106
    },
    {
      "epoch": 0.5856555973981513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2881,
      "step": 17107
    },
    {
      "epoch": 0.5856898322492298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.304,
      "step": 17108
    },
    {
      "epoch": 0.5857240671003081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2968,
      "step": 17109
    },
    {
      "epoch": 0.5857583019513866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2906,
      "step": 17110
    },
    {
      "epoch": 0.5857925368024649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3035,
      "step": 17111
    },
    {
      "epoch": 0.5858267716535434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3104,
      "step": 17112
    },
    {
      "epoch": 0.5858610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.302,
      "step": 17113
    },
    {
      "epoch": 0.5858952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3087,
      "step": 17114
    },
    {
      "epoch": 0.5859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3049,
      "step": 17115
    },
    {
      "epoch": 0.5859637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3219,
      "step": 17116
    },
    {
      "epoch": 0.5859979459089353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2942,
      "step": 17117
    },
    {
      "epoch": 0.5860321807600137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3204,
      "step": 17118
    },
    {
      "epoch": 0.5860664156110921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3187,
      "step": 17119
    },
    {
      "epoch": 0.5861006504621705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3165,
      "step": 17120
    },
    {
      "epoch": 0.5861348853132489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3065,
      "step": 17121
    },
    {
      "epoch": 0.5861691201643273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2916,
      "step": 17122
    },
    {
      "epoch": 0.5862033550154057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2893,
      "step": 17123
    },
    {
      "epoch": 0.5862375898664841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3188,
      "step": 17124
    },
    {
      "epoch": 0.5862718247175625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3026,
      "step": 17125
    },
    {
      "epoch": 0.5863060595686409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.294,
      "step": 17126
    },
    {
      "epoch": 0.5863402944197192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2975,
      "step": 17127
    },
    {
      "epoch": 0.5863745292707977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.298,
      "step": 17128
    },
    {
      "epoch": 0.586408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2976,
      "step": 17129
    },
    {
      "epoch": 0.5864429989729545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2912,
      "step": 17130
    },
    {
      "epoch": 0.5864772338240328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3171,
      "step": 17131
    },
    {
      "epoch": 0.5865114686751113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3069,
      "step": 17132
    },
    {
      "epoch": 0.5865457035261896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3174,
      "step": 17133
    },
    {
      "epoch": 0.5865799383772681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3197,
      "step": 17134
    },
    {
      "epoch": 0.5866141732283464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3012,
      "step": 17135
    },
    {
      "epoch": 0.5866484080794249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3086,
      "step": 17136
    },
    {
      "epoch": 0.5866826429305032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3076,
      "step": 17137
    },
    {
      "epoch": 0.5867168777815817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2982,
      "step": 17138
    },
    {
      "epoch": 0.58675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3034,
      "step": 17139
    },
    {
      "epoch": 0.5867853474837385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.302,
      "step": 17140
    },
    {
      "epoch": 0.5868195823348168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3054,
      "step": 17141
    },
    {
      "epoch": 0.5868538171858952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2979,
      "step": 17142
    },
    {
      "epoch": 0.5868880520369736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3097,
      "step": 17143
    },
    {
      "epoch": 0.586922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2994,
      "step": 17144
    },
    {
      "epoch": 0.5869565217391305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2982,
      "step": 17145
    },
    {
      "epoch": 0.5869907565902088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2963,
      "step": 17146
    },
    {
      "epoch": 0.5870249914412873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.319,
      "step": 17147
    },
    {
      "epoch": 0.5870592262923656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3085,
      "step": 17148
    },
    {
      "epoch": 0.5870934611434441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3159,
      "step": 17149
    },
    {
      "epoch": 0.5871276959945224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3051,
      "step": 17150
    },
    {
      "epoch": 0.5871619308456009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.305,
      "step": 17151
    },
    {
      "epoch": 0.5871961656966792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.303,
      "step": 17152
    },
    {
      "epoch": 0.5872304005477577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2788,
      "step": 17153
    },
    {
      "epoch": 0.587264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2716,
      "step": 17154
    },
    {
      "epoch": 0.5872988702499145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2945,
      "step": 17155
    },
    {
      "epoch": 0.5873331051009928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3046,
      "step": 17156
    },
    {
      "epoch": 0.5873673399520712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2888,
      "step": 17157
    },
    {
      "epoch": 0.5874015748031496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.282,
      "step": 17158
    },
    {
      "epoch": 0.587435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3026,
      "step": 17159
    },
    {
      "epoch": 0.5874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3046,
      "step": 17160
    },
    {
      "epoch": 0.5875042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2998,
      "step": 17161
    },
    {
      "epoch": 0.5875385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3174,
      "step": 17162
    },
    {
      "epoch": 0.5875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3045,
      "step": 17163
    },
    {
      "epoch": 0.58760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3012,
      "step": 17164
    },
    {
      "epoch": 0.5876412187606984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3013,
      "step": 17165
    },
    {
      "epoch": 0.5876754536117768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3048,
      "step": 17166
    },
    {
      "epoch": 0.5877096884628552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3073,
      "step": 17167
    },
    {
      "epoch": 0.5877439233139335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3077,
      "step": 17168
    },
    {
      "epoch": 0.587778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2908,
      "step": 17169
    },
    {
      "epoch": 0.5878123930160903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3285,
      "step": 17170
    },
    {
      "epoch": 0.5878466278671688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2946,
      "step": 17171
    },
    {
      "epoch": 0.5878808627182471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3088,
      "step": 17172
    },
    {
      "epoch": 0.5879150975693256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2963,
      "step": 17173
    },
    {
      "epoch": 0.5879493324204039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2981,
      "step": 17174
    },
    {
      "epoch": 0.5879835672714824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2957,
      "step": 17175
    },
    {
      "epoch": 0.5880178021225607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3,
      "step": 17176
    },
    {
      "epoch": 0.5880520369736392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2988,
      "step": 17177
    },
    {
      "epoch": 0.5880862718247175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2986,
      "step": 17178
    },
    {
      "epoch": 0.588120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3194,
      "step": 17179
    },
    {
      "epoch": 0.5881547415268743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2896,
      "step": 17180
    },
    {
      "epoch": 0.5881889763779528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3155,
      "step": 17181
    },
    {
      "epoch": 0.5882232112290312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2921,
      "step": 17182
    },
    {
      "epoch": 0.5882574460801095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3035,
      "step": 17183
    },
    {
      "epoch": 0.588291680931188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3112,
      "step": 17184
    },
    {
      "epoch": 0.5883259157822663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3057,
      "step": 17185
    },
    {
      "epoch": 0.5883601506333448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3037,
      "step": 17186
    },
    {
      "epoch": 0.5883943854844231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3102,
      "step": 17187
    },
    {
      "epoch": 0.5884286203355016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3052,
      "step": 17188
    },
    {
      "epoch": 0.5884628551865799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2881,
      "step": 17189
    },
    {
      "epoch": 0.5884970900376584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3004,
      "step": 17190
    },
    {
      "epoch": 0.5885313248887367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3119,
      "step": 17191
    },
    {
      "epoch": 0.5885655597398152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3008,
      "step": 17192
    },
    {
      "epoch": 0.5885997945908935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2847,
      "step": 17193
    },
    {
      "epoch": 0.588634029441972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3143,
      "step": 17194
    },
    {
      "epoch": 0.5886682642930503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3026,
      "step": 17195
    },
    {
      "epoch": 0.5887024991441288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3138,
      "step": 17196
    },
    {
      "epoch": 0.5887367339952071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2956,
      "step": 17197
    },
    {
      "epoch": 0.5887709688462855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3107,
      "step": 17198
    },
    {
      "epoch": 0.5888052036973639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3054,
      "step": 17199
    },
    {
      "epoch": 0.5888394385484423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2933,
      "step": 17200
    },
    {
      "epoch": 0.5888736733995207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2993,
      "step": 17201
    },
    {
      "epoch": 0.5889079082505991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2944,
      "step": 17202
    },
    {
      "epoch": 0.5889421431016775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3072,
      "step": 17203
    },
    {
      "epoch": 0.5889763779527559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2964,
      "step": 17204
    },
    {
      "epoch": 0.5890106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3238,
      "step": 17205
    },
    {
      "epoch": 0.5890448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2911,
      "step": 17206
    },
    {
      "epoch": 0.5890790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2833,
      "step": 17207
    },
    {
      "epoch": 0.5891133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.321,
      "step": 17208
    },
    {
      "epoch": 0.5891475522081479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3126,
      "step": 17209
    },
    {
      "epoch": 0.5891817870592263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3111,
      "step": 17210
    },
    {
      "epoch": 0.5892160219103046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.307,
      "step": 17211
    },
    {
      "epoch": 0.5892502567613831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2974,
      "step": 17212
    },
    {
      "epoch": 0.5892844916124614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.317,
      "step": 17213
    },
    {
      "epoch": 0.5893187264635399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2859,
      "step": 17214
    },
    {
      "epoch": 0.5893529613146182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2904,
      "step": 17215
    },
    {
      "epoch": 0.5893871961656967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3018,
      "step": 17216
    },
    {
      "epoch": 0.5894214310167751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3064,
      "step": 17217
    },
    {
      "epoch": 0.5894556658678535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.31,
      "step": 17218
    },
    {
      "epoch": 0.5894899007189319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3062,
      "step": 17219
    },
    {
      "epoch": 0.5895241355700103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3013,
      "step": 17220
    },
    {
      "epoch": 0.5895583704210887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3027,
      "step": 17221
    },
    {
      "epoch": 0.5895926052721671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2909,
      "step": 17222
    },
    {
      "epoch": 0.5896268401232455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3039,
      "step": 17223
    },
    {
      "epoch": 0.5896610749743239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3005,
      "step": 17224
    },
    {
      "epoch": 0.5896953098254023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2886,
      "step": 17225
    },
    {
      "epoch": 0.5897295446764806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3058,
      "step": 17226
    },
    {
      "epoch": 0.5897637795275591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3017,
      "step": 17227
    },
    {
      "epoch": 0.5897980143786374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2988,
      "step": 17228
    },
    {
      "epoch": 0.5898322492297159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.301,
      "step": 17229
    },
    {
      "epoch": 0.5898664840807942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2939,
      "step": 17230
    },
    {
      "epoch": 0.5899007189318727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3016,
      "step": 17231
    },
    {
      "epoch": 0.589934953782951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2907,
      "step": 17232
    },
    {
      "epoch": 0.5899691886340295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2849,
      "step": 17233
    },
    {
      "epoch": 0.5900034234851078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2931,
      "step": 17234
    },
    {
      "epoch": 0.5900376583361863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3078,
      "step": 17235
    },
    {
      "epoch": 0.5900718931872646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2935,
      "step": 17236
    },
    {
      "epoch": 0.5901061280383431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3082,
      "step": 17237
    },
    {
      "epoch": 0.5901403628894214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2958,
      "step": 17238
    },
    {
      "epoch": 0.5901745977404999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2999,
      "step": 17239
    },
    {
      "epoch": 0.5902088325915782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2997,
      "step": 17240
    },
    {
      "epoch": 0.5902430674426566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2999,
      "step": 17241
    },
    {
      "epoch": 0.590277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3015,
      "step": 17242
    },
    {
      "epoch": 0.5903115371448134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.301,
      "step": 17243
    },
    {
      "epoch": 0.5903457719958918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2897,
      "step": 17244
    },
    {
      "epoch": 0.5903800068469702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3021,
      "step": 17245
    },
    {
      "epoch": 0.5904142416980486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2843,
      "step": 17246
    },
    {
      "epoch": 0.590448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3195,
      "step": 17247
    },
    {
      "epoch": 0.5904827114002054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2916,
      "step": 17248
    },
    {
      "epoch": 0.5905169462512838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3141,
      "step": 17249
    },
    {
      "epoch": 0.5905511811023622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2927,
      "step": 17250
    },
    {
      "epoch": 0.5905854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2994,
      "step": 17251
    },
    {
      "epoch": 0.590619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2869,
      "step": 17252
    },
    {
      "epoch": 0.5906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3076,
      "step": 17253
    },
    {
      "epoch": 0.5906881205066759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3117,
      "step": 17254
    },
    {
      "epoch": 0.5907223553577542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2878,
      "step": 17255
    },
    {
      "epoch": 0.5907565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3219,
      "step": 17256
    },
    {
      "epoch": 0.590790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3087,
      "step": 17257
    },
    {
      "epoch": 0.5908250599109894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2838,
      "step": 17258
    },
    {
      "epoch": 0.5908592947620678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3151,
      "step": 17259
    },
    {
      "epoch": 0.5908935296131462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2916,
      "step": 17260
    },
    {
      "epoch": 0.5909277644642246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3075,
      "step": 17261
    },
    {
      "epoch": 0.590961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3,
      "step": 17262
    },
    {
      "epoch": 0.5909962341663814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.307,
      "step": 17263
    },
    {
      "epoch": 0.5910304690174598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2885,
      "step": 17264
    },
    {
      "epoch": 0.5910647038685382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3053,
      "step": 17265
    },
    {
      "epoch": 0.5910989387196166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2883,
      "step": 17266
    },
    {
      "epoch": 0.591133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.302,
      "step": 17267
    },
    {
      "epoch": 0.5911674084217734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2886,
      "step": 17268
    },
    {
      "epoch": 0.5912016432728517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2977,
      "step": 17269
    },
    {
      "epoch": 0.5912358781239302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.294,
      "step": 17270
    },
    {
      "epoch": 0.5912701129750085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2972,
      "step": 17271
    },
    {
      "epoch": 0.591304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2999,
      "step": 17272
    },
    {
      "epoch": 0.5913385826771653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2986,
      "step": 17273
    },
    {
      "epoch": 0.5913728175282438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2985,
      "step": 17274
    },
    {
      "epoch": 0.5914070523793221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3145,
      "step": 17275
    },
    {
      "epoch": 0.5914412872304006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.323,
      "step": 17276
    },
    {
      "epoch": 0.5914755220814789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3101,
      "step": 17277
    },
    {
      "epoch": 0.5915097569325574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2966,
      "step": 17278
    },
    {
      "epoch": 0.5915439917836357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3102,
      "step": 17279
    },
    {
      "epoch": 0.5915782266347142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3003,
      "step": 17280
    },
    {
      "epoch": 0.5916124614857925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2897,
      "step": 17281
    },
    {
      "epoch": 0.591646696336871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3049,
      "step": 17282
    },
    {
      "epoch": 0.5916809311879493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2938,
      "step": 17283
    },
    {
      "epoch": 0.5917151660390277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2862,
      "step": 17284
    },
    {
      "epoch": 0.5917494008901061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2908,
      "step": 17285
    },
    {
      "epoch": 0.5917836357411845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3066,
      "step": 17286
    },
    {
      "epoch": 0.5918178705922629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3004,
      "step": 17287
    },
    {
      "epoch": 0.5918521054433413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2924,
      "step": 17288
    },
    {
      "epoch": 0.5918863402944197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2988,
      "step": 17289
    },
    {
      "epoch": 0.5919205751454981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2948,
      "step": 17290
    },
    {
      "epoch": 0.5919548099965766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3104,
      "step": 17291
    },
    {
      "epoch": 0.5919890448476549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2957,
      "step": 17292
    },
    {
      "epoch": 0.5920232796987334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3014,
      "step": 17293
    },
    {
      "epoch": 0.5920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2995,
      "step": 17294
    },
    {
      "epoch": 0.5920917494008902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3077,
      "step": 17295
    },
    {
      "epoch": 0.5921259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3017,
      "step": 17296
    },
    {
      "epoch": 0.592160219103047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3049,
      "step": 17297
    },
    {
      "epoch": 0.5921944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2869,
      "step": 17298
    },
    {
      "epoch": 0.5922286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3039,
      "step": 17299
    },
    {
      "epoch": 0.5922629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2924,
      "step": 17300
    },
    {
      "epoch": 0.5922971585073605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3149,
      "step": 17301
    },
    {
      "epoch": 0.5923313933584389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.29,
      "step": 17302
    },
    {
      "epoch": 0.5923656282095173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.29,
      "step": 17303
    },
    {
      "epoch": 0.5923998630605957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2978,
      "step": 17304
    },
    {
      "epoch": 0.5924340979116741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3006,
      "step": 17305
    },
    {
      "epoch": 0.5924683327627525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2879,
      "step": 17306
    },
    {
      "epoch": 0.5925025676138309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2876,
      "step": 17307
    },
    {
      "epoch": 0.5925368024649093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2911,
      "step": 17308
    },
    {
      "epoch": 0.5925710373159877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2809,
      "step": 17309
    },
    {
      "epoch": 0.592605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2913,
      "step": 17310
    },
    {
      "epoch": 0.5926395070181445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3156,
      "step": 17311
    },
    {
      "epoch": 0.5926737418692228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2707,
      "step": 17312
    },
    {
      "epoch": 0.5927079767203013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3026,
      "step": 17313
    },
    {
      "epoch": 0.5927422115713796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3002,
      "step": 17314
    },
    {
      "epoch": 0.5927764464224581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2911,
      "step": 17315
    },
    {
      "epoch": 0.5928106812735364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3026,
      "step": 17316
    },
    {
      "epoch": 0.5928449161246149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3017,
      "step": 17317
    },
    {
      "epoch": 0.5928791509756932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2883,
      "step": 17318
    },
    {
      "epoch": 0.5929133858267717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3062,
      "step": 17319
    },
    {
      "epoch": 0.59294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2955,
      "step": 17320
    },
    {
      "epoch": 0.5929818555289285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2968,
      "step": 17321
    },
    {
      "epoch": 0.5930160903800068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.28,
      "step": 17322
    },
    {
      "epoch": 0.5930503252310853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3071,
      "step": 17323
    },
    {
      "epoch": 0.5930845600821636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2902,
      "step": 17324
    },
    {
      "epoch": 0.593118794933242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2906,
      "step": 17325
    },
    {
      "epoch": 0.5931530297843204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2973,
      "step": 17326
    },
    {
      "epoch": 0.5931872646353988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2961,
      "step": 17327
    },
    {
      "epoch": 0.5932214994864773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2872,
      "step": 17328
    },
    {
      "epoch": 0.5932557343375556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.307,
      "step": 17329
    },
    {
      "epoch": 0.5932899691886341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2802,
      "step": 17330
    },
    {
      "epoch": 0.5933242040397124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2868,
      "step": 17331
    },
    {
      "epoch": 0.5933584388907909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3113,
      "step": 17332
    },
    {
      "epoch": 0.5933926737418692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2884,
      "step": 17333
    },
    {
      "epoch": 0.5934269085929477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3012,
      "step": 17334
    },
    {
      "epoch": 0.593461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2807,
      "step": 17335
    },
    {
      "epoch": 0.5934953782951045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2879,
      "step": 17336
    },
    {
      "epoch": 0.5935296131461828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3024,
      "step": 17337
    },
    {
      "epoch": 0.5935638479972613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.291,
      "step": 17338
    },
    {
      "epoch": 0.5935980828483396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2864,
      "step": 17339
    },
    {
      "epoch": 0.593632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2806,
      "step": 17340
    },
    {
      "epoch": 0.5936665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3087,
      "step": 17341
    },
    {
      "epoch": 0.5937007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2764,
      "step": 17342
    },
    {
      "epoch": 0.5937350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2886,
      "step": 17343
    },
    {
      "epoch": 0.5937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3124,
      "step": 17344
    },
    {
      "epoch": 0.59380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2943,
      "step": 17345
    },
    {
      "epoch": 0.5938377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2783,
      "step": 17346
    },
    {
      "epoch": 0.5938719616569668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3022,
      "step": 17347
    },
    {
      "epoch": 0.5939061965080452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2919,
      "step": 17348
    },
    {
      "epoch": 0.5939404313591236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2905,
      "step": 17349
    },
    {
      "epoch": 0.593974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2901,
      "step": 17350
    },
    {
      "epoch": 0.5940089010612803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3148,
      "step": 17351
    },
    {
      "epoch": 0.5940431359123588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2829,
      "step": 17352
    },
    {
      "epoch": 0.5940773707634371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3015,
      "step": 17353
    },
    {
      "epoch": 0.5941116056145156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3005,
      "step": 17354
    },
    {
      "epoch": 0.5941458404655939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.283,
      "step": 17355
    },
    {
      "epoch": 0.5941800753166724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3023,
      "step": 17356
    },
    {
      "epoch": 0.5942143101677507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2838,
      "step": 17357
    },
    {
      "epoch": 0.5942485450188292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2866,
      "step": 17358
    },
    {
      "epoch": 0.5942827798699075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2947,
      "step": 17359
    },
    {
      "epoch": 0.594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3029,
      "step": 17360
    },
    {
      "epoch": 0.5943512495720643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2909,
      "step": 17361
    },
    {
      "epoch": 0.5943854844231428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2888,
      "step": 17362
    },
    {
      "epoch": 0.5944197192742211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2965,
      "step": 17363
    },
    {
      "epoch": 0.5944539541252996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3084,
      "step": 17364
    },
    {
      "epoch": 0.594488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2916,
      "step": 17365
    },
    {
      "epoch": 0.5945224238274563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3089,
      "step": 17366
    },
    {
      "epoch": 0.5945566586785348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2878,
      "step": 17367
    },
    {
      "epoch": 0.5945908935296131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2957,
      "step": 17368
    },
    {
      "epoch": 0.5946251283806916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2789,
      "step": 17369
    },
    {
      "epoch": 0.5946593632317699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3308,
      "step": 17370
    },
    {
      "epoch": 0.5946935980828484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3017,
      "step": 17371
    },
    {
      "epoch": 0.5947278329339267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2969,
      "step": 17372
    },
    {
      "epoch": 0.5947620677850052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2991,
      "step": 17373
    },
    {
      "epoch": 0.5947963026360835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2914,
      "step": 17374
    },
    {
      "epoch": 0.594830537487162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.28,
      "step": 17375
    },
    {
      "epoch": 0.5948647723382403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2996,
      "step": 17376
    },
    {
      "epoch": 0.5948990071893188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2959,
      "step": 17377
    },
    {
      "epoch": 0.5949332420403971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2917,
      "step": 17378
    },
    {
      "epoch": 0.5949674768914756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.325,
      "step": 17379
    },
    {
      "epoch": 0.5950017117425539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2966,
      "step": 17380
    },
    {
      "epoch": 0.5950359465936323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3033,
      "step": 17381
    },
    {
      "epoch": 0.5950701814447107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3062,
      "step": 17382
    },
    {
      "epoch": 0.5951044162957891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2864,
      "step": 17383
    },
    {
      "epoch": 0.5951386511468675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3024,
      "step": 17384
    },
    {
      "epoch": 0.5951728859979459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3084,
      "step": 17385
    },
    {
      "epoch": 0.5952071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2914,
      "step": 17386
    },
    {
      "epoch": 0.5952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2923,
      "step": 17387
    },
    {
      "epoch": 0.5952755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2918,
      "step": 17388
    },
    {
      "epoch": 0.5953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.293,
      "step": 17389
    },
    {
      "epoch": 0.5953440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.295,
      "step": 17390
    },
    {
      "epoch": 0.5953782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2909,
      "step": 17391
    },
    {
      "epoch": 0.5954125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3128,
      "step": 17392
    },
    {
      "epoch": 0.5954467648065731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2934,
      "step": 17393
    },
    {
      "epoch": 0.5954809996576514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2987,
      "step": 17394
    },
    {
      "epoch": 0.5955152345087299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2875,
      "step": 17395
    },
    {
      "epoch": 0.5955494693598082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2803,
      "step": 17396
    },
    {
      "epoch": 0.5955837042108867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2882,
      "step": 17397
    },
    {
      "epoch": 0.595617939061965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2977,
      "step": 17398
    },
    {
      "epoch": 0.5956521739130435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3032,
      "step": 17399
    },
    {
      "epoch": 0.5956864087641218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3029,
      "step": 17400
    },
    {
      "epoch": 0.5957206436152003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2991,
      "step": 17401
    },
    {
      "epoch": 0.5957548784662787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2968,
      "step": 17402
    },
    {
      "epoch": 0.5957891133173571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2895,
      "step": 17403
    },
    {
      "epoch": 0.5958233481684355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2961,
      "step": 17404
    },
    {
      "epoch": 0.5958575830195139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2963,
      "step": 17405
    },
    {
      "epoch": 0.5958918178705923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.277,
      "step": 17406
    },
    {
      "epoch": 0.5959260527216707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2897,
      "step": 17407
    },
    {
      "epoch": 0.5959602875727491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2949,
      "step": 17408
    },
    {
      "epoch": 0.5959945224238274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3053,
      "step": 17409
    },
    {
      "epoch": 0.5960287572749059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2974,
      "step": 17410
    },
    {
      "epoch": 0.5960629921259842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2851,
      "step": 17411
    },
    {
      "epoch": 0.5960972269770627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2843,
      "step": 17412
    },
    {
      "epoch": 0.596131461828141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2896,
      "step": 17413
    },
    {
      "epoch": 0.5961656966792195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2853,
      "step": 17414
    },
    {
      "epoch": 0.5961999315302978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2906,
      "step": 17415
    },
    {
      "epoch": 0.5962341663813763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2754,
      "step": 17416
    },
    {
      "epoch": 0.5962684012324546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2849,
      "step": 17417
    },
    {
      "epoch": 0.5963026360835331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2927,
      "step": 17418
    },
    {
      "epoch": 0.5963368709346114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 17419
    },
    {
      "epoch": 0.5963711057856899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2762,
      "step": 17420
    },
    {
      "epoch": 0.5964053406367682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2889,
      "step": 17421
    },
    {
      "epoch": 0.5964395754878467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2855,
      "step": 17422
    },
    {
      "epoch": 0.596473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2975,
      "step": 17423
    },
    {
      "epoch": 0.5965080451900034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3063,
      "step": 17424
    },
    {
      "epoch": 0.5965422800410818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2932,
      "step": 17425
    },
    {
      "epoch": 0.5965765148921602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.303,
      "step": 17426
    },
    {
      "epoch": 0.5966107497432386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.295,
      "step": 17427
    },
    {
      "epoch": 0.596644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3034,
      "step": 17428
    },
    {
      "epoch": 0.5966792194453954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2998,
      "step": 17429
    },
    {
      "epoch": 0.5967134542964738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2885,
      "step": 17430
    },
    {
      "epoch": 0.5967476891475522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2978,
      "step": 17431
    },
    {
      "epoch": 0.5967819239986306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.286,
      "step": 17432
    },
    {
      "epoch": 0.596816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2822,
      "step": 17433
    },
    {
      "epoch": 0.5968503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2943,
      "step": 17434
    },
    {
      "epoch": 0.5968846285518657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2966,
      "step": 17435
    },
    {
      "epoch": 0.5969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2872,
      "step": 17436
    },
    {
      "epoch": 0.5969530982540225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2852,
      "step": 17437
    },
    {
      "epoch": 0.596987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2937,
      "step": 17438
    },
    {
      "epoch": 0.5970215679561794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3013,
      "step": 17439
    },
    {
      "epoch": 0.5970558028072578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2752,
      "step": 17440
    },
    {
      "epoch": 0.5970900376583362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2991,
      "step": 17441
    },
    {
      "epoch": 0.5971242725094146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.301,
      "step": 17442
    },
    {
      "epoch": 0.597158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2907,
      "step": 17443
    },
    {
      "epoch": 0.5971927422115714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2896,
      "step": 17444
    },
    {
      "epoch": 0.5972269770626498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2991,
      "step": 17445
    },
    {
      "epoch": 0.5972612119137282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3113,
      "step": 17446
    },
    {
      "epoch": 0.5972954467648066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2983,
      "step": 17447
    },
    {
      "epoch": 0.597329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2816,
      "step": 17448
    },
    {
      "epoch": 0.5973639164669634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3083,
      "step": 17449
    },
    {
      "epoch": 0.5973981513180417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2697,
      "step": 17450
    },
    {
      "epoch": 0.5974323861691202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3008,
      "step": 17451
    },
    {
      "epoch": 0.5974666210201985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2975,
      "step": 17452
    },
    {
      "epoch": 0.597500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3137,
      "step": 17453
    },
    {
      "epoch": 0.5975350907223553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3065,
      "step": 17454
    },
    {
      "epoch": 0.5975693255734338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2855,
      "step": 17455
    },
    {
      "epoch": 0.5976035604245121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2847,
      "step": 17456
    },
    {
      "epoch": 0.5976377952755906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2762,
      "step": 17457
    },
    {
      "epoch": 0.5976720301266689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.283,
      "step": 17458
    },
    {
      "epoch": 0.5977062649777474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.295,
      "step": 17459
    },
    {
      "epoch": 0.5977404998288257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2838,
      "step": 17460
    },
    {
      "epoch": 0.5977747346799042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.284,
      "step": 17461
    },
    {
      "epoch": 0.5978089695309825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2963,
      "step": 17462
    },
    {
      "epoch": 0.597843204382061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2966,
      "step": 17463
    },
    {
      "epoch": 0.5978774392331393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2929,
      "step": 17464
    },
    {
      "epoch": 0.5979116740842177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2886,
      "step": 17465
    },
    {
      "epoch": 0.5979459089352961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2846,
      "step": 17466
    },
    {
      "epoch": 0.5979801437863745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2958,
      "step": 17467
    },
    {
      "epoch": 0.5980143786374529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2864,
      "step": 17468
    },
    {
      "epoch": 0.5980486134885313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2841,
      "step": 17469
    },
    {
      "epoch": 0.5980828483396097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2933,
      "step": 17470
    },
    {
      "epoch": 0.5981170831906881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2822,
      "step": 17471
    },
    {
      "epoch": 0.5981513180417665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3081,
      "step": 17472
    },
    {
      "epoch": 0.5981855528928449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2799,
      "step": 17473
    },
    {
      "epoch": 0.5982197877439233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2892,
      "step": 17474
    },
    {
      "epoch": 0.5982540225950017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3023,
      "step": 17475
    },
    {
      "epoch": 0.5982882574460802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2687,
      "step": 17476
    },
    {
      "epoch": 0.5983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2843,
      "step": 17477
    },
    {
      "epoch": 0.598356727148237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.288,
      "step": 17478
    },
    {
      "epoch": 0.5983909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2955,
      "step": 17479
    },
    {
      "epoch": 0.5984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2855,
      "step": 17480
    },
    {
      "epoch": 0.5984594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2843,
      "step": 17481
    },
    {
      "epoch": 0.5984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2875,
      "step": 17482
    },
    {
      "epoch": 0.5985279014036289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2939,
      "step": 17483
    },
    {
      "epoch": 0.5985621362547073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3029,
      "step": 17484
    },
    {
      "epoch": 0.5985963711057857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3058,
      "step": 17485
    },
    {
      "epoch": 0.5986306059568641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2977,
      "step": 17486
    },
    {
      "epoch": 0.5986648408079425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2956,
      "step": 17487
    },
    {
      "epoch": 0.5986990756590209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2859,
      "step": 17488
    },
    {
      "epoch": 0.5987333105100993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.294,
      "step": 17489
    },
    {
      "epoch": 0.5987675453611777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2923,
      "step": 17490
    },
    {
      "epoch": 0.598801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3008,
      "step": 17491
    },
    {
      "epoch": 0.5988360150633345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2891,
      "step": 17492
    },
    {
      "epoch": 0.5988702499144128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2775,
      "step": 17493
    },
    {
      "epoch": 0.5989044847654913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2906,
      "step": 17494
    },
    {
      "epoch": 0.5989387196165696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2807,
      "step": 17495
    },
    {
      "epoch": 0.5989729544676481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.289,
      "step": 17496
    },
    {
      "epoch": 0.5990071893187264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2727,
      "step": 17497
    },
    {
      "epoch": 0.5990414241698049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3114,
      "step": 17498
    },
    {
      "epoch": 0.5990756590208832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.304,
      "step": 17499
    },
    {
      "epoch": 0.5991098938719617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.284,
      "step": 17500
    },
    {
      "epoch": 0.59914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2871,
      "step": 17501
    },
    {
      "epoch": 0.5991783635741185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.293,
      "step": 17502
    },
    {
      "epoch": 0.5992125984251968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2875,
      "step": 17503
    },
    {
      "epoch": 0.5992468332762753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2722,
      "step": 17504
    },
    {
      "epoch": 0.5992810681273536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2933,
      "step": 17505
    },
    {
      "epoch": 0.599315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.296,
      "step": 17506
    },
    {
      "epoch": 0.5993495378295104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3033,
      "step": 17507
    },
    {
      "epoch": 0.5993837726805888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2876,
      "step": 17508
    },
    {
      "epoch": 0.5994180075316672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2972,
      "step": 17509
    },
    {
      "epoch": 0.5994522423827456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2814,
      "step": 17510
    },
    {
      "epoch": 0.599486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3053,
      "step": 17511
    },
    {
      "epoch": 0.5995207120849024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2897,
      "step": 17512
    },
    {
      "epoch": 0.5995549469359809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.299,
      "step": 17513
    },
    {
      "epoch": 0.5995891817870592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2969,
      "step": 17514
    },
    {
      "epoch": 0.5996234166381377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.309,
      "step": 17515
    },
    {
      "epoch": 0.599657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2826,
      "step": 17516
    },
    {
      "epoch": 0.5996918863402945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2863,
      "step": 17517
    },
    {
      "epoch": 0.5997261211913728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2907,
      "step": 17518
    },
    {
      "epoch": 0.5997603560424513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2905,
      "step": 17519
    },
    {
      "epoch": 0.5997945908935296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2912,
      "step": 17520
    },
    {
      "epoch": 0.599828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2948,
      "step": 17521
    },
    {
      "epoch": 0.5998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2891,
      "step": 17522
    },
    {
      "epoch": 0.5998972954467648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3017,
      "step": 17523
    },
    {
      "epoch": 0.5999315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2875,
      "step": 17524
    },
    {
      "epoch": 0.5999657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2955,
      "step": 17525
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2797,
      "step": 17526
    },
    {
      "epoch": 0.6000342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2863,
      "step": 17527
    },
    {
      "epoch": 0.6000684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2932,
      "step": 17528
    },
    {
      "epoch": 0.6001027045532352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2966,
      "step": 17529
    },
    {
      "epoch": 0.6001369394043136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2879,
      "step": 17530
    },
    {
      "epoch": 0.600171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2756,
      "step": 17531
    },
    {
      "epoch": 0.6002054091064704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2831,
      "step": 17532
    },
    {
      "epoch": 0.6002396439575488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3002,
      "step": 17533
    },
    {
      "epoch": 0.6002738788086271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2881,
      "step": 17534
    },
    {
      "epoch": 0.6003081136597056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.302,
      "step": 17535
    },
    {
      "epoch": 0.6003423485107839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2963,
      "step": 17536
    },
    {
      "epoch": 0.6003765833618624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2899,
      "step": 17537
    },
    {
      "epoch": 0.6004108182129407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.284,
      "step": 17538
    },
    {
      "epoch": 0.6004450530640192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2895,
      "step": 17539
    },
    {
      "epoch": 0.6004792879150975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3115,
      "step": 17540
    },
    {
      "epoch": 0.600513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2687,
      "step": 17541
    },
    {
      "epoch": 0.6005477576172543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2868,
      "step": 17542
    },
    {
      "epoch": 0.6005819924683328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2923,
      "step": 17543
    },
    {
      "epoch": 0.6006162273194111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.287,
      "step": 17544
    },
    {
      "epoch": 0.6006504621704896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2911,
      "step": 17545
    },
    {
      "epoch": 0.6006846970215679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2958,
      "step": 17546
    },
    {
      "epoch": 0.6007189318726464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3019,
      "step": 17547
    },
    {
      "epoch": 0.6007531667237247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2881,
      "step": 17548
    },
    {
      "epoch": 0.6007874015748031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2938,
      "step": 17549
    },
    {
      "epoch": 0.6008216364258816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2874,
      "step": 17550
    },
    {
      "epoch": 0.6008558712769599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2943,
      "step": 17551
    },
    {
      "epoch": 0.6008901061280384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2811,
      "step": 17552
    },
    {
      "epoch": 0.6009243409791167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.284,
      "step": 17553
    },
    {
      "epoch": 0.6009585758301952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2823,
      "step": 17554
    },
    {
      "epoch": 0.6009928106812735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2599,
      "step": 17555
    },
    {
      "epoch": 0.601027045532352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2963,
      "step": 17556
    },
    {
      "epoch": 0.6010612803834303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.284,
      "step": 17557
    },
    {
      "epoch": 0.6010955152345088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2944,
      "step": 17558
    },
    {
      "epoch": 0.6011297500855871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.289,
      "step": 17559
    },
    {
      "epoch": 0.6011639849366656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2961,
      "step": 17560
    },
    {
      "epoch": 0.6011982197877439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2853,
      "step": 17561
    },
    {
      "epoch": 0.6012324546388224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2768,
      "step": 17562
    },
    {
      "epoch": 0.6012666894899007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2881,
      "step": 17563
    },
    {
      "epoch": 0.6013009243409791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3004,
      "step": 17564
    },
    {
      "epoch": 0.6013351591920575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3011,
      "step": 17565
    },
    {
      "epoch": 0.6013693940431359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2968,
      "step": 17566
    },
    {
      "epoch": 0.6014036288942143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2849,
      "step": 17567
    },
    {
      "epoch": 0.6014378637452927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3008,
      "step": 17568
    },
    {
      "epoch": 0.6014720985963711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2901,
      "step": 17569
    },
    {
      "epoch": 0.6015063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2842,
      "step": 17570
    },
    {
      "epoch": 0.6015405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2798,
      "step": 17571
    },
    {
      "epoch": 0.6015748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2969,
      "step": 17572
    },
    {
      "epoch": 0.6016090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2744,
      "step": 17573
    },
    {
      "epoch": 0.6016432728517631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2854,
      "step": 17574
    },
    {
      "epoch": 0.6016775077028415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2785,
      "step": 17575
    },
    {
      "epoch": 0.6017117425539199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2964,
      "step": 17576
    },
    {
      "epoch": 0.6017459774049982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2811,
      "step": 17577
    },
    {
      "epoch": 0.6017802122560767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2984,
      "step": 17578
    },
    {
      "epoch": 0.601814447107155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2897,
      "step": 17579
    },
    {
      "epoch": 0.6018486819582335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2932,
      "step": 17580
    },
    {
      "epoch": 0.6018829168093118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2797,
      "step": 17581
    },
    {
      "epoch": 0.6019171516603903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2877,
      "step": 17582
    },
    {
      "epoch": 0.6019513865114686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2872,
      "step": 17583
    },
    {
      "epoch": 0.6019856213625471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 17584
    },
    {
      "epoch": 0.6020198562136255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2982,
      "step": 17585
    },
    {
      "epoch": 0.6020540910647039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2996,
      "step": 17586
    },
    {
      "epoch": 0.6020883259157823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2911,
      "step": 17587
    },
    {
      "epoch": 0.6021225607668607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2766,
      "step": 17588
    },
    {
      "epoch": 0.6021567956179391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.288,
      "step": 17589
    },
    {
      "epoch": 0.6021910304690175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2864,
      "step": 17590
    },
    {
      "epoch": 0.6022252653200959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2795,
      "step": 17591
    },
    {
      "epoch": 0.6022595001711742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2814,
      "step": 17592
    },
    {
      "epoch": 0.6022937350222527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2858,
      "step": 17593
    },
    {
      "epoch": 0.602327969873331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2816,
      "step": 17594
    },
    {
      "epoch": 0.6023622047244095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2899,
      "step": 17595
    },
    {
      "epoch": 0.6023964395754878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2835,
      "step": 17596
    },
    {
      "epoch": 0.6024306744265663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 17597
    },
    {
      "epoch": 0.6024649092776446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2872,
      "step": 17598
    },
    {
      "epoch": 0.6024991441287231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2818,
      "step": 17599
    },
    {
      "epoch": 0.6025333789798014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2993,
      "step": 17600
    },
    {
      "epoch": 0.6025676138308799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2963,
      "step": 17601
    },
    {
      "epoch": 0.6026018486819582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2841,
      "step": 17602
    },
    {
      "epoch": 0.6026360835330367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3024,
      "step": 17603
    },
    {
      "epoch": 0.602670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2692,
      "step": 17604
    },
    {
      "epoch": 0.6027045532351935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2802,
      "step": 17605
    },
    {
      "epoch": 0.6027387880862718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2723,
      "step": 17606
    },
    {
      "epoch": 0.6027730229373502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2899,
      "step": 17607
    },
    {
      "epoch": 0.6028072577884286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3117,
      "step": 17608
    },
    {
      "epoch": 0.602841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2969,
      "step": 17609
    },
    {
      "epoch": 0.6028757274905854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3001,
      "step": 17610
    },
    {
      "epoch": 0.6029099623416638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2894,
      "step": 17611
    },
    {
      "epoch": 0.6029441971927422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2705,
      "step": 17612
    },
    {
      "epoch": 0.6029784320438206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2857,
      "step": 17613
    },
    {
      "epoch": 0.603012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3039,
      "step": 17614
    },
    {
      "epoch": 0.6030469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2853,
      "step": 17615
    },
    {
      "epoch": 0.6030811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.299,
      "step": 17616
    },
    {
      "epoch": 0.6031153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2759,
      "step": 17617
    },
    {
      "epoch": 0.6031496062992125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2855,
      "step": 17618
    },
    {
      "epoch": 0.603183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2856,
      "step": 17619
    },
    {
      "epoch": 0.6032180760013693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2807,
      "step": 17620
    },
    {
      "epoch": 0.6032523108524478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 17621
    },
    {
      "epoch": 0.6032865457035262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2838,
      "step": 17622
    },
    {
      "epoch": 0.6033207805546046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2684,
      "step": 17623
    },
    {
      "epoch": 0.603355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3027,
      "step": 17624
    },
    {
      "epoch": 0.6033892502567614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2746,
      "step": 17625
    },
    {
      "epoch": 0.6034234851078398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2933,
      "step": 17626
    },
    {
      "epoch": 0.6034577199589182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3015,
      "step": 17627
    },
    {
      "epoch": 0.6034919548099966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.263,
      "step": 17628
    },
    {
      "epoch": 0.603526189661075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2816,
      "step": 17629
    },
    {
      "epoch": 0.6035604245121534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.288,
      "step": 17630
    },
    {
      "epoch": 0.6035946593632318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2903,
      "step": 17631
    },
    {
      "epoch": 0.6036288942143102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 17632
    },
    {
      "epoch": 0.6036631290653885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3077,
      "step": 17633
    },
    {
      "epoch": 0.603697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3072,
      "step": 17634
    },
    {
      "epoch": 0.6037315987675453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2875,
      "step": 17635
    },
    {
      "epoch": 0.6037658336186238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2833,
      "step": 17636
    },
    {
      "epoch": 0.6038000684697021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.287,
      "step": 17637
    },
    {
      "epoch": 0.6038343033207806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2825,
      "step": 17638
    },
    {
      "epoch": 0.6038685381718589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3048,
      "step": 17639
    },
    {
      "epoch": 0.6039027730229374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2798,
      "step": 17640
    },
    {
      "epoch": 0.6039370078740157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.296,
      "step": 17641
    },
    {
      "epoch": 0.6039712427250942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2727,
      "step": 17642
    },
    {
      "epoch": 0.6040054775761725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3085,
      "step": 17643
    },
    {
      "epoch": 0.604039712427251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2844,
      "step": 17644
    },
    {
      "epoch": 0.6040739472783293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.295,
      "step": 17645
    },
    {
      "epoch": 0.6041081821294078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2699,
      "step": 17646
    },
    {
      "epoch": 0.6041424169804861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2761,
      "step": 17647
    },
    {
      "epoch": 0.6041766518315645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2845,
      "step": 17648
    },
    {
      "epoch": 0.6042108866826429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2659,
      "step": 17649
    },
    {
      "epoch": 0.6042451215337213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2896,
      "step": 17650
    },
    {
      "epoch": 0.6042793563847997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2915,
      "step": 17651
    },
    {
      "epoch": 0.6043135912358781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2857,
      "step": 17652
    },
    {
      "epoch": 0.6043478260869565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2969,
      "step": 17653
    },
    {
      "epoch": 0.6043820609380349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2788,
      "step": 17654
    },
    {
      "epoch": 0.6044162957891133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.289,
      "step": 17655
    },
    {
      "epoch": 0.6044505306401917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2845,
      "step": 17656
    },
    {
      "epoch": 0.6044847654912701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2674,
      "step": 17657
    },
    {
      "epoch": 0.6045190003423485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.291,
      "step": 17658
    },
    {
      "epoch": 0.604553235193427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2801,
      "step": 17659
    },
    {
      "epoch": 0.6045874700445053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2778,
      "step": 17660
    },
    {
      "epoch": 0.6046217048955838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2911,
      "step": 17661
    },
    {
      "epoch": 0.6046559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3018,
      "step": 17662
    },
    {
      "epoch": 0.6046901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3049,
      "step": 17663
    },
    {
      "epoch": 0.6047244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2805,
      "step": 17664
    },
    {
      "epoch": 0.6047586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2864,
      "step": 17665
    },
    {
      "epoch": 0.6047928791509757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2875,
      "step": 17666
    },
    {
      "epoch": 0.6048271140020541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2975,
      "step": 17667
    },
    {
      "epoch": 0.6048613488531325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2935,
      "step": 17668
    },
    {
      "epoch": 0.6048955837042109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.296,
      "step": 17669
    },
    {
      "epoch": 0.6049298185552893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2639,
      "step": 17670
    },
    {
      "epoch": 0.6049640534063677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.287,
      "step": 17671
    },
    {
      "epoch": 0.6049982882574461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2734,
      "step": 17672
    },
    {
      "epoch": 0.6050325231085245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2913,
      "step": 17673
    },
    {
      "epoch": 0.6050667579596029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2917,
      "step": 17674
    },
    {
      "epoch": 0.6051009928106813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2934,
      "step": 17675
    },
    {
      "epoch": 0.6051352276617596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3009,
      "step": 17676
    },
    {
      "epoch": 0.6051694625128381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2734,
      "step": 17677
    },
    {
      "epoch": 0.6052036973639164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.287,
      "step": 17678
    },
    {
      "epoch": 0.6052379322149949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2824,
      "step": 17679
    },
    {
      "epoch": 0.6052721670660732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.292,
      "step": 17680
    },
    {
      "epoch": 0.6053064019171517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.306,
      "step": 17681
    },
    {
      "epoch": 0.60534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2957,
      "step": 17682
    },
    {
      "epoch": 0.6053748716193085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3002,
      "step": 17683
    },
    {
      "epoch": 0.6054091064703868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2873,
      "step": 17684
    },
    {
      "epoch": 0.6054433413214653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2781,
      "step": 17685
    },
    {
      "epoch": 0.6054775761725436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2961,
      "step": 17686
    },
    {
      "epoch": 0.6055118110236221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2697,
      "step": 17687
    },
    {
      "epoch": 0.6055460458747004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2748,
      "step": 17688
    },
    {
      "epoch": 0.6055802807257789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 17689
    },
    {
      "epoch": 0.6056145155768572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2685,
      "step": 17690
    },
    {
      "epoch": 0.6056487504279356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.281,
      "step": 17691
    },
    {
      "epoch": 0.605682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2794,
      "step": 17692
    },
    {
      "epoch": 0.6057172201300924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2736,
      "step": 17693
    },
    {
      "epoch": 0.6057514549811708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3046,
      "step": 17694
    },
    {
      "epoch": 0.6057856898322492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2659,
      "step": 17695
    },
    {
      "epoch": 0.6058199246833277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2841,
      "step": 17696
    },
    {
      "epoch": 0.605854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3153,
      "step": 17697
    },
    {
      "epoch": 0.6058883943854845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 17698
    },
    {
      "epoch": 0.6059226292365628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.272,
      "step": 17699
    },
    {
      "epoch": 0.6059568640876413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2802,
      "step": 17700
    },
    {
      "epoch": 0.6059910989387196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2774,
      "step": 17701
    },
    {
      "epoch": 0.6060253337897981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2878,
      "step": 17702
    },
    {
      "epoch": 0.6060595686408764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2878,
      "step": 17703
    },
    {
      "epoch": 0.6060938034919549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2952,
      "step": 17704
    },
    {
      "epoch": 0.6061280383430332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2921,
      "step": 17705
    },
    {
      "epoch": 0.6061622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2841,
      "step": 17706
    },
    {
      "epoch": 0.60619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2896,
      "step": 17707
    },
    {
      "epoch": 0.6062307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.283,
      "step": 17708
    },
    {
      "epoch": 0.6062649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2779,
      "step": 17709
    },
    {
      "epoch": 0.6062992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3009,
      "step": 17710
    },
    {
      "epoch": 0.6063334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2811,
      "step": 17711
    },
    {
      "epoch": 0.606367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2931,
      "step": 17712
    },
    {
      "epoch": 0.6064019171516604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2725,
      "step": 17713
    },
    {
      "epoch": 0.6064361520027388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2833,
      "step": 17714
    },
    {
      "epoch": 0.6064703868538172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2953,
      "step": 17715
    },
    {
      "epoch": 0.6065046217048956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2794,
      "step": 17716
    },
    {
      "epoch": 0.606538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2789,
      "step": 17717
    },
    {
      "epoch": 0.6065730914070524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2779,
      "step": 17718
    },
    {
      "epoch": 0.6066073262581307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2839,
      "step": 17719
    },
    {
      "epoch": 0.6066415611092092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2927,
      "step": 17720
    },
    {
      "epoch": 0.6066757959602875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.273,
      "step": 17721
    },
    {
      "epoch": 0.606710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2986,
      "step": 17722
    },
    {
      "epoch": 0.6067442656624443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2938,
      "step": 17723
    },
    {
      "epoch": 0.6067785005135228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.274,
      "step": 17724
    },
    {
      "epoch": 0.6068127353646011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2825,
      "step": 17725
    },
    {
      "epoch": 0.6068469702156796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2792,
      "step": 17726
    },
    {
      "epoch": 0.6068812050667579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2758,
      "step": 17727
    },
    {
      "epoch": 0.6069154399178364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2805,
      "step": 17728
    },
    {
      "epoch": 0.6069496747689147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2905,
      "step": 17729
    },
    {
      "epoch": 0.6069839096199932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2869,
      "step": 17730
    },
    {
      "epoch": 0.6070181444710715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2682,
      "step": 17731
    },
    {
      "epoch": 0.60705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2808,
      "step": 17732
    },
    {
      "epoch": 0.6070866141732284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2882,
      "step": 17733
    },
    {
      "epoch": 0.6071208490243067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2893,
      "step": 17734
    },
    {
      "epoch": 0.6071550838753852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2915,
      "step": 17735
    },
    {
      "epoch": 0.6071893187264635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2995,
      "step": 17736
    },
    {
      "epoch": 0.607223553577542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2766,
      "step": 17737
    },
    {
      "epoch": 0.6072577884286203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2819,
      "step": 17738
    },
    {
      "epoch": 0.6072920232796988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2798,
      "step": 17739
    },
    {
      "epoch": 0.6073262581307771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2955,
      "step": 17740
    },
    {
      "epoch": 0.6073604929818556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.287,
      "step": 17741
    },
    {
      "epoch": 0.6073947278329339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2678,
      "step": 17742
    },
    {
      "epoch": 0.6074289626840124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2717,
      "step": 17743
    },
    {
      "epoch": 0.6074631975350907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2792,
      "step": 17744
    },
    {
      "epoch": 0.6074974323861692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2803,
      "step": 17745
    },
    {
      "epoch": 0.6075316672372475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2932,
      "step": 17746
    },
    {
      "epoch": 0.607565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.291,
      "step": 17747
    },
    {
      "epoch": 0.6076001369394043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.305,
      "step": 17748
    },
    {
      "epoch": 0.6076343717904827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2883,
      "step": 17749
    },
    {
      "epoch": 0.6076686066415611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2769,
      "step": 17750
    },
    {
      "epoch": 0.6077028414926395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2771,
      "step": 17751
    },
    {
      "epoch": 0.6077370763437179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.272,
      "step": 17752
    },
    {
      "epoch": 0.6077713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 17753
    },
    {
      "epoch": 0.6078055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2892,
      "step": 17754
    },
    {
      "epoch": 0.6078397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2891,
      "step": 17755
    },
    {
      "epoch": 0.6078740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2823,
      "step": 17756
    },
    {
      "epoch": 0.6079082505991099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2848,
      "step": 17757
    },
    {
      "epoch": 0.6079424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2803,
      "step": 17758
    },
    {
      "epoch": 0.6079767203012667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2727,
      "step": 17759
    },
    {
      "epoch": 0.608010955152345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2891,
      "step": 17760
    },
    {
      "epoch": 0.6080451900034235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2607,
      "step": 17761
    },
    {
      "epoch": 0.6080794248545018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2731,
      "step": 17762
    },
    {
      "epoch": 0.6081136597055803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2733,
      "step": 17763
    },
    {
      "epoch": 0.6081478945566586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2826,
      "step": 17764
    },
    {
      "epoch": 0.6081821294077371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2691,
      "step": 17765
    },
    {
      "epoch": 0.6082163642588154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2832,
      "step": 17766
    },
    {
      "epoch": 0.6082505991098939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2753,
      "step": 17767
    },
    {
      "epoch": 0.6082848339609722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2974,
      "step": 17768
    },
    {
      "epoch": 0.6083190688120507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2736,
      "step": 17769
    },
    {
      "epoch": 0.6083533036631291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2905,
      "step": 17770
    },
    {
      "epoch": 0.6083875385142075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.286,
      "step": 17771
    },
    {
      "epoch": 0.6084217733652859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2609,
      "step": 17772
    },
    {
      "epoch": 0.6084560082163643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2981,
      "step": 17773
    },
    {
      "epoch": 0.6084902430674427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2811,
      "step": 17774
    },
    {
      "epoch": 0.608524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2856,
      "step": 17775
    },
    {
      "epoch": 0.6085587127695995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2815,
      "step": 17776
    },
    {
      "epoch": 0.6085929476206778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2817,
      "step": 17777
    },
    {
      "epoch": 0.6086271824717563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2939,
      "step": 17778
    },
    {
      "epoch": 0.6086614173228346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2854,
      "step": 17779
    },
    {
      "epoch": 0.6086956521739131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.305,
      "step": 17780
    },
    {
      "epoch": 0.6087298870249914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2912,
      "step": 17781
    },
    {
      "epoch": 0.6087641218760699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2713,
      "step": 17782
    },
    {
      "epoch": 0.6087983567271482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2937,
      "step": 17783
    },
    {
      "epoch": 0.6088325915782267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2764,
      "step": 17784
    },
    {
      "epoch": 0.608866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2851,
      "step": 17785
    },
    {
      "epoch": 0.6089010612803835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.311,
      "step": 17786
    },
    {
      "epoch": 0.6089352961314618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2784,
      "step": 17787
    },
    {
      "epoch": 0.6089695309825403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2885,
      "step": 17788
    },
    {
      "epoch": 0.6090037658336186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2874,
      "step": 17789
    },
    {
      "epoch": 0.609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2749,
      "step": 17790
    },
    {
      "epoch": 0.6090722355357754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.293,
      "step": 17791
    },
    {
      "epoch": 0.6091064703868538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2973,
      "step": 17792
    },
    {
      "epoch": 0.6091407052379322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2952,
      "step": 17793
    },
    {
      "epoch": 0.6091749400890106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2856,
      "step": 17794
    },
    {
      "epoch": 0.609209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2723,
      "step": 17795
    },
    {
      "epoch": 0.6092434097911674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2865,
      "step": 17796
    },
    {
      "epoch": 0.6092776446422458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2941,
      "step": 17797
    },
    {
      "epoch": 0.6093118794933242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2933,
      "step": 17798
    },
    {
      "epoch": 0.6093461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3051,
      "step": 17799
    },
    {
      "epoch": 0.609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.299,
      "step": 17800
    },
    {
      "epoch": 0.6094145840465593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2842,
      "step": 17801
    },
    {
      "epoch": 0.6094488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2773,
      "step": 17802
    },
    {
      "epoch": 0.6094830537487161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2906,
      "step": 17803
    },
    {
      "epoch": 0.6095172885997946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.29,
      "step": 17804
    },
    {
      "epoch": 0.6095515234508729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2832,
      "step": 17805
    },
    {
      "epoch": 0.6095857583019514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2522,
      "step": 17806
    },
    {
      "epoch": 0.6096199931530298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2981,
      "step": 17807
    },
    {
      "epoch": 0.6096542280041082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2817,
      "step": 17808
    },
    {
      "epoch": 0.6096884628551866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2692,
      "step": 17809
    },
    {
      "epoch": 0.609722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2885,
      "step": 17810
    },
    {
      "epoch": 0.6097569325573434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2812,
      "step": 17811
    },
    {
      "epoch": 0.6097911674084218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2728,
      "step": 17812
    },
    {
      "epoch": 0.6098254022595002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2758,
      "step": 17813
    },
    {
      "epoch": 0.6098596371105786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2831,
      "step": 17814
    },
    {
      "epoch": 0.609893871961657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2721,
      "step": 17815
    },
    {
      "epoch": 0.6099281068127353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2841,
      "step": 17816
    },
    {
      "epoch": 0.6099623416638138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2663,
      "step": 17817
    },
    {
      "epoch": 0.6099965765148921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2846,
      "step": 17818
    },
    {
      "epoch": 0.6100308113659706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2812,
      "step": 17819
    },
    {
      "epoch": 0.6100650462170489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2807,
      "step": 17820
    },
    {
      "epoch": 0.6100992810681274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2815,
      "step": 17821
    },
    {
      "epoch": 0.6101335159192057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2962,
      "step": 17822
    },
    {
      "epoch": 0.6101677507702842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2679,
      "step": 17823
    },
    {
      "epoch": 0.6102019856213625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2869,
      "step": 17824
    },
    {
      "epoch": 0.610236220472441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2875,
      "step": 17825
    },
    {
      "epoch": 0.6102704553235193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2918,
      "step": 17826
    },
    {
      "epoch": 0.6103046901745978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2639,
      "step": 17827
    },
    {
      "epoch": 0.6103389250256761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2939,
      "step": 17828
    },
    {
      "epoch": 0.6103731598767546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2911,
      "step": 17829
    },
    {
      "epoch": 0.6104073947278329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2778,
      "step": 17830
    },
    {
      "epoch": 0.6104416295789113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2868,
      "step": 17831
    },
    {
      "epoch": 0.6104758644299897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2894,
      "step": 17832
    },
    {
      "epoch": 0.6105100992810681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2826,
      "step": 17833
    },
    {
      "epoch": 0.6105443341321465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2892,
      "step": 17834
    },
    {
      "epoch": 0.6105785689832249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2761,
      "step": 17835
    },
    {
      "epoch": 0.6106128038343033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2735,
      "step": 17836
    },
    {
      "epoch": 0.6106470386853817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2832,
      "step": 17837
    },
    {
      "epoch": 0.6106812735364601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2769,
      "step": 17838
    },
    {
      "epoch": 0.6107155083875385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2719,
      "step": 17839
    },
    {
      "epoch": 0.6107497432386169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2775,
      "step": 17840
    },
    {
      "epoch": 0.6107839780896953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2874,
      "step": 17841
    },
    {
      "epoch": 0.6108182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2969,
      "step": 17842
    },
    {
      "epoch": 0.6108524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2888,
      "step": 17843
    },
    {
      "epoch": 0.6108866826429306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2758,
      "step": 17844
    },
    {
      "epoch": 0.6109209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2766,
      "step": 17845
    },
    {
      "epoch": 0.6109551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2889,
      "step": 17846
    },
    {
      "epoch": 0.6109893871961657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2741,
      "step": 17847
    },
    {
      "epoch": 0.6110236220472441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2685,
      "step": 17848
    },
    {
      "epoch": 0.6110578568983225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2646,
      "step": 17849
    },
    {
      "epoch": 0.6110920917494009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2824,
      "step": 17850
    },
    {
      "epoch": 0.6111263266004793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2802,
      "step": 17851
    },
    {
      "epoch": 0.6111605614515577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2892,
      "step": 17852
    },
    {
      "epoch": 0.6111947963026361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2974,
      "step": 17853
    },
    {
      "epoch": 0.6112290311537145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2625,
      "step": 17854
    },
    {
      "epoch": 0.6112632660047929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2725,
      "step": 17855
    },
    {
      "epoch": 0.6112975008558713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2741,
      "step": 17856
    },
    {
      "epoch": 0.6113317357069497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2872,
      "step": 17857
    },
    {
      "epoch": 0.6113659705580281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2792,
      "step": 17858
    },
    {
      "epoch": 0.6114002054091064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2708,
      "step": 17859
    },
    {
      "epoch": 0.6114344402601849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2857,
      "step": 17860
    },
    {
      "epoch": 0.6114686751112632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2725,
      "step": 17861
    },
    {
      "epoch": 0.6115029099623417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2848,
      "step": 17862
    },
    {
      "epoch": 0.61153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.279,
      "step": 17863
    },
    {
      "epoch": 0.6115713796644985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2759,
      "step": 17864
    },
    {
      "epoch": 0.6116056145155768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2689,
      "step": 17865
    },
    {
      "epoch": 0.6116398493666553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.273,
      "step": 17866
    },
    {
      "epoch": 0.6116740842177336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2879,
      "step": 17867
    },
    {
      "epoch": 0.6117083190688121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3001,
      "step": 17868
    },
    {
      "epoch": 0.6117425539198904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2668,
      "step": 17869
    },
    {
      "epoch": 0.6117767887709689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2782,
      "step": 17870
    },
    {
      "epoch": 0.6118110236220472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2898,
      "step": 17871
    },
    {
      "epoch": 0.6118452584731257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2653,
      "step": 17872
    },
    {
      "epoch": 0.611879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2777,
      "step": 17873
    },
    {
      "epoch": 0.6119137281752824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2674,
      "step": 17874
    },
    {
      "epoch": 0.6119479630263608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2951,
      "step": 17875
    },
    {
      "epoch": 0.6119821978774392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3016,
      "step": 17876
    },
    {
      "epoch": 0.6120164327285176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2778,
      "step": 17877
    },
    {
      "epoch": 0.612050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2933,
      "step": 17878
    },
    {
      "epoch": 0.6120849024306744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2905,
      "step": 17879
    },
    {
      "epoch": 0.6121191372817528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2901,
      "step": 17880
    },
    {
      "epoch": 0.6121533721328313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2739,
      "step": 17881
    },
    {
      "epoch": 0.6121876069839096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2818,
      "step": 17882
    },
    {
      "epoch": 0.6122218418349881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2869,
      "step": 17883
    },
    {
      "epoch": 0.6122560766860664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3096,
      "step": 17884
    },
    {
      "epoch": 0.6122903115371449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2959,
      "step": 17885
    },
    {
      "epoch": 0.6123245463882232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2667,
      "step": 17886
    },
    {
      "epoch": 0.6123587812393017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2811,
      "step": 17887
    },
    {
      "epoch": 0.61239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2767,
      "step": 17888
    },
    {
      "epoch": 0.6124272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2681,
      "step": 17889
    },
    {
      "epoch": 0.6124614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2805,
      "step": 17890
    },
    {
      "epoch": 0.6124957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2756,
      "step": 17891
    },
    {
      "epoch": 0.6125299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2751,
      "step": 17892
    },
    {
      "epoch": 0.612564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2845,
      "step": 17893
    },
    {
      "epoch": 0.6125984251968504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2675,
      "step": 17894
    },
    {
      "epoch": 0.6126326600479288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2792,
      "step": 17895
    },
    {
      "epoch": 0.6126668948990072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2882,
      "step": 17896
    },
    {
      "epoch": 0.6127011297500856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3015,
      "step": 17897
    },
    {
      "epoch": 0.612735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2819,
      "step": 17898
    },
    {
      "epoch": 0.6127695994522424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2736,
      "step": 17899
    },
    {
      "epoch": 0.6128038343033207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2844,
      "step": 17900
    },
    {
      "epoch": 0.6128380691543992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2689,
      "step": 17901
    },
    {
      "epoch": 0.6128723040054775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2888,
      "step": 17902
    },
    {
      "epoch": 0.612906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2903,
      "step": 17903
    },
    {
      "epoch": 0.6129407737076343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3041,
      "step": 17904
    },
    {
      "epoch": 0.6129750085587128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2868,
      "step": 17905
    },
    {
      "epoch": 0.6130092434097911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2794,
      "step": 17906
    },
    {
      "epoch": 0.6130434782608696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2758,
      "step": 17907
    },
    {
      "epoch": 0.6130777131119479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2865,
      "step": 17908
    },
    {
      "epoch": 0.6131119479630264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2659,
      "step": 17909
    },
    {
      "epoch": 0.6131461828141047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2767,
      "step": 17910
    },
    {
      "epoch": 0.6131804176651832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.29,
      "step": 17911
    },
    {
      "epoch": 0.6132146525162615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2941,
      "step": 17912
    },
    {
      "epoch": 0.61324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2845,
      "step": 17913
    },
    {
      "epoch": 0.6132831222184183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2823,
      "step": 17914
    },
    {
      "epoch": 0.6133173570694967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2813,
      "step": 17915
    },
    {
      "epoch": 0.6133515919205752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2737,
      "step": 17916
    },
    {
      "epoch": 0.6133858267716535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.29,
      "step": 17917
    },
    {
      "epoch": 0.613420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2814,
      "step": 17918
    },
    {
      "epoch": 0.6134542964738103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2617,
      "step": 17919
    },
    {
      "epoch": 0.6134885313248888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2808,
      "step": 17920
    },
    {
      "epoch": 0.6135227661759671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2718,
      "step": 17921
    },
    {
      "epoch": 0.6135570010270456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2855,
      "step": 17922
    },
    {
      "epoch": 0.6135912358781239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.273,
      "step": 17923
    },
    {
      "epoch": 0.6136254707292024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.287,
      "step": 17924
    },
    {
      "epoch": 0.6136597055802807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2792,
      "step": 17925
    },
    {
      "epoch": 0.6136939404313592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2721,
      "step": 17926
    },
    {
      "epoch": 0.6137281752824375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2786,
      "step": 17927
    },
    {
      "epoch": 0.613762410133516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2672,
      "step": 17928
    },
    {
      "epoch": 0.6137966449845943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2681,
      "step": 17929
    },
    {
      "epoch": 0.6138308798356727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2685,
      "step": 17930
    },
    {
      "epoch": 0.6138651146867511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2721,
      "step": 17931
    },
    {
      "epoch": 0.6138993495378295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2764,
      "step": 17932
    },
    {
      "epoch": 0.6139335843889079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2871,
      "step": 17933
    },
    {
      "epoch": 0.6139678192399863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2711,
      "step": 17934
    },
    {
      "epoch": 0.6140020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2836,
      "step": 17935
    },
    {
      "epoch": 0.6140362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2755,
      "step": 17936
    },
    {
      "epoch": 0.6140705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 17937
    },
    {
      "epoch": 0.6141047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2771,
      "step": 17938
    },
    {
      "epoch": 0.6141389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2858,
      "step": 17939
    },
    {
      "epoch": 0.6141732283464567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2804,
      "step": 17940
    },
    {
      "epoch": 0.614207463197535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2851,
      "step": 17941
    },
    {
      "epoch": 0.6142416980486135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3022,
      "step": 17942
    },
    {
      "epoch": 0.6142759328996918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2648,
      "step": 17943
    },
    {
      "epoch": 0.6143101677507703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2775,
      "step": 17944
    },
    {
      "epoch": 0.6143444026018486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2751,
      "step": 17945
    },
    {
      "epoch": 0.6143786374529271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2646,
      "step": 17946
    },
    {
      "epoch": 0.6144128723040054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3002,
      "step": 17947
    },
    {
      "epoch": 0.6144471071550839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.296,
      "step": 17948
    },
    {
      "epoch": 0.6144813420061622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2813,
      "step": 17949
    },
    {
      "epoch": 0.6145155768572407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2904,
      "step": 17950
    },
    {
      "epoch": 0.614549811708319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2698,
      "step": 17951
    },
    {
      "epoch": 0.6145840465593975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2876,
      "step": 17952
    },
    {
      "epoch": 0.6146182814104759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2878,
      "step": 17953
    },
    {
      "epoch": 0.6146525162615543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2914,
      "step": 17954
    },
    {
      "epoch": 0.6146867511126327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2844,
      "step": 17955
    },
    {
      "epoch": 0.614720985963711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2657,
      "step": 17956
    },
    {
      "epoch": 0.6147552208147895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2856,
      "step": 17957
    },
    {
      "epoch": 0.6147894556658678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2659,
      "step": 17958
    },
    {
      "epoch": 0.6148236905169463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2712,
      "step": 17959
    },
    {
      "epoch": 0.6148579253680246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2788,
      "step": 17960
    },
    {
      "epoch": 0.6148921602191031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2972,
      "step": 17961
    },
    {
      "epoch": 0.6149263950701814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2651,
      "step": 17962
    },
    {
      "epoch": 0.6149606299212599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2795,
      "step": 17963
    },
    {
      "epoch": 0.6149948647723382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2791,
      "step": 17964
    },
    {
      "epoch": 0.6150290996234167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2712,
      "step": 17965
    },
    {
      "epoch": 0.615063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.282,
      "step": 17966
    },
    {
      "epoch": 0.6150975693255735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2639,
      "step": 17967
    },
    {
      "epoch": 0.6151318041766518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2842,
      "step": 17968
    },
    {
      "epoch": 0.6151660390277303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2803,
      "step": 17969
    },
    {
      "epoch": 0.6152002738788086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2766,
      "step": 17970
    },
    {
      "epoch": 0.615234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.3058,
      "step": 17971
    },
    {
      "epoch": 0.6152687435809654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2753,
      "step": 17972
    },
    {
      "epoch": 0.6153029784320438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2786,
      "step": 17973
    },
    {
      "epoch": 0.6153372132831222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.272,
      "step": 17974
    },
    {
      "epoch": 0.6153714481342006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2688,
      "step": 17975
    },
    {
      "epoch": 0.615405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2911,
      "step": 17976
    },
    {
      "epoch": 0.6154399178363574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2867,
      "step": 17977
    },
    {
      "epoch": 0.6154741526874358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2787,
      "step": 17978
    },
    {
      "epoch": 0.6155083875385142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2924,
      "step": 17979
    },
    {
      "epoch": 0.6155426223895926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2592,
      "step": 17980
    },
    {
      "epoch": 0.615576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2789,
      "step": 17981
    },
    {
      "epoch": 0.6156110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2698,
      "step": 17982
    },
    {
      "epoch": 0.6156453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2731,
      "step": 17983
    },
    {
      "epoch": 0.6156795617939061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.291,
      "step": 17984
    },
    {
      "epoch": 0.6157137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2749,
      "step": 17985
    },
    {
      "epoch": 0.6157480314960629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2531,
      "step": 17986
    },
    {
      "epoch": 0.6157822663471414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2843,
      "step": 17987
    },
    {
      "epoch": 0.6158165011982197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.291,
      "step": 17988
    },
    {
      "epoch": 0.6158507360492982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2802,
      "step": 17989
    },
    {
      "epoch": 0.6158849709003766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2598,
      "step": 17990
    },
    {
      "epoch": 0.615919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2715,
      "step": 17991
    },
    {
      "epoch": 0.6159534406025334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2824,
      "step": 17992
    },
    {
      "epoch": 0.6159876754536118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2797,
      "step": 17993
    },
    {
      "epoch": 0.6160219103046902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2773,
      "step": 17994
    },
    {
      "epoch": 0.6160561451557686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.279,
      "step": 17995
    },
    {
      "epoch": 0.616090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2643,
      "step": 17996
    },
    {
      "epoch": 0.6161246148579254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2845,
      "step": 17997
    },
    {
      "epoch": 0.6161588497090038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2839,
      "step": 17998
    },
    {
      "epoch": 0.6161930845600821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.271,
      "step": 17999
    },
    {
      "epoch": 0.6162273194111606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2885,
      "step": 18000
    },
    {
      "epoch": 0.6162615542622389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2822,
      "step": 18001
    },
    {
      "epoch": 0.6162957891133174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.274,
      "step": 18002
    },
    {
      "epoch": 0.6163300239643957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 18003
    },
    {
      "epoch": 0.6163642588154742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2603,
      "step": 18004
    },
    {
      "epoch": 0.6163984936665525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2628,
      "step": 18005
    },
    {
      "epoch": 0.616432728517631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2657,
      "step": 18006
    },
    {
      "epoch": 0.6164669633687093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2919,
      "step": 18007
    },
    {
      "epoch": 0.6165011982197878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2692,
      "step": 18008
    },
    {
      "epoch": 0.6165354330708661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2909,
      "step": 18009
    },
    {
      "epoch": 0.6165696679219446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2712,
      "step": 18010
    },
    {
      "epoch": 0.6166039027730229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.282,
      "step": 18011
    },
    {
      "epoch": 0.6166381376241014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2811,
      "step": 18012
    },
    {
      "epoch": 0.6166723724751797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2806,
      "step": 18013
    },
    {
      "epoch": 0.6167066073262581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2795,
      "step": 18014
    },
    {
      "epoch": 0.6167408421773365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2819,
      "step": 18015
    },
    {
      "epoch": 0.6167750770284149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2706,
      "step": 18016
    },
    {
      "epoch": 0.6168093118794933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2689,
      "step": 18017
    },
    {
      "epoch": 0.6168435467305717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2757,
      "step": 18018
    },
    {
      "epoch": 0.6168777815816501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2791,
      "step": 18019
    },
    {
      "epoch": 0.6169120164327285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.273,
      "step": 18020
    },
    {
      "epoch": 0.6169462512838069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2807,
      "step": 18021
    },
    {
      "epoch": 0.6169804861348853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2883,
      "step": 18022
    },
    {
      "epoch": 0.6170147209859637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2773,
      "step": 18023
    },
    {
      "epoch": 0.6170489558370421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2739,
      "step": 18024
    },
    {
      "epoch": 0.6170831906881205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2714,
      "step": 18025
    },
    {
      "epoch": 0.6171174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2974,
      "step": 18026
    },
    {
      "epoch": 0.6171516603902774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2692,
      "step": 18027
    },
    {
      "epoch": 0.6171858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2806,
      "step": 18028
    },
    {
      "epoch": 0.6172201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2619,
      "step": 18029
    },
    {
      "epoch": 0.6172543649435125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2724,
      "step": 18030
    },
    {
      "epoch": 0.6172885997945909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2737,
      "step": 18031
    },
    {
      "epoch": 0.6173228346456693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2801,
      "step": 18032
    },
    {
      "epoch": 0.6173570694967477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2869,
      "step": 18033
    },
    {
      "epoch": 0.6173913043478261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2615,
      "step": 18034
    },
    {
      "epoch": 0.6174255391989045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2786,
      "step": 18035
    },
    {
      "epoch": 0.6174597740499829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2809,
      "step": 18036
    },
    {
      "epoch": 0.6174940089010613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2965,
      "step": 18037
    },
    {
      "epoch": 0.6175282437521397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2905,
      "step": 18038
    },
    {
      "epoch": 0.6175624786032181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.279,
      "step": 18039
    },
    {
      "epoch": 0.6175967134542965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2842,
      "step": 18040
    },
    {
      "epoch": 0.6176309483053749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2604,
      "step": 18041
    },
    {
      "epoch": 0.6176651831564532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2786,
      "step": 18042
    },
    {
      "epoch": 0.6176994180075317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.275,
      "step": 18043
    },
    {
      "epoch": 0.61773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2853,
      "step": 18044
    },
    {
      "epoch": 0.6177678877096885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2864,
      "step": 18045
    },
    {
      "epoch": 0.6178021225607668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.275,
      "step": 18046
    },
    {
      "epoch": 0.6178363574118453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2791,
      "step": 18047
    },
    {
      "epoch": 0.6178705922629236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2939,
      "step": 18048
    },
    {
      "epoch": 0.6179048271140021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2777,
      "step": 18049
    },
    {
      "epoch": 0.6179390619650804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2762,
      "step": 18050
    },
    {
      "epoch": 0.6179732968161589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2756,
      "step": 18051
    },
    {
      "epoch": 0.6180075316672372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2821,
      "step": 18052
    },
    {
      "epoch": 0.6180417665183157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.275,
      "step": 18053
    },
    {
      "epoch": 0.618076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2664,
      "step": 18054
    },
    {
      "epoch": 0.6181102362204725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2715,
      "step": 18055
    },
    {
      "epoch": 0.6181444710715508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2814,
      "step": 18056
    },
    {
      "epoch": 0.6181787059226292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2804,
      "step": 18057
    },
    {
      "epoch": 0.6182129407737076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.297,
      "step": 18058
    },
    {
      "epoch": 0.618247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2739,
      "step": 18059
    },
    {
      "epoch": 0.6182814104758644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2733,
      "step": 18060
    },
    {
      "epoch": 0.6183156453269428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2958,
      "step": 18061
    },
    {
      "epoch": 0.6183498801780212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.281,
      "step": 18062
    },
    {
      "epoch": 0.6183841150290996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2605,
      "step": 18063
    },
    {
      "epoch": 0.6184183498801781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2529,
      "step": 18064
    },
    {
      "epoch": 0.6184525847312564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2769,
      "step": 18065
    },
    {
      "epoch": 0.6184868195823349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2831,
      "step": 18066
    },
    {
      "epoch": 0.6185210544334132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2556,
      "step": 18067
    },
    {
      "epoch": 0.6185552892844917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2714,
      "step": 18068
    },
    {
      "epoch": 0.61858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2583,
      "step": 18069
    },
    {
      "epoch": 0.6186237589866485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2715,
      "step": 18070
    },
    {
      "epoch": 0.6186579938377268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2725,
      "step": 18071
    },
    {
      "epoch": 0.6186922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2673,
      "step": 18072
    },
    {
      "epoch": 0.6187264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2973,
      "step": 18073
    },
    {
      "epoch": 0.618760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2815,
      "step": 18074
    },
    {
      "epoch": 0.6187949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2876,
      "step": 18075
    },
    {
      "epoch": 0.6188291680931188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2845,
      "step": 18076
    },
    {
      "epoch": 0.6188634029441972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2831,
      "step": 18077
    },
    {
      "epoch": 0.6188976377952756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2661,
      "step": 18078
    },
    {
      "epoch": 0.618931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2791,
      "step": 18079
    },
    {
      "epoch": 0.6189661074974324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2773,
      "step": 18080
    },
    {
      "epoch": 0.6190003423485108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2806,
      "step": 18081
    },
    {
      "epoch": 0.6190345771995892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.26,
      "step": 18082
    },
    {
      "epoch": 0.6190688120506675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.268,
      "step": 18083
    },
    {
      "epoch": 0.619103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2628,
      "step": 18084
    },
    {
      "epoch": 0.6191372817528243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2711,
      "step": 18085
    },
    {
      "epoch": 0.6191715166039028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2823,
      "step": 18086
    },
    {
      "epoch": 0.6192057514549811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2588,
      "step": 18087
    },
    {
      "epoch": 0.6192399863060596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2783,
      "step": 18088
    },
    {
      "epoch": 0.6192742211571379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2643,
      "step": 18089
    },
    {
      "epoch": 0.6193084560082164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2848,
      "step": 18090
    },
    {
      "epoch": 0.6193426908592947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2731,
      "step": 18091
    },
    {
      "epoch": 0.6193769257103732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.269,
      "step": 18092
    },
    {
      "epoch": 0.6194111605614515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2735,
      "step": 18093
    },
    {
      "epoch": 0.61944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2802,
      "step": 18094
    },
    {
      "epoch": 0.6194796302636083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2752,
      "step": 18095
    },
    {
      "epoch": 0.6195138651146868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2667,
      "step": 18096
    },
    {
      "epoch": 0.6195480999657651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2799,
      "step": 18097
    },
    {
      "epoch": 0.6195823348168435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.28,
      "step": 18098
    },
    {
      "epoch": 0.6196165696679219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2723,
      "step": 18099
    },
    {
      "epoch": 0.6196508045190003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2553,
      "step": 18100
    },
    {
      "epoch": 0.6196850393700788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2745,
      "step": 18101
    },
    {
      "epoch": 0.6197192742211571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2716,
      "step": 18102
    },
    {
      "epoch": 0.6197535090722356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2755,
      "step": 18103
    },
    {
      "epoch": 0.6197877439233139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2741,
      "step": 18104
    },
    {
      "epoch": 0.6198219787743924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2704,
      "step": 18105
    },
    {
      "epoch": 0.6198562136254707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2695,
      "step": 18106
    },
    {
      "epoch": 0.6198904484765492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2802,
      "step": 18107
    },
    {
      "epoch": 0.6199246833276275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2596,
      "step": 18108
    },
    {
      "epoch": 0.619958918178706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2851,
      "step": 18109
    },
    {
      "epoch": 0.6199931530297843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2884,
      "step": 18110
    },
    {
      "epoch": 0.6200273878808628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2654,
      "step": 18111
    },
    {
      "epoch": 0.6200616227319411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2583,
      "step": 18112
    },
    {
      "epoch": 0.6200958575830195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2845,
      "step": 18113
    },
    {
      "epoch": 0.6201300924340979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2542,
      "step": 18114
    },
    {
      "epoch": 0.6201643272851763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.271,
      "step": 18115
    },
    {
      "epoch": 0.6201985621362547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2744,
      "step": 18116
    },
    {
      "epoch": 0.6202327969873331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.285,
      "step": 18117
    },
    {
      "epoch": 0.6202670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2681,
      "step": 18118
    },
    {
      "epoch": 0.6203012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2856,
      "step": 18119
    },
    {
      "epoch": 0.6203355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2643,
      "step": 18120
    },
    {
      "epoch": 0.6203697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2847,
      "step": 18121
    },
    {
      "epoch": 0.6204039712427251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2946,
      "step": 18122
    },
    {
      "epoch": 0.6204382060938035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2734,
      "step": 18123
    },
    {
      "epoch": 0.6204724409448819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2724,
      "step": 18124
    },
    {
      "epoch": 0.6205066757959603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2781,
      "step": 18125
    },
    {
      "epoch": 0.6205409106470386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2617,
      "step": 18126
    },
    {
      "epoch": 0.6205751454981171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2664,
      "step": 18127
    },
    {
      "epoch": 0.6206093803491954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2783,
      "step": 18128
    },
    {
      "epoch": 0.6206436152002739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2516,
      "step": 18129
    },
    {
      "epoch": 0.6206778500513522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2728,
      "step": 18130
    },
    {
      "epoch": 0.6207120849024307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2661,
      "step": 18131
    },
    {
      "epoch": 0.620746319753509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2694,
      "step": 18132
    },
    {
      "epoch": 0.6207805546045875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2573,
      "step": 18133
    },
    {
      "epoch": 0.6208147894556658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2802,
      "step": 18134
    },
    {
      "epoch": 0.6208490243067443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2825,
      "step": 18135
    },
    {
      "epoch": 0.6208832591578226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2819,
      "step": 18136
    },
    {
      "epoch": 0.6209174940089011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2663,
      "step": 18137
    },
    {
      "epoch": 0.6209517288599795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2775,
      "step": 18138
    },
    {
      "epoch": 0.6209859637110579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2643,
      "step": 18139
    },
    {
      "epoch": 0.6210201985621363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2842,
      "step": 18140
    },
    {
      "epoch": 0.6210544334132146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2771,
      "step": 18141
    },
    {
      "epoch": 0.6210886682642931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2744,
      "step": 18142
    },
    {
      "epoch": 0.6211229031153714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2984,
      "step": 18143
    },
    {
      "epoch": 0.6211571379664499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2751,
      "step": 18144
    },
    {
      "epoch": 0.6211913728175282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2754,
      "step": 18145
    },
    {
      "epoch": 0.6212256076686067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2797,
      "step": 18146
    },
    {
      "epoch": 0.621259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2788,
      "step": 18147
    },
    {
      "epoch": 0.6212940773707635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2699,
      "step": 18148
    },
    {
      "epoch": 0.6213283122218418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2685,
      "step": 18149
    },
    {
      "epoch": 0.6213625470729203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2711,
      "step": 18150
    },
    {
      "epoch": 0.6213967819239986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2849,
      "step": 18151
    },
    {
      "epoch": 0.6214310167750771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2825,
      "step": 18152
    },
    {
      "epoch": 0.6214652516261554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2666,
      "step": 18153
    },
    {
      "epoch": 0.6214994864772339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2719,
      "step": 18154
    },
    {
      "epoch": 0.6215337213283122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2703,
      "step": 18155
    },
    {
      "epoch": 0.6215679561793906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2709,
      "step": 18156
    },
    {
      "epoch": 0.621602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2948,
      "step": 18157
    },
    {
      "epoch": 0.6216364258815474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.26,
      "step": 18158
    },
    {
      "epoch": 0.6216706607326258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2658,
      "step": 18159
    },
    {
      "epoch": 0.6217048955837042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.28,
      "step": 18160
    },
    {
      "epoch": 0.6217391304347826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2671,
      "step": 18161
    },
    {
      "epoch": 0.621773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2605,
      "step": 18162
    },
    {
      "epoch": 0.6218076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2888,
      "step": 18163
    },
    {
      "epoch": 0.6218418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.27,
      "step": 18164
    },
    {
      "epoch": 0.6218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2689,
      "step": 18165
    },
    {
      "epoch": 0.6219103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2683,
      "step": 18166
    },
    {
      "epoch": 0.621944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.277,
      "step": 18167
    },
    {
      "epoch": 0.6219787743923314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2602,
      "step": 18168
    },
    {
      "epoch": 0.6220130092434097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2606,
      "step": 18169
    },
    {
      "epoch": 0.6220472440944882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2701,
      "step": 18170
    },
    {
      "epoch": 0.6220814789455665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2669,
      "step": 18171
    },
    {
      "epoch": 0.622115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2734,
      "step": 18172
    },
    {
      "epoch": 0.6221499486477233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2789,
      "step": 18173
    },
    {
      "epoch": 0.6221841834988018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2748,
      "step": 18174
    },
    {
      "epoch": 0.6222184183498802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2836,
      "step": 18175
    },
    {
      "epoch": 0.6222526532009586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.267,
      "step": 18176
    },
    {
      "epoch": 0.622286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2634,
      "step": 18177
    },
    {
      "epoch": 0.6223211229031154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2719,
      "step": 18178
    },
    {
      "epoch": 0.6223553577541938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2616,
      "step": 18179
    },
    {
      "epoch": 0.6223895926052722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2511,
      "step": 18180
    },
    {
      "epoch": 0.6224238274563506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2781,
      "step": 18181
    },
    {
      "epoch": 0.622458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2846,
      "step": 18182
    },
    {
      "epoch": 0.6224922971585074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.278,
      "step": 18183
    },
    {
      "epoch": 0.6225265320095857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2842,
      "step": 18184
    },
    {
      "epoch": 0.6225607668606642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2699,
      "step": 18185
    },
    {
      "epoch": 0.6225950017117425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2536,
      "step": 18186
    },
    {
      "epoch": 0.622629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2668,
      "step": 18187
    },
    {
      "epoch": 0.6226634714138993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2683,
      "step": 18188
    },
    {
      "epoch": 0.6226977062649778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2676,
      "step": 18189
    },
    {
      "epoch": 0.6227319411160561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2784,
      "step": 18190
    },
    {
      "epoch": 0.6227661759671346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2675,
      "step": 18191
    },
    {
      "epoch": 0.6228004108182129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.275,
      "step": 18192
    },
    {
      "epoch": 0.6228346456692914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2674,
      "step": 18193
    },
    {
      "epoch": 0.6228688805203697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.259,
      "step": 18194
    },
    {
      "epoch": 0.6229031153714482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2509,
      "step": 18195
    },
    {
      "epoch": 0.6229373502225265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2514,
      "step": 18196
    },
    {
      "epoch": 0.622971585073605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2679,
      "step": 18197
    },
    {
      "epoch": 0.6230058199246833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2755,
      "step": 18198
    },
    {
      "epoch": 0.6230400547757617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2858,
      "step": 18199
    },
    {
      "epoch": 0.6230742896268401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2637,
      "step": 18200
    },
    {
      "epoch": 0.6231085244779185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2547,
      "step": 18201
    },
    {
      "epoch": 0.6231427593289969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2667,
      "step": 18202
    },
    {
      "epoch": 0.6231769941800753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2632,
      "step": 18203
    },
    {
      "epoch": 0.6232112290311537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2766,
      "step": 18204
    },
    {
      "epoch": 0.6232454638822321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2728,
      "step": 18205
    },
    {
      "epoch": 0.6232796987333105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.28,
      "step": 18206
    },
    {
      "epoch": 0.6233139335843889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2752,
      "step": 18207
    },
    {
      "epoch": 0.6233481684354673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2754,
      "step": 18208
    },
    {
      "epoch": 0.6233824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.261,
      "step": 18209
    },
    {
      "epoch": 0.623416638137624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2649,
      "step": 18210
    },
    {
      "epoch": 0.6234508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2603,
      "step": 18211
    },
    {
      "epoch": 0.623485107839781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2765,
      "step": 18212
    },
    {
      "epoch": 0.6235193426908593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2772,
      "step": 18213
    },
    {
      "epoch": 0.6235535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2638,
      "step": 18214
    },
    {
      "epoch": 0.6235878123930161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2587,
      "step": 18215
    },
    {
      "epoch": 0.6236220472440945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2791,
      "step": 18216
    },
    {
      "epoch": 0.6236562820951729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2706,
      "step": 18217
    },
    {
      "epoch": 0.6236905169462513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2698,
      "step": 18218
    },
    {
      "epoch": 0.6237247517973297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2762,
      "step": 18219
    },
    {
      "epoch": 0.6237589866484081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2836,
      "step": 18220
    },
    {
      "epoch": 0.6237932214994865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.27,
      "step": 18221
    },
    {
      "epoch": 0.6238274563505649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 18222
    },
    {
      "epoch": 0.6238616912016433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2732,
      "step": 18223
    },
    {
      "epoch": 0.6238959260527217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2695,
      "step": 18224
    },
    {
      "epoch": 0.6239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.273,
      "step": 18225
    },
    {
      "epoch": 0.6239643957548785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2634,
      "step": 18226
    },
    {
      "epoch": 0.6239986306059568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2517,
      "step": 18227
    },
    {
      "epoch": 0.6240328654570353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2769,
      "step": 18228
    },
    {
      "epoch": 0.6240671003081136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.278,
      "step": 18229
    },
    {
      "epoch": 0.6241013351591921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2711,
      "step": 18230
    },
    {
      "epoch": 0.6241355700102704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2654,
      "step": 18231
    },
    {
      "epoch": 0.6241698048613489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2647,
      "step": 18232
    },
    {
      "epoch": 0.6242040397124272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2592,
      "step": 18233
    },
    {
      "epoch": 0.6242382745635057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2789,
      "step": 18234
    },
    {
      "epoch": 0.624272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2705,
      "step": 18235
    },
    {
      "epoch": 0.6243067442656625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2655,
      "step": 18236
    },
    {
      "epoch": 0.6243409791167408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2809,
      "step": 18237
    },
    {
      "epoch": 0.6243752139678193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2741,
      "step": 18238
    },
    {
      "epoch": 0.6244094488188976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2698,
      "step": 18239
    },
    {
      "epoch": 0.624443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2637,
      "step": 18240
    },
    {
      "epoch": 0.6244779185210544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2726,
      "step": 18241
    },
    {
      "epoch": 0.6245121533721328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2832,
      "step": 18242
    },
    {
      "epoch": 0.6245463882232112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2731,
      "step": 18243
    },
    {
      "epoch": 0.6245806230742896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2658,
      "step": 18244
    },
    {
      "epoch": 0.624614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2795,
      "step": 18245
    },
    {
      "epoch": 0.6246490927764464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2715,
      "step": 18246
    },
    {
      "epoch": 0.6246833276275248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2891,
      "step": 18247
    },
    {
      "epoch": 0.6247175624786032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2578,
      "step": 18248
    },
    {
      "epoch": 0.6247517973296817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2639,
      "step": 18249
    },
    {
      "epoch": 0.62478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2764,
      "step": 18250
    },
    {
      "epoch": 0.6248202670318385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2607,
      "step": 18251
    },
    {
      "epoch": 0.6248545018829168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2676,
      "step": 18252
    },
    {
      "epoch": 0.6248887367339953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2594,
      "step": 18253
    },
    {
      "epoch": 0.6249229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2648,
      "step": 18254
    },
    {
      "epoch": 0.624957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2726,
      "step": 18255
    },
    {
      "epoch": 0.6249914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2812,
      "step": 18256
    },
    {
      "epoch": 0.6250256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2853,
      "step": 18257
    },
    {
      "epoch": 0.6250599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2784,
      "step": 18258
    },
    {
      "epoch": 0.6250941458404656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2999,
      "step": 18259
    },
    {
      "epoch": 0.625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2615,
      "step": 18260
    },
    {
      "epoch": 0.6251626155426224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2839,
      "step": 18261
    },
    {
      "epoch": 0.6251968503937008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2688,
      "step": 18262
    },
    {
      "epoch": 0.6252310852447792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2781,
      "step": 18263
    },
    {
      "epoch": 0.6252653200958576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2647,
      "step": 18264
    },
    {
      "epoch": 0.625299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2651,
      "step": 18265
    },
    {
      "epoch": 0.6253337897980144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2911,
      "step": 18266
    },
    {
      "epoch": 0.6253680246490928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2718,
      "step": 18267
    },
    {
      "epoch": 0.6254022595001711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2825,
      "step": 18268
    },
    {
      "epoch": 0.6254364943512496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2654,
      "step": 18269
    },
    {
      "epoch": 0.6254707292023279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2825,
      "step": 18270
    },
    {
      "epoch": 0.6255049640534064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2731,
      "step": 18271
    },
    {
      "epoch": 0.6255391989044847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2659,
      "step": 18272
    },
    {
      "epoch": 0.6255734337555632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2683,
      "step": 18273
    },
    {
      "epoch": 0.6256076686066415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2657,
      "step": 18274
    },
    {
      "epoch": 0.62564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2667,
      "step": 18275
    },
    {
      "epoch": 0.6256761383087983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2729,
      "step": 18276
    },
    {
      "epoch": 0.6257103731598768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2864,
      "step": 18277
    },
    {
      "epoch": 0.6257446080109551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2646,
      "step": 18278
    },
    {
      "epoch": 0.6257788428620336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2571,
      "step": 18279
    },
    {
      "epoch": 0.6258130777131119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2708,
      "step": 18280
    },
    {
      "epoch": 0.6258473125641904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2701,
      "step": 18281
    },
    {
      "epoch": 0.6258815474152687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2688,
      "step": 18282
    },
    {
      "epoch": 0.6259157822663471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2759,
      "step": 18283
    },
    {
      "epoch": 0.6259500171174256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2732,
      "step": 18284
    },
    {
      "epoch": 0.6259842519685039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2839,
      "step": 18285
    },
    {
      "epoch": 0.6260184868195824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2584,
      "step": 18286
    },
    {
      "epoch": 0.6260527216706607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2682,
      "step": 18287
    },
    {
      "epoch": 0.6260869565217392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2726,
      "step": 18288
    },
    {
      "epoch": 0.6261211913728175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2632,
      "step": 18289
    },
    {
      "epoch": 0.626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2659,
      "step": 18290
    },
    {
      "epoch": 0.6261896610749743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.286,
      "step": 18291
    },
    {
      "epoch": 0.6262238959260528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2494,
      "step": 18292
    },
    {
      "epoch": 0.6262581307771311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2807,
      "step": 18293
    },
    {
      "epoch": 0.6262923656282096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2764,
      "step": 18294
    },
    {
      "epoch": 0.6263266004792879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2591,
      "step": 18295
    },
    {
      "epoch": 0.6263608353303664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2788,
      "step": 18296
    },
    {
      "epoch": 0.6263950701814447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2752,
      "step": 18297
    },
    {
      "epoch": 0.6264293050325231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2463,
      "step": 18298
    },
    {
      "epoch": 0.6264635398836015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2625,
      "step": 18299
    },
    {
      "epoch": 0.6264977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2832,
      "step": 18300
    },
    {
      "epoch": 0.6265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2729,
      "step": 18301
    },
    {
      "epoch": 0.6265662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2403,
      "step": 18302
    },
    {
      "epoch": 0.6266004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2685,
      "step": 18303
    },
    {
      "epoch": 0.6266347141389935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2758,
      "step": 18304
    },
    {
      "epoch": 0.6266689489900719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2789,
      "step": 18305
    },
    {
      "epoch": 0.6267031838411503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2721,
      "step": 18306
    },
    {
      "epoch": 0.6267374186922287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2576,
      "step": 18307
    },
    {
      "epoch": 0.6267716535433071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2584,
      "step": 18308
    },
    {
      "epoch": 0.6268058883943854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2651,
      "step": 18309
    },
    {
      "epoch": 0.6268401232454639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2637,
      "step": 18310
    },
    {
      "epoch": 0.6268743580965422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2432,
      "step": 18311
    },
    {
      "epoch": 0.6269085929476207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2449,
      "step": 18312
    },
    {
      "epoch": 0.626942827798699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2503,
      "step": 18313
    },
    {
      "epoch": 0.6269770626497775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2728,
      "step": 18314
    },
    {
      "epoch": 0.6270112975008558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2786,
      "step": 18315
    },
    {
      "epoch": 0.6270455323519343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.267,
      "step": 18316
    },
    {
      "epoch": 0.6270797672030126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2687,
      "step": 18317
    },
    {
      "epoch": 0.6271140020540911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2756,
      "step": 18318
    },
    {
      "epoch": 0.6271482369051694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2544,
      "step": 18319
    },
    {
      "epoch": 0.6271824717562479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.253,
      "step": 18320
    },
    {
      "epoch": 0.6272167066073263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2464,
      "step": 18321
    },
    {
      "epoch": 0.6272509414584047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2773,
      "step": 18322
    },
    {
      "epoch": 0.6272851763094831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2713,
      "step": 18323
    },
    {
      "epoch": 0.6273194111605614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2748,
      "step": 18324
    },
    {
      "epoch": 0.6273536460116399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2614,
      "step": 18325
    },
    {
      "epoch": 0.6273878808627182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2644,
      "step": 18326
    },
    {
      "epoch": 0.6274221157137967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2596,
      "step": 18327
    },
    {
      "epoch": 0.627456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2594,
      "step": 18328
    },
    {
      "epoch": 0.6274905854159535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2882,
      "step": 18329
    },
    {
      "epoch": 0.6275248202670318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2792,
      "step": 18330
    },
    {
      "epoch": 0.6275590551181103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2749,
      "step": 18331
    },
    {
      "epoch": 0.6275932899691886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2732,
      "step": 18332
    },
    {
      "epoch": 0.6276275248202671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2851,
      "step": 18333
    },
    {
      "epoch": 0.6276617596713454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.274,
      "step": 18334
    },
    {
      "epoch": 0.6276959945224239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2582,
      "step": 18335
    },
    {
      "epoch": 0.6277302293735022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2691,
      "step": 18336
    },
    {
      "epoch": 0.6277644642245807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2803,
      "step": 18337
    },
    {
      "epoch": 0.627798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2538,
      "step": 18338
    },
    {
      "epoch": 0.6278329339267374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.262,
      "step": 18339
    },
    {
      "epoch": 0.6278671687778158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2526,
      "step": 18340
    },
    {
      "epoch": 0.6279014036288942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2649,
      "step": 18341
    },
    {
      "epoch": 0.6279356384799726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2682,
      "step": 18342
    },
    {
      "epoch": 0.627969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2596,
      "step": 18343
    },
    {
      "epoch": 0.6280041081821294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2658,
      "step": 18344
    },
    {
      "epoch": 0.6280383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2913,
      "step": 18345
    },
    {
      "epoch": 0.6280725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2699,
      "step": 18346
    },
    {
      "epoch": 0.6281068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2721,
      "step": 18347
    },
    {
      "epoch": 0.628141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2796,
      "step": 18348
    },
    {
      "epoch": 0.6281752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2706,
      "step": 18349
    },
    {
      "epoch": 0.6282095172885998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2639,
      "step": 18350
    },
    {
      "epoch": 0.6282437521396782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2824,
      "step": 18351
    },
    {
      "epoch": 0.6282779869907565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2708,
      "step": 18352
    },
    {
      "epoch": 0.628312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2715,
      "step": 18353
    },
    {
      "epoch": 0.6283464566929133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2643,
      "step": 18354
    },
    {
      "epoch": 0.6283806915439918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2601,
      "step": 18355
    },
    {
      "epoch": 0.6284149263950701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2767,
      "step": 18356
    },
    {
      "epoch": 0.6284491612461486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2662,
      "step": 18357
    },
    {
      "epoch": 0.628483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2658,
      "step": 18358
    },
    {
      "epoch": 0.6285176309483054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2574,
      "step": 18359
    },
    {
      "epoch": 0.6285518657993838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2576,
      "step": 18360
    },
    {
      "epoch": 0.6285861006504622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2643,
      "step": 18361
    },
    {
      "epoch": 0.6286203355015406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2797,
      "step": 18362
    },
    {
      "epoch": 0.628654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2645,
      "step": 18363
    },
    {
      "epoch": 0.6286888052036974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2458,
      "step": 18364
    },
    {
      "epoch": 0.6287230400547758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2825,
      "step": 18365
    },
    {
      "epoch": 0.6287572749058542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2685,
      "step": 18366
    },
    {
      "epoch": 0.6287915097569325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2648,
      "step": 18367
    },
    {
      "epoch": 0.628825744608011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2524,
      "step": 18368
    },
    {
      "epoch": 0.6288599794590893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2605,
      "step": 18369
    },
    {
      "epoch": 0.6288942143101678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2544,
      "step": 18370
    },
    {
      "epoch": 0.6289284491612461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.253,
      "step": 18371
    },
    {
      "epoch": 0.6289626840123246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2646,
      "step": 18372
    },
    {
      "epoch": 0.6289969188634029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2925,
      "step": 18373
    },
    {
      "epoch": 0.6290311537144814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.272,
      "step": 18374
    },
    {
      "epoch": 0.6290653885655597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2531,
      "step": 18375
    },
    {
      "epoch": 0.6290996234166382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2858,
      "step": 18376
    },
    {
      "epoch": 0.6291338582677165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2571,
      "step": 18377
    },
    {
      "epoch": 0.629168093118795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2787,
      "step": 18378
    },
    {
      "epoch": 0.6292023279698733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2662,
      "step": 18379
    },
    {
      "epoch": 0.6292365628209518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2647,
      "step": 18380
    },
    {
      "epoch": 0.6292707976720301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2605,
      "step": 18381
    },
    {
      "epoch": 0.6293050325231085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2764,
      "step": 18382
    },
    {
      "epoch": 0.6293392673741869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2567,
      "step": 18383
    },
    {
      "epoch": 0.6293735022252653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2533,
      "step": 18384
    },
    {
      "epoch": 0.6294077370763437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.272,
      "step": 18385
    },
    {
      "epoch": 0.6294419719274221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2717,
      "step": 18386
    },
    {
      "epoch": 0.6294762067785005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2524,
      "step": 18387
    },
    {
      "epoch": 0.6295104416295789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2639,
      "step": 18388
    },
    {
      "epoch": 0.6295446764806573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2638,
      "step": 18389
    },
    {
      "epoch": 0.6295789113317357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2745,
      "step": 18390
    },
    {
      "epoch": 0.629613146182814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2617,
      "step": 18391
    },
    {
      "epoch": 0.6296473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2751,
      "step": 18392
    },
    {
      "epoch": 0.6296816158849708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2473,
      "step": 18393
    },
    {
      "epoch": 0.6297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2593,
      "step": 18394
    },
    {
      "epoch": 0.6297500855871278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2665,
      "step": 18395
    },
    {
      "epoch": 0.6297843204382061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2811,
      "step": 18396
    },
    {
      "epoch": 0.6298185552892845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2562,
      "step": 18397
    },
    {
      "epoch": 0.6298527901403629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2668,
      "step": 18398
    },
    {
      "epoch": 0.6298870249914413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2781,
      "step": 18399
    },
    {
      "epoch": 0.6299212598425197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2732,
      "step": 18400
    },
    {
      "epoch": 0.6299554946935981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2598,
      "step": 18401
    },
    {
      "epoch": 0.6299897295446765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2709,
      "step": 18402
    },
    {
      "epoch": 0.6300239643957549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2865,
      "step": 18403
    },
    {
      "epoch": 0.6300581992468333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2592,
      "step": 18404
    },
    {
      "epoch": 0.6300924340979117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.273,
      "step": 18405
    },
    {
      "epoch": 0.63012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2386,
      "step": 18406
    },
    {
      "epoch": 0.6301609038000685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2577,
      "step": 18407
    },
    {
      "epoch": 0.6301951386511468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2592,
      "step": 18408
    },
    {
      "epoch": 0.6302293735022253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2649,
      "step": 18409
    },
    {
      "epoch": 0.6302636083533036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2689,
      "step": 18410
    },
    {
      "epoch": 0.6302978432043821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2793,
      "step": 18411
    },
    {
      "epoch": 0.6303320780554604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2638,
      "step": 18412
    },
    {
      "epoch": 0.6303663129065389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2697,
      "step": 18413
    },
    {
      "epoch": 0.6304005477576172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2702,
      "step": 18414
    },
    {
      "epoch": 0.6304347826086957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2544,
      "step": 18415
    },
    {
      "epoch": 0.630469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2704,
      "step": 18416
    },
    {
      "epoch": 0.6305032523108525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2632,
      "step": 18417
    },
    {
      "epoch": 0.6305374871619308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2815,
      "step": 18418
    },
    {
      "epoch": 0.6305717220130093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2736,
      "step": 18419
    },
    {
      "epoch": 0.6306059568640876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2789,
      "step": 18420
    },
    {
      "epoch": 0.630640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2485,
      "step": 18421
    },
    {
      "epoch": 0.6306744265662444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.274,
      "step": 18422
    },
    {
      "epoch": 0.6307086614173228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2738,
      "step": 18423
    },
    {
      "epoch": 0.6307428962684012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2469,
      "step": 18424
    },
    {
      "epoch": 0.6307771311194796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2565,
      "step": 18425
    },
    {
      "epoch": 0.630811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2592,
      "step": 18426
    },
    {
      "epoch": 0.6308456008216364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2661,
      "step": 18427
    },
    {
      "epoch": 0.6308798356727148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2694,
      "step": 18428
    },
    {
      "epoch": 0.6309140705237932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2653,
      "step": 18429
    },
    {
      "epoch": 0.6309483053748716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2834,
      "step": 18430
    },
    {
      "epoch": 0.63098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2721,
      "step": 18431
    },
    {
      "epoch": 0.6310167750770285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2683,
      "step": 18432
    },
    {
      "epoch": 0.6310510099281068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2797,
      "step": 18433
    },
    {
      "epoch": 0.6310852447791853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.264,
      "step": 18434
    },
    {
      "epoch": 0.6311194796302636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2473,
      "step": 18435
    },
    {
      "epoch": 0.631153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2601,
      "step": 18436
    },
    {
      "epoch": 0.6311879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2673,
      "step": 18437
    },
    {
      "epoch": 0.6312221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2629,
      "step": 18438
    },
    {
      "epoch": 0.6312564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2601,
      "step": 18439
    },
    {
      "epoch": 0.6312906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2723,
      "step": 18440
    },
    {
      "epoch": 0.631324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2615,
      "step": 18441
    },
    {
      "epoch": 0.6313591235878124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2736,
      "step": 18442
    },
    {
      "epoch": 0.6313933584388908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2559,
      "step": 18443
    },
    {
      "epoch": 0.6314275932899692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.269,
      "step": 18444
    },
    {
      "epoch": 0.6314618281410476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2993,
      "step": 18445
    },
    {
      "epoch": 0.631496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2521,
      "step": 18446
    },
    {
      "epoch": 0.6315302978432044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2639,
      "step": 18447
    },
    {
      "epoch": 0.6315645326942828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.255,
      "step": 18448
    },
    {
      "epoch": 0.6315987675453612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2525,
      "step": 18449
    },
    {
      "epoch": 0.6316330023964396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2733,
      "step": 18450
    },
    {
      "epoch": 0.6316672372475179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2466,
      "step": 18451
    },
    {
      "epoch": 0.6317014720985964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2833,
      "step": 18452
    },
    {
      "epoch": 0.6317357069496747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.282,
      "step": 18453
    },
    {
      "epoch": 0.6317699418007532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2551,
      "step": 18454
    },
    {
      "epoch": 0.6318041766518315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2679,
      "step": 18455
    },
    {
      "epoch": 0.63183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2817,
      "step": 18456
    },
    {
      "epoch": 0.6318726463539883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2705,
      "step": 18457
    },
    {
      "epoch": 0.6319068812050668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.277,
      "step": 18458
    },
    {
      "epoch": 0.6319411160561451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2627,
      "step": 18459
    },
    {
      "epoch": 0.6319753509072236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 18460
    },
    {
      "epoch": 0.6320095857583019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2672,
      "step": 18461
    },
    {
      "epoch": 0.6320438206093804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2488,
      "step": 18462
    },
    {
      "epoch": 0.6320780554604587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2769,
      "step": 18463
    },
    {
      "epoch": 0.6321122903115372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2686,
      "step": 18464
    },
    {
      "epoch": 0.6321465251626155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2717,
      "step": 18465
    },
    {
      "epoch": 0.6321807600136939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2708,
      "step": 18466
    },
    {
      "epoch": 0.6322149948647723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2574,
      "step": 18467
    },
    {
      "epoch": 0.6322492297158507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2565,
      "step": 18468
    },
    {
      "epoch": 0.6322834645669292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.273,
      "step": 18469
    },
    {
      "epoch": 0.6323176994180075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2638,
      "step": 18470
    },
    {
      "epoch": 0.632351934269086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2675,
      "step": 18471
    },
    {
      "epoch": 0.6323861691201643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2488,
      "step": 18472
    },
    {
      "epoch": 0.6324204039712428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2708,
      "step": 18473
    },
    {
      "epoch": 0.6324546388223211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2567,
      "step": 18474
    },
    {
      "epoch": 0.6324888736733996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2423,
      "step": 18475
    },
    {
      "epoch": 0.6325231085244779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2758,
      "step": 18476
    },
    {
      "epoch": 0.6325573433755564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2581,
      "step": 18477
    },
    {
      "epoch": 0.6325915782266347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2723,
      "step": 18478
    },
    {
      "epoch": 0.6326258130777132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2756,
      "step": 18479
    },
    {
      "epoch": 0.6326600479287915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2623,
      "step": 18480
    },
    {
      "epoch": 0.6326942827798699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2747,
      "step": 18481
    },
    {
      "epoch": 0.6327285176309483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2627,
      "step": 18482
    },
    {
      "epoch": 0.6327627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2574,
      "step": 18483
    },
    {
      "epoch": 0.6327969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2497,
      "step": 18484
    },
    {
      "epoch": 0.6328312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2759,
      "step": 18485
    },
    {
      "epoch": 0.6328654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2497,
      "step": 18486
    },
    {
      "epoch": 0.6328996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2644,
      "step": 18487
    },
    {
      "epoch": 0.6329339267374187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2546,
      "step": 18488
    },
    {
      "epoch": 0.6329681615884971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.271,
      "step": 18489
    },
    {
      "epoch": 0.6330023964395755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2572,
      "step": 18490
    },
    {
      "epoch": 0.6330366312906539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2711,
      "step": 18491
    },
    {
      "epoch": 0.6330708661417322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2705,
      "step": 18492
    },
    {
      "epoch": 0.6331051009928107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2734,
      "step": 18493
    },
    {
      "epoch": 0.633139335843889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2649,
      "step": 18494
    },
    {
      "epoch": 0.6331735706949675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2526,
      "step": 18495
    },
    {
      "epoch": 0.6332078055460458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2699,
      "step": 18496
    },
    {
      "epoch": 0.6332420403971243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2709,
      "step": 18497
    },
    {
      "epoch": 0.6332762752482026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2802,
      "step": 18498
    },
    {
      "epoch": 0.6333105100992811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2555,
      "step": 18499
    },
    {
      "epoch": 0.6333447449503594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2653,
      "step": 18500
    },
    {
      "epoch": 0.6333789798014379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.244,
      "step": 18501
    },
    {
      "epoch": 0.6334132146525162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2759,
      "step": 18502
    },
    {
      "epoch": 0.6334474495035947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2446,
      "step": 18503
    },
    {
      "epoch": 0.633481684354673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.255,
      "step": 18504
    },
    {
      "epoch": 0.6335159192057515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2691,
      "step": 18505
    },
    {
      "epoch": 0.6335501540568299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2675,
      "step": 18506
    },
    {
      "epoch": 0.6335843889079082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2859,
      "step": 18507
    },
    {
      "epoch": 0.6336186237589867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2531,
      "step": 18508
    },
    {
      "epoch": 0.633652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2497,
      "step": 18509
    },
    {
      "epoch": 0.6336870934611435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2517,
      "step": 18510
    },
    {
      "epoch": 0.6337213283122218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.268,
      "step": 18511
    },
    {
      "epoch": 0.6337555631633003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2267,
      "step": 18512
    },
    {
      "epoch": 0.6337897980143786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2587,
      "step": 18513
    },
    {
      "epoch": 0.6338240328654571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2536,
      "step": 18514
    },
    {
      "epoch": 0.6338582677165354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2645,
      "step": 18515
    },
    {
      "epoch": 0.6338925025676139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2866,
      "step": 18516
    },
    {
      "epoch": 0.6339267374186922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2639,
      "step": 18517
    },
    {
      "epoch": 0.6339609722697707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2666,
      "step": 18518
    },
    {
      "epoch": 0.633995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2533,
      "step": 18519
    },
    {
      "epoch": 0.6340294419719275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2828,
      "step": 18520
    },
    {
      "epoch": 0.6340636768230058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2683,
      "step": 18521
    },
    {
      "epoch": 0.6340979116740842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2664,
      "step": 18522
    },
    {
      "epoch": 0.6341321465251626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2705,
      "step": 18523
    },
    {
      "epoch": 0.634166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2538,
      "step": 18524
    },
    {
      "epoch": 0.6342006162273194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2606,
      "step": 18525
    },
    {
      "epoch": 0.6342348510783978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2787,
      "step": 18526
    },
    {
      "epoch": 0.6342690859294762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2607,
      "step": 18527
    },
    {
      "epoch": 0.6343033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2628,
      "step": 18528
    },
    {
      "epoch": 0.634337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2625,
      "step": 18529
    },
    {
      "epoch": 0.6343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2756,
      "step": 18530
    },
    {
      "epoch": 0.6344060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2746,
      "step": 18531
    },
    {
      "epoch": 0.6344402601848682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2605,
      "step": 18532
    },
    {
      "epoch": 0.6344744950359466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2673,
      "step": 18533
    },
    {
      "epoch": 0.634508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2469,
      "step": 18534
    },
    {
      "epoch": 0.6345429647381033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2643,
      "step": 18535
    },
    {
      "epoch": 0.6345771995891818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.274,
      "step": 18536
    },
    {
      "epoch": 0.6346114344402601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2503,
      "step": 18537
    },
    {
      "epoch": 0.6346456692913386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2812,
      "step": 18538
    },
    {
      "epoch": 0.6346799041424169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2602,
      "step": 18539
    },
    {
      "epoch": 0.6347141389934954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2681,
      "step": 18540
    },
    {
      "epoch": 0.6347483738445737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2521,
      "step": 18541
    },
    {
      "epoch": 0.6347826086956522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2615,
      "step": 18542
    },
    {
      "epoch": 0.6348168435467306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2427,
      "step": 18543
    },
    {
      "epoch": 0.634851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2576,
      "step": 18544
    },
    {
      "epoch": 0.6348853132488874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2583,
      "step": 18545
    },
    {
      "epoch": 0.6349195480999658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2642,
      "step": 18546
    },
    {
      "epoch": 0.6349537829510442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2653,
      "step": 18547
    },
    {
      "epoch": 0.6349880178021226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2702,
      "step": 18548
    },
    {
      "epoch": 0.635022252653201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2534,
      "step": 18549
    },
    {
      "epoch": 0.6350564875042793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2488,
      "step": 18550
    },
    {
      "epoch": 0.6350907223553578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2515,
      "step": 18551
    },
    {
      "epoch": 0.6351249572064361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.274,
      "step": 18552
    },
    {
      "epoch": 0.6351591920575146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2645,
      "step": 18553
    },
    {
      "epoch": 0.6351934269085929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2513,
      "step": 18554
    },
    {
      "epoch": 0.6352276617596714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2613,
      "step": 18555
    },
    {
      "epoch": 0.6352618966107497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2466,
      "step": 18556
    },
    {
      "epoch": 0.6352961314618282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2596,
      "step": 18557
    },
    {
      "epoch": 0.6353303663129065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2786,
      "step": 18558
    },
    {
      "epoch": 0.635364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.268,
      "step": 18559
    },
    {
      "epoch": 0.6353988360150633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.262,
      "step": 18560
    },
    {
      "epoch": 0.6354330708661418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.269,
      "step": 18561
    },
    {
      "epoch": 0.6354673057172201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2468,
      "step": 18562
    },
    {
      "epoch": 0.6355015405682986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2731,
      "step": 18563
    },
    {
      "epoch": 0.6355357754193769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2689,
      "step": 18564
    },
    {
      "epoch": 0.6355700102704553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2861,
      "step": 18565
    },
    {
      "epoch": 0.6356042451215337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2581,
      "step": 18566
    },
    {
      "epoch": 0.6356384799726121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2535,
      "step": 18567
    },
    {
      "epoch": 0.6356727148236905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2603,
      "step": 18568
    },
    {
      "epoch": 0.6357069496747689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2524,
      "step": 18569
    },
    {
      "epoch": 0.6357411845258473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2692,
      "step": 18570
    },
    {
      "epoch": 0.6357754193769257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2422,
      "step": 18571
    },
    {
      "epoch": 0.6358096542280041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2604,
      "step": 18572
    },
    {
      "epoch": 0.6358438890790825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2591,
      "step": 18573
    },
    {
      "epoch": 0.6358781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2566,
      "step": 18574
    },
    {
      "epoch": 0.6359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2681,
      "step": 18575
    },
    {
      "epoch": 0.6359465936323176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2528,
      "step": 18576
    },
    {
      "epoch": 0.6359808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2651,
      "step": 18577
    },
    {
      "epoch": 0.6360150633344744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2479,
      "step": 18578
    },
    {
      "epoch": 0.6360492981855529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2694,
      "step": 18579
    },
    {
      "epoch": 0.6360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2527,
      "step": 18580
    },
    {
      "epoch": 0.6361177678877097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.268,
      "step": 18581
    },
    {
      "epoch": 0.6361520027387881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.271,
      "step": 18582
    },
    {
      "epoch": 0.6361862375898665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2684,
      "step": 18583
    },
    {
      "epoch": 0.6362204724409449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.252,
      "step": 18584
    },
    {
      "epoch": 0.6362547072920233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2633,
      "step": 18585
    },
    {
      "epoch": 0.6362889421431017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2523,
      "step": 18586
    },
    {
      "epoch": 0.6363231769941801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2438,
      "step": 18587
    },
    {
      "epoch": 0.6363574118452585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2531,
      "step": 18588
    },
    {
      "epoch": 0.6363916466963369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2723,
      "step": 18589
    },
    {
      "epoch": 0.6364258815474153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2518,
      "step": 18590
    },
    {
      "epoch": 0.6364601163984936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2334,
      "step": 18591
    },
    {
      "epoch": 0.6364943512495721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2574,
      "step": 18592
    },
    {
      "epoch": 0.6365285861006504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.27,
      "step": 18593
    },
    {
      "epoch": 0.6365628209517289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2425,
      "step": 18594
    },
    {
      "epoch": 0.6365970558028072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2595,
      "step": 18595
    },
    {
      "epoch": 0.6366312906538857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2503,
      "step": 18596
    },
    {
      "epoch": 0.636665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2532,
      "step": 18597
    },
    {
      "epoch": 0.6366997603560425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2415,
      "step": 18598
    },
    {
      "epoch": 0.6367339952071208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2618,
      "step": 18599
    },
    {
      "epoch": 0.6367682300581993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2567,
      "step": 18600
    },
    {
      "epoch": 0.6368024649092776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2614,
      "step": 18601
    },
    {
      "epoch": 0.6368366997603561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.261,
      "step": 18602
    },
    {
      "epoch": 0.6368709346114344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2565,
      "step": 18603
    },
    {
      "epoch": 0.6369051694625129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2684,
      "step": 18604
    },
    {
      "epoch": 0.6369394043135912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2838,
      "step": 18605
    },
    {
      "epoch": 0.6369736391646696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.255,
      "step": 18606
    },
    {
      "epoch": 0.637007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2602,
      "step": 18607
    },
    {
      "epoch": 0.6370421088668264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2591,
      "step": 18608
    },
    {
      "epoch": 0.6370763437179048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2691,
      "step": 18609
    },
    {
      "epoch": 0.6371105785689832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2702,
      "step": 18610
    },
    {
      "epoch": 0.6371448134200616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2422,
      "step": 18611
    },
    {
      "epoch": 0.63717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2725,
      "step": 18612
    },
    {
      "epoch": 0.6372132831222184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.248,
      "step": 18613
    },
    {
      "epoch": 0.6372475179732968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2721,
      "step": 18614
    },
    {
      "epoch": 0.6372817528243753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2529,
      "step": 18615
    },
    {
      "epoch": 0.6373159876754536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2506,
      "step": 18616
    },
    {
      "epoch": 0.6373502225265321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2553,
      "step": 18617
    },
    {
      "epoch": 0.6373844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2671,
      "step": 18618
    },
    {
      "epoch": 0.6374186922286889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2384,
      "step": 18619
    },
    {
      "epoch": 0.6374529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2352,
      "step": 18620
    },
    {
      "epoch": 0.6374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2513,
      "step": 18621
    },
    {
      "epoch": 0.637521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2474,
      "step": 18622
    },
    {
      "epoch": 0.6375556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.272,
      "step": 18623
    },
    {
      "epoch": 0.6375898664840808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2644,
      "step": 18624
    },
    {
      "epoch": 0.6376241013351592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2646,
      "step": 18625
    },
    {
      "epoch": 0.6376583361862376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2662,
      "step": 18626
    },
    {
      "epoch": 0.637692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2701,
      "step": 18627
    },
    {
      "epoch": 0.6377268058883944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2614,
      "step": 18628
    },
    {
      "epoch": 0.6377610407394728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2572,
      "step": 18629
    },
    {
      "epoch": 0.6377952755905512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2651,
      "step": 18630
    },
    {
      "epoch": 0.6378295104416296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2578,
      "step": 18631
    },
    {
      "epoch": 0.637863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2433,
      "step": 18632
    },
    {
      "epoch": 0.6378979801437864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2584,
      "step": 18633
    },
    {
      "epoch": 0.6379322149948647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2733,
      "step": 18634
    },
    {
      "epoch": 0.6379664498459432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2649,
      "step": 18635
    },
    {
      "epoch": 0.6380006846970215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2617,
      "step": 18636
    },
    {
      "epoch": 0.6380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2764,
      "step": 18637
    },
    {
      "epoch": 0.6380691543991783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2593,
      "step": 18638
    },
    {
      "epoch": 0.6381033892502568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2618,
      "step": 18639
    },
    {
      "epoch": 0.6381376241013351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2474,
      "step": 18640
    },
    {
      "epoch": 0.6381718589524136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2542,
      "step": 18641
    },
    {
      "epoch": 0.6382060938034919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2658,
      "step": 18642
    },
    {
      "epoch": 0.6382403286545704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2616,
      "step": 18643
    },
    {
      "epoch": 0.6382745635056487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2612,
      "step": 18644
    },
    {
      "epoch": 0.6383087983567272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2657,
      "step": 18645
    },
    {
      "epoch": 0.6383430332078055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2527,
      "step": 18646
    },
    {
      "epoch": 0.638377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2572,
      "step": 18647
    },
    {
      "epoch": 0.6384115029099623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2548,
      "step": 18648
    },
    {
      "epoch": 0.6384457377610407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2837,
      "step": 18649
    },
    {
      "epoch": 0.6384799726121191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2752,
      "step": 18650
    },
    {
      "epoch": 0.6385142074631975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2623,
      "step": 18651
    },
    {
      "epoch": 0.638548442314276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2769,
      "step": 18652
    },
    {
      "epoch": 0.6385826771653543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2444,
      "step": 18653
    },
    {
      "epoch": 0.6386169120164328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2708,
      "step": 18654
    },
    {
      "epoch": 0.6386511468675111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2681,
      "step": 18655
    },
    {
      "epoch": 0.6386853817185896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2629,
      "step": 18656
    },
    {
      "epoch": 0.6387196165696679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2694,
      "step": 18657
    },
    {
      "epoch": 0.6387538514207464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2656,
      "step": 18658
    },
    {
      "epoch": 0.6387880862718247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.256,
      "step": 18659
    },
    {
      "epoch": 0.6388223211229032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2641,
      "step": 18660
    },
    {
      "epoch": 0.6388565559739815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.274,
      "step": 18661
    },
    {
      "epoch": 0.63889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2706,
      "step": 18662
    },
    {
      "epoch": 0.6389250256761383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2556,
      "step": 18663
    },
    {
      "epoch": 0.6389592605272167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2636,
      "step": 18664
    },
    {
      "epoch": 0.6389934953782951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.285,
      "step": 18665
    },
    {
      "epoch": 0.6390277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2654,
      "step": 18666
    },
    {
      "epoch": 0.6390619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2657,
      "step": 18667
    },
    {
      "epoch": 0.6390961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2605,
      "step": 18668
    },
    {
      "epoch": 0.6391304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2458,
      "step": 18669
    },
    {
      "epoch": 0.6391646696336871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2593,
      "step": 18670
    },
    {
      "epoch": 0.6391989044847655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2607,
      "step": 18671
    },
    {
      "epoch": 0.6392331393358439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2601,
      "step": 18672
    },
    {
      "epoch": 0.6392673741869223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2734,
      "step": 18673
    },
    {
      "epoch": 0.6393016090380007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2503,
      "step": 18674
    },
    {
      "epoch": 0.639335843889079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2468,
      "step": 18675
    },
    {
      "epoch": 0.6393700787401575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2695,
      "step": 18676
    },
    {
      "epoch": 0.6394043135912358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2658,
      "step": 18677
    },
    {
      "epoch": 0.6394385484423143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2588,
      "step": 18678
    },
    {
      "epoch": 0.6394727832933926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2562,
      "step": 18679
    },
    {
      "epoch": 0.6395070181444711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2458,
      "step": 18680
    },
    {
      "epoch": 0.6395412529955494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2494,
      "step": 18681
    },
    {
      "epoch": 0.6395754878466279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2693,
      "step": 18682
    },
    {
      "epoch": 0.6396097226977062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2762,
      "step": 18683
    },
    {
      "epoch": 0.6396439575487847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2554,
      "step": 18684
    },
    {
      "epoch": 0.639678192399863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2706,
      "step": 18685
    },
    {
      "epoch": 0.6397124272509415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2589,
      "step": 18686
    },
    {
      "epoch": 0.6397466621020198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2434,
      "step": 18687
    },
    {
      "epoch": 0.6397808969530983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.263,
      "step": 18688
    },
    {
      "epoch": 0.6398151318041767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2818,
      "step": 18689
    },
    {
      "epoch": 0.639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2684,
      "step": 18690
    },
    {
      "epoch": 0.6398836015063335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.259,
      "step": 18691
    },
    {
      "epoch": 0.6399178363574118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2543,
      "step": 18692
    },
    {
      "epoch": 0.6399520712084903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2719,
      "step": 18693
    },
    {
      "epoch": 0.6399863060595686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2446,
      "step": 18694
    },
    {
      "epoch": 0.6400205409106471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2559,
      "step": 18695
    },
    {
      "epoch": 0.6400547757617254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2616,
      "step": 18696
    },
    {
      "epoch": 0.6400890106128039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2493,
      "step": 18697
    },
    {
      "epoch": 0.6401232454638822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2696,
      "step": 18698
    },
    {
      "epoch": 0.6401574803149607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2427,
      "step": 18699
    },
    {
      "epoch": 0.640191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2476,
      "step": 18700
    },
    {
      "epoch": 0.6402259500171175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2522,
      "step": 18701
    },
    {
      "epoch": 0.6402601848681958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2631,
      "step": 18702
    },
    {
      "epoch": 0.6402944197192743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2476,
      "step": 18703
    },
    {
      "epoch": 0.6403286545703526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2582,
      "step": 18704
    },
    {
      "epoch": 0.640362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2578,
      "step": 18705
    },
    {
      "epoch": 0.6403971242725094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2686,
      "step": 18706
    },
    {
      "epoch": 0.6404313591235878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2759,
      "step": 18707
    },
    {
      "epoch": 0.6404655939746662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.27,
      "step": 18708
    },
    {
      "epoch": 0.6404998288257446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2679,
      "step": 18709
    },
    {
      "epoch": 0.640534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2613,
      "step": 18710
    },
    {
      "epoch": 0.6405682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2571,
      "step": 18711
    },
    {
      "epoch": 0.6406025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2559,
      "step": 18712
    },
    {
      "epoch": 0.6406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2567,
      "step": 18713
    },
    {
      "epoch": 0.6406710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2614,
      "step": 18714
    },
    {
      "epoch": 0.640705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2574,
      "step": 18715
    },
    {
      "epoch": 0.6407394727832934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2567,
      "step": 18716
    },
    {
      "epoch": 0.6407737076343718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2543,
      "step": 18717
    },
    {
      "epoch": 0.6408079424854501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2427,
      "step": 18718
    },
    {
      "epoch": 0.6408421773365286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2621,
      "step": 18719
    },
    {
      "epoch": 0.6408764121876069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2592,
      "step": 18720
    },
    {
      "epoch": 0.6409106470386854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2754,
      "step": 18721
    },
    {
      "epoch": 0.6409448818897637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2693,
      "step": 18722
    },
    {
      "epoch": 0.6409791167408422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2511,
      "step": 18723
    },
    {
      "epoch": 0.6410133515919205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2596,
      "step": 18724
    },
    {
      "epoch": 0.641047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2728,
      "step": 18725
    },
    {
      "epoch": 0.6410818212940774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2701,
      "step": 18726
    },
    {
      "epoch": 0.6411160561451558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2625,
      "step": 18727
    },
    {
      "epoch": 0.6411502909962342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2441,
      "step": 18728
    },
    {
      "epoch": 0.6411845258473126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2447,
      "step": 18729
    },
    {
      "epoch": 0.641218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.269,
      "step": 18730
    },
    {
      "epoch": 0.6412529955494694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2528,
      "step": 18731
    },
    {
      "epoch": 0.6412872304005478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2561,
      "step": 18732
    },
    {
      "epoch": 0.6413214652516261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2478,
      "step": 18733
    },
    {
      "epoch": 0.6413557001027046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2653,
      "step": 18734
    },
    {
      "epoch": 0.6413899349537829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2639,
      "step": 18735
    },
    {
      "epoch": 0.6414241698048614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2544,
      "step": 18736
    },
    {
      "epoch": 0.6414584046559397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2562,
      "step": 18737
    },
    {
      "epoch": 0.6414926395070182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2531,
      "step": 18738
    },
    {
      "epoch": 0.6415268743580965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2654,
      "step": 18739
    },
    {
      "epoch": 0.641561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2617,
      "step": 18740
    },
    {
      "epoch": 0.6415953440602533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2478,
      "step": 18741
    },
    {
      "epoch": 0.6416295789113318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2501,
      "step": 18742
    },
    {
      "epoch": 0.6416638137624101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2655,
      "step": 18743
    },
    {
      "epoch": 0.6416980486134886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2689,
      "step": 18744
    },
    {
      "epoch": 0.6417322834645669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2536,
      "step": 18745
    },
    {
      "epoch": 0.6417665183156454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2571,
      "step": 18746
    },
    {
      "epoch": 0.6418007531667237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2612,
      "step": 18747
    },
    {
      "epoch": 0.6418349880178021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.259,
      "step": 18748
    },
    {
      "epoch": 0.6418692228688805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2661,
      "step": 18749
    },
    {
      "epoch": 0.6419034577199589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2664,
      "step": 18750
    },
    {
      "epoch": 0.6419376925710373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2696,
      "step": 18751
    },
    {
      "epoch": 0.6419719274221157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2556,
      "step": 18752
    },
    {
      "epoch": 0.6420061622731941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2545,
      "step": 18753
    },
    {
      "epoch": 0.6420403971242725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2446,
      "step": 18754
    },
    {
      "epoch": 0.6420746319753509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2518,
      "step": 18755
    },
    {
      "epoch": 0.6421088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2797,
      "step": 18756
    },
    {
      "epoch": 0.6421431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2424,
      "step": 18757
    },
    {
      "epoch": 0.6421773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2614,
      "step": 18758
    },
    {
      "epoch": 0.6422115713796644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2627,
      "step": 18759
    },
    {
      "epoch": 0.6422458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.261,
      "step": 18760
    },
    {
      "epoch": 0.6422800410818212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2627,
      "step": 18761
    },
    {
      "epoch": 0.6423142759328997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2682,
      "step": 18762
    },
    {
      "epoch": 0.6423485107839781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2626,
      "step": 18763
    },
    {
      "epoch": 0.6423827456350565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2528,
      "step": 18764
    },
    {
      "epoch": 0.6424169804861349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2603,
      "step": 18765
    },
    {
      "epoch": 0.6424512153372133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2695,
      "step": 18766
    },
    {
      "epoch": 0.6424854501882917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2533,
      "step": 18767
    },
    {
      "epoch": 0.6425196850393701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2766,
      "step": 18768
    },
    {
      "epoch": 0.6425539198904485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2803,
      "step": 18769
    },
    {
      "epoch": 0.6425881547415269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2791,
      "step": 18770
    },
    {
      "epoch": 0.6426223895926053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2672,
      "step": 18771
    },
    {
      "epoch": 0.6426566244436837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.249,
      "step": 18772
    },
    {
      "epoch": 0.6426908592947621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2716,
      "step": 18773
    },
    {
      "epoch": 0.6427250941458404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2642,
      "step": 18774
    },
    {
      "epoch": 0.6427593289969189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2614,
      "step": 18775
    },
    {
      "epoch": 0.6427935638479972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2409,
      "step": 18776
    },
    {
      "epoch": 0.6428277986990757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2536,
      "step": 18777
    },
    {
      "epoch": 0.642862033550154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.256,
      "step": 18778
    },
    {
      "epoch": 0.6428962684012325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2618,
      "step": 18779
    },
    {
      "epoch": 0.6429305032523108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2703,
      "step": 18780
    },
    {
      "epoch": 0.6429647381033893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2438,
      "step": 18781
    },
    {
      "epoch": 0.6429989729544676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2554,
      "step": 18782
    },
    {
      "epoch": 0.6430332078055461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2641,
      "step": 18783
    },
    {
      "epoch": 0.6430674426566244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2453,
      "step": 18784
    },
    {
      "epoch": 0.6431016775077029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2237,
      "step": 18785
    },
    {
      "epoch": 0.6431359123587812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2588,
      "step": 18786
    },
    {
      "epoch": 0.6431701472098597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2512,
      "step": 18787
    },
    {
      "epoch": 0.643204382060938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2651,
      "step": 18788
    },
    {
      "epoch": 0.6432386169120164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2348,
      "step": 18789
    },
    {
      "epoch": 0.6432728517630948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.256,
      "step": 18790
    },
    {
      "epoch": 0.6433070866141732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2621,
      "step": 18791
    },
    {
      "epoch": 0.6433413214652516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2518,
      "step": 18792
    },
    {
      "epoch": 0.64337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2322,
      "step": 18793
    },
    {
      "epoch": 0.6434097911674084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.26,
      "step": 18794
    },
    {
      "epoch": 0.6434440260184868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2516,
      "step": 18795
    },
    {
      "epoch": 0.6434782608695652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2465,
      "step": 18796
    },
    {
      "epoch": 0.6435124957206436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2606,
      "step": 18797
    },
    {
      "epoch": 0.643546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2584,
      "step": 18798
    },
    {
      "epoch": 0.6435809654228004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2701,
      "step": 18799
    },
    {
      "epoch": 0.6436152002738789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.274,
      "step": 18800
    },
    {
      "epoch": 0.6436494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2492,
      "step": 18801
    },
    {
      "epoch": 0.6436836699760357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2623,
      "step": 18802
    },
    {
      "epoch": 0.643717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2589,
      "step": 18803
    },
    {
      "epoch": 0.6437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2604,
      "step": 18804
    },
    {
      "epoch": 0.6437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2437,
      "step": 18805
    },
    {
      "epoch": 0.6438206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2711,
      "step": 18806
    },
    {
      "epoch": 0.6438548442314276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2389,
      "step": 18807
    },
    {
      "epoch": 0.643889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2546,
      "step": 18808
    },
    {
      "epoch": 0.6439233139335844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2599,
      "step": 18809
    },
    {
      "epoch": 0.6439575487846628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2524,
      "step": 18810
    },
    {
      "epoch": 0.6439917836357412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2502,
      "step": 18811
    },
    {
      "epoch": 0.6440260184868196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.241,
      "step": 18812
    },
    {
      "epoch": 0.644060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2426,
      "step": 18813
    },
    {
      "epoch": 0.6440944881889764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2586,
      "step": 18814
    },
    {
      "epoch": 0.6441287230400548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2426,
      "step": 18815
    },
    {
      "epoch": 0.6441629578911332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2213,
      "step": 18816
    },
    {
      "epoch": 0.6441971927422115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.243,
      "step": 18817
    },
    {
      "epoch": 0.64423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2707,
      "step": 18818
    },
    {
      "epoch": 0.6442656624443683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2413,
      "step": 18819
    },
    {
      "epoch": 0.6442998972954468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2597,
      "step": 18820
    },
    {
      "epoch": 0.6443341321465251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.271,
      "step": 18821
    },
    {
      "epoch": 0.6443683669976036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2537,
      "step": 18822
    },
    {
      "epoch": 0.6444026018486819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2541,
      "step": 18823
    },
    {
      "epoch": 0.6444368366997604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2578,
      "step": 18824
    },
    {
      "epoch": 0.6444710715508387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2502,
      "step": 18825
    },
    {
      "epoch": 0.6445053064019172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2544,
      "step": 18826
    },
    {
      "epoch": 0.6445395412529955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2595,
      "step": 18827
    },
    {
      "epoch": 0.644573776104074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2606,
      "step": 18828
    },
    {
      "epoch": 0.6446080109551523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2342,
      "step": 18829
    },
    {
      "epoch": 0.6446422458062308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.256,
      "step": 18830
    },
    {
      "epoch": 0.6446764806573091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2521,
      "step": 18831
    },
    {
      "epoch": 0.6447107155083875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2532,
      "step": 18832
    },
    {
      "epoch": 0.6447449503594659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.252,
      "step": 18833
    },
    {
      "epoch": 0.6447791852105443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2704,
      "step": 18834
    },
    {
      "epoch": 0.6448134200616227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2737,
      "step": 18835
    },
    {
      "epoch": 0.6448476549127011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2429,
      "step": 18836
    },
    {
      "epoch": 0.6448818897637796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2503,
      "step": 18837
    },
    {
      "epoch": 0.6449161246148579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2568,
      "step": 18838
    },
    {
      "epoch": 0.6449503594659364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.243,
      "step": 18839
    },
    {
      "epoch": 0.6449845943170147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2511,
      "step": 18840
    },
    {
      "epoch": 0.6450188291680932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2506,
      "step": 18841
    },
    {
      "epoch": 0.6450530640191715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.254,
      "step": 18842
    },
    {
      "epoch": 0.64508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2549,
      "step": 18843
    },
    {
      "epoch": 0.6451215337213283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2571,
      "step": 18844
    },
    {
      "epoch": 0.6451557685724068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2576,
      "step": 18845
    },
    {
      "epoch": 0.6451900034234851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2732,
      "step": 18846
    },
    {
      "epoch": 0.6452242382745635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2948,
      "step": 18847
    },
    {
      "epoch": 0.6452584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.259,
      "step": 18848
    },
    {
      "epoch": 0.6452927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2601,
      "step": 18849
    },
    {
      "epoch": 0.6453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.282,
      "step": 18850
    },
    {
      "epoch": 0.6453611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2661,
      "step": 18851
    },
    {
      "epoch": 0.6453954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2449,
      "step": 18852
    },
    {
      "epoch": 0.6454296473810339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2285,
      "step": 18853
    },
    {
      "epoch": 0.6454638822321123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2397,
      "step": 18854
    },
    {
      "epoch": 0.6454981170831907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2466,
      "step": 18855
    },
    {
      "epoch": 0.645532351934269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2632,
      "step": 18856
    },
    {
      "epoch": 0.6455665867853475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2606,
      "step": 18857
    },
    {
      "epoch": 0.6456008216364258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2788,
      "step": 18858
    },
    {
      "epoch": 0.6456350564875043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2518,
      "step": 18859
    },
    {
      "epoch": 0.6456692913385826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2637,
      "step": 18860
    },
    {
      "epoch": 0.6457035261896611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2661,
      "step": 18861
    },
    {
      "epoch": 0.6457377610407394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2516,
      "step": 18862
    },
    {
      "epoch": 0.6457719958918179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2628,
      "step": 18863
    },
    {
      "epoch": 0.6458062307428962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2634,
      "step": 18864
    },
    {
      "epoch": 0.6458404655939747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2521,
      "step": 18865
    },
    {
      "epoch": 0.645874700445053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2483,
      "step": 18866
    },
    {
      "epoch": 0.6459089352961315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2645,
      "step": 18867
    },
    {
      "epoch": 0.6459431701472098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2368,
      "step": 18868
    },
    {
      "epoch": 0.6459774049982883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2609,
      "step": 18869
    },
    {
      "epoch": 0.6460116398493666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2527,
      "step": 18870
    },
    {
      "epoch": 0.646045874700445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2679,
      "step": 18871
    },
    {
      "epoch": 0.6460801095515234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2523,
      "step": 18872
    },
    {
      "epoch": 0.6461143444026018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2728,
      "step": 18873
    },
    {
      "epoch": 0.6461485792536803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.259,
      "step": 18874
    },
    {
      "epoch": 0.6461828141047586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2533,
      "step": 18875
    },
    {
      "epoch": 0.6462170489558371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2369,
      "step": 18876
    },
    {
      "epoch": 0.6462512838069154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2444,
      "step": 18877
    },
    {
      "epoch": 0.6462855186579939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2479,
      "step": 18878
    },
    {
      "epoch": 0.6463197535090722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2481,
      "step": 18879
    },
    {
      "epoch": 0.6463539883601507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2489,
      "step": 18880
    },
    {
      "epoch": 0.646388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2449,
      "step": 18881
    },
    {
      "epoch": 0.6464224580623075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2538,
      "step": 18882
    },
    {
      "epoch": 0.6464566929133858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2428,
      "step": 18883
    },
    {
      "epoch": 0.6464909277644643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2565,
      "step": 18884
    },
    {
      "epoch": 0.6465251626155426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2507,
      "step": 18885
    },
    {
      "epoch": 0.646559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2542,
      "step": 18886
    },
    {
      "epoch": 0.6465936323176994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2657,
      "step": 18887
    },
    {
      "epoch": 0.6466278671687778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2366,
      "step": 18888
    },
    {
      "epoch": 0.6466621020198562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2426,
      "step": 18889
    },
    {
      "epoch": 0.6466963368709346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2656,
      "step": 18890
    },
    {
      "epoch": 0.646730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2567,
      "step": 18891
    },
    {
      "epoch": 0.6467648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2492,
      "step": 18892
    },
    {
      "epoch": 0.6467990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2519,
      "step": 18893
    },
    {
      "epoch": 0.6468332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2606,
      "step": 18894
    },
    {
      "epoch": 0.6468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2515,
      "step": 18895
    },
    {
      "epoch": 0.646901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.256,
      "step": 18896
    },
    {
      "epoch": 0.6469359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2678,
      "step": 18897
    },
    {
      "epoch": 0.6469702156795618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2444,
      "step": 18898
    },
    {
      "epoch": 0.6470044505306402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2327,
      "step": 18899
    },
    {
      "epoch": 0.6470386853817186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2656,
      "step": 18900
    },
    {
      "epoch": 0.6470729202327969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2426,
      "step": 18901
    },
    {
      "epoch": 0.6471071550838754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2587,
      "step": 18902
    },
    {
      "epoch": 0.6471413899349537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2531,
      "step": 18903
    },
    {
      "epoch": 0.6471756247860322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.263,
      "step": 18904
    },
    {
      "epoch": 0.6472098596371105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.254,
      "step": 18905
    },
    {
      "epoch": 0.647244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2489,
      "step": 18906
    },
    {
      "epoch": 0.6472783293392673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2534,
      "step": 18907
    },
    {
      "epoch": 0.6473125641903458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2424,
      "step": 18908
    },
    {
      "epoch": 0.6473467990414241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2513,
      "step": 18909
    },
    {
      "epoch": 0.6473810338925026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2741,
      "step": 18910
    },
    {
      "epoch": 0.647415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2576,
      "step": 18911
    },
    {
      "epoch": 0.6474495035946594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2491,
      "step": 18912
    },
    {
      "epoch": 0.6474837384457378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2541,
      "step": 18913
    },
    {
      "epoch": 0.6475179732968162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2345,
      "step": 18914
    },
    {
      "epoch": 0.6475522081478946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2554,
      "step": 18915
    },
    {
      "epoch": 0.6475864429989729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2531,
      "step": 18916
    },
    {
      "epoch": 0.6476206778500514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2682,
      "step": 18917
    },
    {
      "epoch": 0.6476549127011297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2407,
      "step": 18918
    },
    {
      "epoch": 0.6476891475522082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2521,
      "step": 18919
    },
    {
      "epoch": 0.6477233824032865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2483,
      "step": 18920
    },
    {
      "epoch": 0.647757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2443,
      "step": 18921
    },
    {
      "epoch": 0.6477918521054433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2578,
      "step": 18922
    },
    {
      "epoch": 0.6478260869565218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2792,
      "step": 18923
    },
    {
      "epoch": 0.6478603218076001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2397,
      "step": 18924
    },
    {
      "epoch": 0.6478945566586786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.27,
      "step": 18925
    },
    {
      "epoch": 0.6479287915097569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.259,
      "step": 18926
    },
    {
      "epoch": 0.6479630263608354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2321,
      "step": 18927
    },
    {
      "epoch": 0.6479972612119137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.25,
      "step": 18928
    },
    {
      "epoch": 0.6480314960629922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2547,
      "step": 18929
    },
    {
      "epoch": 0.6480657309140705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2727,
      "step": 18930
    },
    {
      "epoch": 0.6480999657651489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.252,
      "step": 18931
    },
    {
      "epoch": 0.6481342006162273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2362,
      "step": 18932
    },
    {
      "epoch": 0.6481684354673057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2552,
      "step": 18933
    },
    {
      "epoch": 0.6482026703183841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2545,
      "step": 18934
    },
    {
      "epoch": 0.6482369051694625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2572,
      "step": 18935
    },
    {
      "epoch": 0.6482711400205409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2643,
      "step": 18936
    },
    {
      "epoch": 0.6483053748716193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2532,
      "step": 18937
    },
    {
      "epoch": 0.6483396097226977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2478,
      "step": 18938
    },
    {
      "epoch": 0.6483738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2488,
      "step": 18939
    },
    {
      "epoch": 0.6484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.252,
      "step": 18940
    },
    {
      "epoch": 0.6484423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2514,
      "step": 18941
    },
    {
      "epoch": 0.6484765491270112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2481,
      "step": 18942
    },
    {
      "epoch": 0.6485107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2381,
      "step": 18943
    },
    {
      "epoch": 0.648545018829168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2381,
      "step": 18944
    },
    {
      "epoch": 0.6485792536802465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2646,
      "step": 18945
    },
    {
      "epoch": 0.6486134885313248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2416,
      "step": 18946
    },
    {
      "epoch": 0.6486477233824033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2529,
      "step": 18947
    },
    {
      "epoch": 0.6486819582334817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2552,
      "step": 18948
    },
    {
      "epoch": 0.6487161930845601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2493,
      "step": 18949
    },
    {
      "epoch": 0.6487504279356385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2522,
      "step": 18950
    },
    {
      "epoch": 0.6487846627867169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2585,
      "step": 18951
    },
    {
      "epoch": 0.6488188976377953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2582,
      "step": 18952
    },
    {
      "epoch": 0.6488531324888737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2559,
      "step": 18953
    },
    {
      "epoch": 0.6488873673399521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2547,
      "step": 18954
    },
    {
      "epoch": 0.6489216021910305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.261,
      "step": 18955
    },
    {
      "epoch": 0.6489558370421089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2465,
      "step": 18956
    },
    {
      "epoch": 0.6489900718931872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2538,
      "step": 18957
    },
    {
      "epoch": 0.6490243067442657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2549,
      "step": 18958
    },
    {
      "epoch": 0.649058541595344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2668,
      "step": 18959
    },
    {
      "epoch": 0.6490927764464225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2726,
      "step": 18960
    },
    {
      "epoch": 0.6491270112975008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2474,
      "step": 18961
    },
    {
      "epoch": 0.6491612461485793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2409,
      "step": 18962
    },
    {
      "epoch": 0.6491954809996576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2466,
      "step": 18963
    },
    {
      "epoch": 0.6492297158507361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2851,
      "step": 18964
    },
    {
      "epoch": 0.6492639507018144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.252,
      "step": 18965
    },
    {
      "epoch": 0.6492981855528929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2539,
      "step": 18966
    },
    {
      "epoch": 0.6493324204039712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.259,
      "step": 18967
    },
    {
      "epoch": 0.6493666552550497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2459,
      "step": 18968
    },
    {
      "epoch": 0.649400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2574,
      "step": 18969
    },
    {
      "epoch": 0.6494351249572065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2499,
      "step": 18970
    },
    {
      "epoch": 0.6494693598082848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2544,
      "step": 18971
    },
    {
      "epoch": 0.6495035946593632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2524,
      "step": 18972
    },
    {
      "epoch": 0.6495378295104416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2518,
      "step": 18973
    },
    {
      "epoch": 0.64957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2418,
      "step": 18974
    },
    {
      "epoch": 0.6496062992125984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2426,
      "step": 18975
    },
    {
      "epoch": 0.6496405340636768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2602,
      "step": 18976
    },
    {
      "epoch": 0.6496747689147552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2464,
      "step": 18977
    },
    {
      "epoch": 0.6497090037658336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2473,
      "step": 18978
    },
    {
      "epoch": 0.649743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2501,
      "step": 18979
    },
    {
      "epoch": 0.6497774734679904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2416,
      "step": 18980
    },
    {
      "epoch": 0.6498117083190688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2686,
      "step": 18981
    },
    {
      "epoch": 0.6498459431701472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2512,
      "step": 18982
    },
    {
      "epoch": 0.6498801780212257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2489,
      "step": 18983
    },
    {
      "epoch": 0.649914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2401,
      "step": 18984
    },
    {
      "epoch": 0.6499486477233825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2401,
      "step": 18985
    },
    {
      "epoch": 0.6499828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2291,
      "step": 18986
    },
    {
      "epoch": 0.6500171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2442,
      "step": 18987
    },
    {
      "epoch": 0.6500513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2623,
      "step": 18988
    },
    {
      "epoch": 0.650085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2398,
      "step": 18989
    },
    {
      "epoch": 0.6501198219787744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2486,
      "step": 18990
    },
    {
      "epoch": 0.6501540568298528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2373,
      "step": 18991
    },
    {
      "epoch": 0.6501882916809312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.261,
      "step": 18992
    },
    {
      "epoch": 0.6502225265320096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2675,
      "step": 18993
    },
    {
      "epoch": 0.650256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2501,
      "step": 18994
    },
    {
      "epoch": 0.6502909962341664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2451,
      "step": 18995
    },
    {
      "epoch": 0.6503252310852448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2593,
      "step": 18996
    },
    {
      "epoch": 0.6503594659363232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2449,
      "step": 18997
    },
    {
      "epoch": 0.6503937007874016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2409,
      "step": 18998
    },
    {
      "epoch": 0.65042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2362,
      "step": 18999
    },
    {
      "epoch": 0.6504621704895583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2572,
      "step": 19000
    },
    {
      "epoch": 0.6504964053406368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2358,
      "step": 19001
    },
    {
      "epoch": 0.6505306401917151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2528,
      "step": 19002
    },
    {
      "epoch": 0.6505648750427936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2573,
      "step": 19003
    },
    {
      "epoch": 0.6505991098938719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2557,
      "step": 19004
    },
    {
      "epoch": 0.6506333447449504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2289,
      "step": 19005
    },
    {
      "epoch": 0.6506675795960287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2488,
      "step": 19006
    },
    {
      "epoch": 0.6507018144471072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2493,
      "step": 19007
    },
    {
      "epoch": 0.6507360492981855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2724,
      "step": 19008
    },
    {
      "epoch": 0.650770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2554,
      "step": 19009
    },
    {
      "epoch": 0.6508045190003423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2463,
      "step": 19010
    },
    {
      "epoch": 0.6508387538514208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2312,
      "step": 19011
    },
    {
      "epoch": 0.6508729887024991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2542,
      "step": 19012
    },
    {
      "epoch": 0.6509072235535776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2305,
      "step": 19013
    },
    {
      "epoch": 0.6509414584046559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2483,
      "step": 19014
    },
    {
      "epoch": 0.6509756932557343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2675,
      "step": 19015
    },
    {
      "epoch": 0.6510099281068127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2417,
      "step": 19016
    },
    {
      "epoch": 0.6510441629578911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2479,
      "step": 19017
    },
    {
      "epoch": 0.6510783978089695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2598,
      "step": 19018
    },
    {
      "epoch": 0.6511126326600479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2578,
      "step": 19019
    },
    {
      "epoch": 0.6511468675111264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.24,
      "step": 19020
    },
    {
      "epoch": 0.6511811023622047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2505,
      "step": 19021
    },
    {
      "epoch": 0.6512153372132832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2415,
      "step": 19022
    },
    {
      "epoch": 0.6512495720643615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2745,
      "step": 19023
    },
    {
      "epoch": 0.65128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2569,
      "step": 19024
    },
    {
      "epoch": 0.6513180417665183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2605,
      "step": 19025
    },
    {
      "epoch": 0.6513522766175968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2563,
      "step": 19026
    },
    {
      "epoch": 0.6513865114686751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2413,
      "step": 19027
    },
    {
      "epoch": 0.6514207463197536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2438,
      "step": 19028
    },
    {
      "epoch": 0.6514549811708319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2421,
      "step": 19029
    },
    {
      "epoch": 0.6514892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2478,
      "step": 19030
    },
    {
      "epoch": 0.6515234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2604,
      "step": 19031
    },
    {
      "epoch": 0.6515576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.244,
      "step": 19032
    },
    {
      "epoch": 0.6515919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2549,
      "step": 19033
    },
    {
      "epoch": 0.6516261554262239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2447,
      "step": 19034
    },
    {
      "epoch": 0.6516603902773023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2523,
      "step": 19035
    },
    {
      "epoch": 0.6516946251283807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2659,
      "step": 19036
    },
    {
      "epoch": 0.6517288599794591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2306,
      "step": 19037
    },
    {
      "epoch": 0.6517630948305375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2498,
      "step": 19038
    },
    {
      "epoch": 0.6517973296816159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2327,
      "step": 19039
    },
    {
      "epoch": 0.6518315645326943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2465,
      "step": 19040
    },
    {
      "epoch": 0.6518657993837726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2593,
      "step": 19041
    },
    {
      "epoch": 0.6519000342348511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2543,
      "step": 19042
    },
    {
      "epoch": 0.6519342690859294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2413,
      "step": 19043
    },
    {
      "epoch": 0.6519685039370079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2357,
      "step": 19044
    },
    {
      "epoch": 0.6520027387880862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2407,
      "step": 19045
    },
    {
      "epoch": 0.6520369736391647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2453,
      "step": 19046
    },
    {
      "epoch": 0.652071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2539,
      "step": 19047
    },
    {
      "epoch": 0.6521054433413215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2399,
      "step": 19048
    },
    {
      "epoch": 0.6521396781923998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2605,
      "step": 19049
    },
    {
      "epoch": 0.6521739130434783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2468,
      "step": 19050
    },
    {
      "epoch": 0.6522081478945566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2521,
      "step": 19051
    },
    {
      "epoch": 0.6522423827456351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2673,
      "step": 19052
    },
    {
      "epoch": 0.6522766175967134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2531,
      "step": 19053
    },
    {
      "epoch": 0.6523108524477919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2488,
      "step": 19054
    },
    {
      "epoch": 0.6523450872988702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2364,
      "step": 19055
    },
    {
      "epoch": 0.6523793221499486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2506,
      "step": 19056
    },
    {
      "epoch": 0.6524135570010271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2402,
      "step": 19057
    },
    {
      "epoch": 0.6524477918521054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2476,
      "step": 19058
    },
    {
      "epoch": 0.6524820267031839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2463,
      "step": 19059
    },
    {
      "epoch": 0.6525162615542622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2505,
      "step": 19060
    },
    {
      "epoch": 0.6525504964053407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2564,
      "step": 19061
    },
    {
      "epoch": 0.652584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2384,
      "step": 19062
    },
    {
      "epoch": 0.6526189661074975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2358,
      "step": 19063
    },
    {
      "epoch": 0.6526532009585758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.247,
      "step": 19064
    },
    {
      "epoch": 0.6526874358096543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2429,
      "step": 19065
    },
    {
      "epoch": 0.6527216706607326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2495,
      "step": 19066
    },
    {
      "epoch": 0.6527559055118111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2531,
      "step": 19067
    },
    {
      "epoch": 0.6527901403628894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.225,
      "step": 19068
    },
    {
      "epoch": 0.6528243752139679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2392,
      "step": 19069
    },
    {
      "epoch": 0.6528586100650462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2471,
      "step": 19070
    },
    {
      "epoch": 0.6528928449161246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2463,
      "step": 19071
    },
    {
      "epoch": 0.652927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2626,
      "step": 19072
    },
    {
      "epoch": 0.6529613146182814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2525,
      "step": 19073
    },
    {
      "epoch": 0.6529955494693598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2282,
      "step": 19074
    },
    {
      "epoch": 0.6530297843204382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2322,
      "step": 19075
    },
    {
      "epoch": 0.6530640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2492,
      "step": 19076
    },
    {
      "epoch": 0.653098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2567,
      "step": 19077
    },
    {
      "epoch": 0.6531324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2558,
      "step": 19078
    },
    {
      "epoch": 0.6531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.261,
      "step": 19079
    },
    {
      "epoch": 0.6532009585758302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2527,
      "step": 19080
    },
    {
      "epoch": 0.6532351934269086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2535,
      "step": 19081
    },
    {
      "epoch": 0.653269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2388,
      "step": 19082
    },
    {
      "epoch": 0.6533036631290654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2396,
      "step": 19083
    },
    {
      "epoch": 0.6533378979801437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.245,
      "step": 19084
    },
    {
      "epoch": 0.6533721328312222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2695,
      "step": 19085
    },
    {
      "epoch": 0.6534063676823005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2285,
      "step": 19086
    },
    {
      "epoch": 0.653440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2423,
      "step": 19087
    },
    {
      "epoch": 0.6534748373844573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2468,
      "step": 19088
    },
    {
      "epoch": 0.6535090722355358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2441,
      "step": 19089
    },
    {
      "epoch": 0.6535433070866141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2593,
      "step": 19090
    },
    {
      "epoch": 0.6535775419376926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2398,
      "step": 19091
    },
    {
      "epoch": 0.6536117767887709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2527,
      "step": 19092
    },
    {
      "epoch": 0.6536460116398494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2438,
      "step": 19093
    },
    {
      "epoch": 0.6536802464909278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2527,
      "step": 19094
    },
    {
      "epoch": 0.6537144813420062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2504,
      "step": 19095
    },
    {
      "epoch": 0.6537487161930846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2556,
      "step": 19096
    },
    {
      "epoch": 0.653782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.251,
      "step": 19097
    },
    {
      "epoch": 0.6538171858952414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2619,
      "step": 19098
    },
    {
      "epoch": 0.6538514207463197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2478,
      "step": 19099
    },
    {
      "epoch": 0.6538856555973982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2363,
      "step": 19100
    },
    {
      "epoch": 0.6539198904484765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2673,
      "step": 19101
    },
    {
      "epoch": 0.653954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2476,
      "step": 19102
    },
    {
      "epoch": 0.6539883601506333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2363,
      "step": 19103
    },
    {
      "epoch": 0.6540225950017118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2614,
      "step": 19104
    },
    {
      "epoch": 0.6540568298527901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2442,
      "step": 19105
    },
    {
      "epoch": 0.6540910647038686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2364,
      "step": 19106
    },
    {
      "epoch": 0.6541252995549469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2343,
      "step": 19107
    },
    {
      "epoch": 0.6541595344060254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2437,
      "step": 19108
    },
    {
      "epoch": 0.6541937692571037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2445,
      "step": 19109
    },
    {
      "epoch": 0.6542280041081822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2303,
      "step": 19110
    },
    {
      "epoch": 0.6542622389592605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2455,
      "step": 19111
    },
    {
      "epoch": 0.654296473810339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2375,
      "step": 19112
    },
    {
      "epoch": 0.6543307086614173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2512,
      "step": 19113
    },
    {
      "epoch": 0.6543649435124957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.243,
      "step": 19114
    },
    {
      "epoch": 0.6543991783635741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2257,
      "step": 19115
    },
    {
      "epoch": 0.6544334132146525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2374,
      "step": 19116
    },
    {
      "epoch": 0.6544676480657309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2605,
      "step": 19117
    },
    {
      "epoch": 0.6545018829168093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.247,
      "step": 19118
    },
    {
      "epoch": 0.6545361177678877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2476,
      "step": 19119
    },
    {
      "epoch": 0.6545703526189661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2623,
      "step": 19120
    },
    {
      "epoch": 0.6546045874700445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2615,
      "step": 19121
    },
    {
      "epoch": 0.6546388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.264,
      "step": 19122
    },
    {
      "epoch": 0.6546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.237,
      "step": 19123
    },
    {
      "epoch": 0.6547072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2384,
      "step": 19124
    },
    {
      "epoch": 0.654741526874358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2352,
      "step": 19125
    },
    {
      "epoch": 0.6547757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2329,
      "step": 19126
    },
    {
      "epoch": 0.6548099965765148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.249,
      "step": 19127
    },
    {
      "epoch": 0.6548442314275933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2609,
      "step": 19128
    },
    {
      "epoch": 0.6548784662786716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2593,
      "step": 19129
    },
    {
      "epoch": 0.6549127011297501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2362,
      "step": 19130
    },
    {
      "epoch": 0.6549469359808285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2401,
      "step": 19131
    },
    {
      "epoch": 0.6549811708319069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2338,
      "step": 19132
    },
    {
      "epoch": 0.6550154056829853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2372,
      "step": 19133
    },
    {
      "epoch": 0.6550496405340637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2321,
      "step": 19134
    },
    {
      "epoch": 0.6550838753851421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2471,
      "step": 19135
    },
    {
      "epoch": 0.6551181102362205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2636,
      "step": 19136
    },
    {
      "epoch": 0.6551523450872989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2437,
      "step": 19137
    },
    {
      "epoch": 0.6551865799383773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2227,
      "step": 19138
    },
    {
      "epoch": 0.6552208147894557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2322,
      "step": 19139
    },
    {
      "epoch": 0.655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2471,
      "step": 19140
    },
    {
      "epoch": 0.6552892844916125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2494,
      "step": 19141
    },
    {
      "epoch": 0.6553235193426908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2587,
      "step": 19142
    },
    {
      "epoch": 0.6553577541937693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2543,
      "step": 19143
    },
    {
      "epoch": 0.6553919890448476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.251,
      "step": 19144
    },
    {
      "epoch": 0.6554262238959261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2564,
      "step": 19145
    },
    {
      "epoch": 0.6554604587470044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2368,
      "step": 19146
    },
    {
      "epoch": 0.6554946935980829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2557,
      "step": 19147
    },
    {
      "epoch": 0.6555289284491612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2489,
      "step": 19148
    },
    {
      "epoch": 0.6555631633002397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2498,
      "step": 19149
    },
    {
      "epoch": 0.655597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2314,
      "step": 19150
    },
    {
      "epoch": 0.6556316330023965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2554,
      "step": 19151
    },
    {
      "epoch": 0.6556658678534748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2482,
      "step": 19152
    },
    {
      "epoch": 0.6557001027045533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2286,
      "step": 19153
    },
    {
      "epoch": 0.6557343375556316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2471,
      "step": 19154
    },
    {
      "epoch": 0.65576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2553,
      "step": 19155
    },
    {
      "epoch": 0.6558028072577884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2537,
      "step": 19156
    },
    {
      "epoch": 0.6558370421088668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2516,
      "step": 19157
    },
    {
      "epoch": 0.6558712769599452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2614,
      "step": 19158
    },
    {
      "epoch": 0.6559055118110236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2523,
      "step": 19159
    },
    {
      "epoch": 0.655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2448,
      "step": 19160
    },
    {
      "epoch": 0.6559739815131804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2392,
      "step": 19161
    },
    {
      "epoch": 0.6560082163642588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2513,
      "step": 19162
    },
    {
      "epoch": 0.6560424512153372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2382,
      "step": 19163
    },
    {
      "epoch": 0.6560766860664156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.253,
      "step": 19164
    },
    {
      "epoch": 0.656110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2344,
      "step": 19165
    },
    {
      "epoch": 0.6561451557685724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2421,
      "step": 19166
    },
    {
      "epoch": 0.6561793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2434,
      "step": 19167
    },
    {
      "epoch": 0.6562136254707293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2319,
      "step": 19168
    },
    {
      "epoch": 0.6562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2573,
      "step": 19169
    },
    {
      "epoch": 0.656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2411,
      "step": 19170
    },
    {
      "epoch": 0.6563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2427,
      "step": 19171
    },
    {
      "epoch": 0.6563505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2378,
      "step": 19172
    },
    {
      "epoch": 0.6563847997261212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2581,
      "step": 19173
    },
    {
      "epoch": 0.6564190345771996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.236,
      "step": 19174
    },
    {
      "epoch": 0.656453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2672,
      "step": 19175
    },
    {
      "epoch": 0.6564875042793564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2397,
      "step": 19176
    },
    {
      "epoch": 0.6565217391304348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2339,
      "step": 19177
    },
    {
      "epoch": 0.6565559739815132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2662,
      "step": 19178
    },
    {
      "epoch": 0.6565902088325916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2305,
      "step": 19179
    },
    {
      "epoch": 0.65662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2348,
      "step": 19180
    },
    {
      "epoch": 0.6566586785347484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2312,
      "step": 19181
    },
    {
      "epoch": 0.6566929133858268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2645,
      "step": 19182
    },
    {
      "epoch": 0.6567271482369051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2559,
      "step": 19183
    },
    {
      "epoch": 0.6567613830879836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.255,
      "step": 19184
    },
    {
      "epoch": 0.6567956179390619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2343,
      "step": 19185
    },
    {
      "epoch": 0.6568298527901404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2463,
      "step": 19186
    },
    {
      "epoch": 0.6568640876412187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2401,
      "step": 19187
    },
    {
      "epoch": 0.6568983224922972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2486,
      "step": 19188
    },
    {
      "epoch": 0.6569325573433755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2433,
      "step": 19189
    },
    {
      "epoch": 0.656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2442,
      "step": 19190
    },
    {
      "epoch": 0.6570010270455323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2562,
      "step": 19191
    },
    {
      "epoch": 0.6570352618966108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.248,
      "step": 19192
    },
    {
      "epoch": 0.6570694967476891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2336,
      "step": 19193
    },
    {
      "epoch": 0.6571037315987676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2584,
      "step": 19194
    },
    {
      "epoch": 0.6571379664498459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2368,
      "step": 19195
    },
    {
      "epoch": 0.6571722013009244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2565,
      "step": 19196
    },
    {
      "epoch": 0.6572064361520027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2544,
      "step": 19197
    },
    {
      "epoch": 0.6572406710030811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2445,
      "step": 19198
    },
    {
      "epoch": 0.6572749058541595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2328,
      "step": 19199
    },
    {
      "epoch": 0.6573091407052379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2531,
      "step": 19200
    },
    {
      "epoch": 0.6573433755563163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.252,
      "step": 19201
    },
    {
      "epoch": 0.6573776104073947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2379,
      "step": 19202
    },
    {
      "epoch": 0.6574118452584731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2349,
      "step": 19203
    },
    {
      "epoch": 0.6574460801095515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2397,
      "step": 19204
    },
    {
      "epoch": 0.65748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2239,
      "step": 19205
    },
    {
      "epoch": 0.6575145498117083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2458,
      "step": 19206
    },
    {
      "epoch": 0.6575487846627868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2513,
      "step": 19207
    },
    {
      "epoch": 0.6575830195138651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2567,
      "step": 19208
    },
    {
      "epoch": 0.6576172543649436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2544,
      "step": 19209
    },
    {
      "epoch": 0.6576514892160219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.239,
      "step": 19210
    },
    {
      "epoch": 0.6576857240671004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2344,
      "step": 19211
    },
    {
      "epoch": 0.6577199589181787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2538,
      "step": 19212
    },
    {
      "epoch": 0.6577541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2377,
      "step": 19213
    },
    {
      "epoch": 0.6577884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2451,
      "step": 19214
    },
    {
      "epoch": 0.6578226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2319,
      "step": 19215
    },
    {
      "epoch": 0.6578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2427,
      "step": 19216
    },
    {
      "epoch": 0.6578911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2434,
      "step": 19217
    },
    {
      "epoch": 0.6579253680246491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2399,
      "step": 19218
    },
    {
      "epoch": 0.6579596028757275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2515,
      "step": 19219
    },
    {
      "epoch": 0.6579938377268059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2365,
      "step": 19220
    },
    {
      "epoch": 0.6580280725778843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2565,
      "step": 19221
    },
    {
      "epoch": 0.6580623074289627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.246,
      "step": 19222
    },
    {
      "epoch": 0.6580965422800411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2478,
      "step": 19223
    },
    {
      "epoch": 0.6581307771311194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2612,
      "step": 19224
    },
    {
      "epoch": 0.6581650119821979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2503,
      "step": 19225
    },
    {
      "epoch": 0.6581992468332762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2395,
      "step": 19226
    },
    {
      "epoch": 0.6582334816843547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.246,
      "step": 19227
    },
    {
      "epoch": 0.658267716535433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2391,
      "step": 19228
    },
    {
      "epoch": 0.6583019513865115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2435,
      "step": 19229
    },
    {
      "epoch": 0.6583361862375898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2307,
      "step": 19230
    },
    {
      "epoch": 0.6583704210886683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2375,
      "step": 19231
    },
    {
      "epoch": 0.6584046559397466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2396,
      "step": 19232
    },
    {
      "epoch": 0.6584388907908251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2516,
      "step": 19233
    },
    {
      "epoch": 0.6584731256419034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.262,
      "step": 19234
    },
    {
      "epoch": 0.6585073604929819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2488,
      "step": 19235
    },
    {
      "epoch": 0.6585415953440602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2401,
      "step": 19236
    },
    {
      "epoch": 0.6585758301951387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.254,
      "step": 19237
    },
    {
      "epoch": 0.658610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2432,
      "step": 19238
    },
    {
      "epoch": 0.6586442998972954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2286,
      "step": 19239
    },
    {
      "epoch": 0.6586785347483738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2492,
      "step": 19240
    },
    {
      "epoch": 0.6587127695994522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2372,
      "step": 19241
    },
    {
      "epoch": 0.6587470044505307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.239,
      "step": 19242
    },
    {
      "epoch": 0.658781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2427,
      "step": 19243
    },
    {
      "epoch": 0.6588154741526875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2376,
      "step": 19244
    },
    {
      "epoch": 0.6588497090037658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2467,
      "step": 19245
    },
    {
      "epoch": 0.6588839438548443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2397,
      "step": 19246
    },
    {
      "epoch": 0.6589181787059226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2436,
      "step": 19247
    },
    {
      "epoch": 0.6589524135570011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2381,
      "step": 19248
    },
    {
      "epoch": 0.6589866484080794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2342,
      "step": 19249
    },
    {
      "epoch": 0.6590208832591579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.249,
      "step": 19250
    },
    {
      "epoch": 0.6590551181102362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.242,
      "step": 19251
    },
    {
      "epoch": 0.6590893529613147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2221,
      "step": 19252
    },
    {
      "epoch": 0.659123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2296,
      "step": 19253
    },
    {
      "epoch": 0.6591578226634714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2273,
      "step": 19254
    },
    {
      "epoch": 0.6591920575145498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2431,
      "step": 19255
    },
    {
      "epoch": 0.6592262923656282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2444,
      "step": 19256
    },
    {
      "epoch": 0.6592605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2385,
      "step": 19257
    },
    {
      "epoch": 0.659294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2535,
      "step": 19258
    },
    {
      "epoch": 0.6593289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2336,
      "step": 19259
    },
    {
      "epoch": 0.6593632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2344,
      "step": 19260
    },
    {
      "epoch": 0.6593974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2482,
      "step": 19261
    },
    {
      "epoch": 0.6594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2227,
      "step": 19262
    },
    {
      "epoch": 0.659465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2471,
      "step": 19263
    },
    {
      "epoch": 0.6595001711742554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2303,
      "step": 19264
    },
    {
      "epoch": 0.6595344060253338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2318,
      "step": 19265
    },
    {
      "epoch": 0.6595686408764122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2454,
      "step": 19266
    },
    {
      "epoch": 0.6596028757274905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2335,
      "step": 19267
    },
    {
      "epoch": 0.659637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2444,
      "step": 19268
    },
    {
      "epoch": 0.6596713454296473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2419,
      "step": 19269
    },
    {
      "epoch": 0.6597055802807258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2402,
      "step": 19270
    },
    {
      "epoch": 0.6597398151318041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2568,
      "step": 19271
    },
    {
      "epoch": 0.6597740499828826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2576,
      "step": 19272
    },
    {
      "epoch": 0.6598082848339609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2396,
      "step": 19273
    },
    {
      "epoch": 0.6598425196850394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2296,
      "step": 19274
    },
    {
      "epoch": 0.6598767545361177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2442,
      "step": 19275
    },
    {
      "epoch": 0.6599109893871962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.26,
      "step": 19276
    },
    {
      "epoch": 0.6599452242382745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2505,
      "step": 19277
    },
    {
      "epoch": 0.659979459089353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2518,
      "step": 19278
    },
    {
      "epoch": 0.6600136939404314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2332,
      "step": 19279
    },
    {
      "epoch": 0.6600479287915098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2341,
      "step": 19280
    },
    {
      "epoch": 0.6600821636425882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2522,
      "step": 19281
    },
    {
      "epoch": 0.6601163984936665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2498,
      "step": 19282
    },
    {
      "epoch": 0.660150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2342,
      "step": 19283
    },
    {
      "epoch": 0.6601848681958233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2492,
      "step": 19284
    },
    {
      "epoch": 0.6602191030469018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2477,
      "step": 19285
    },
    {
      "epoch": 0.6602533378979801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2414,
      "step": 19286
    },
    {
      "epoch": 0.6602875727490586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2418,
      "step": 19287
    },
    {
      "epoch": 0.6603218076001369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2705,
      "step": 19288
    },
    {
      "epoch": 0.6603560424512154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2383,
      "step": 19289
    },
    {
      "epoch": 0.6603902773022937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2313,
      "step": 19290
    },
    {
      "epoch": 0.6604245121533722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2409,
      "step": 19291
    },
    {
      "epoch": 0.6604587470044505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2424,
      "step": 19292
    },
    {
      "epoch": 0.660492981855529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2477,
      "step": 19293
    },
    {
      "epoch": 0.6605272167066073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2545,
      "step": 19294
    },
    {
      "epoch": 0.6605614515576858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2423,
      "step": 19295
    },
    {
      "epoch": 0.6605956864087641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2302,
      "step": 19296
    },
    {
      "epoch": 0.6606299212598425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2385,
      "step": 19297
    },
    {
      "epoch": 0.6606641561109209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2474,
      "step": 19298
    },
    {
      "epoch": 0.6606983909619993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2362,
      "step": 19299
    },
    {
      "epoch": 0.6607326258130777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2495,
      "step": 19300
    },
    {
      "epoch": 0.6607668606641561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2385,
      "step": 19301
    },
    {
      "epoch": 0.6608010955152345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.233,
      "step": 19302
    },
    {
      "epoch": 0.6608353303663129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2477,
      "step": 19303
    },
    {
      "epoch": 0.6608695652173913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2474,
      "step": 19304
    },
    {
      "epoch": 0.6609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2461,
      "step": 19305
    },
    {
      "epoch": 0.6609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2313,
      "step": 19306
    },
    {
      "epoch": 0.6609722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2394,
      "step": 19307
    },
    {
      "epoch": 0.6610065046217048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2465,
      "step": 19308
    },
    {
      "epoch": 0.6610407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2593,
      "step": 19309
    },
    {
      "epoch": 0.6610749743238616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.272,
      "step": 19310
    },
    {
      "epoch": 0.6611092091749401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2467,
      "step": 19311
    },
    {
      "epoch": 0.6611434440260184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2636,
      "step": 19312
    },
    {
      "epoch": 0.6611776788770969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2323,
      "step": 19313
    },
    {
      "epoch": 0.6612119137281753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2484,
      "step": 19314
    },
    {
      "epoch": 0.6612461485792537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2367,
      "step": 19315
    },
    {
      "epoch": 0.6612803834303321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2514,
      "step": 19316
    },
    {
      "epoch": 0.6613146182814105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2424,
      "step": 19317
    },
    {
      "epoch": 0.6613488531324889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2369,
      "step": 19318
    },
    {
      "epoch": 0.6613830879835673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.246,
      "step": 19319
    },
    {
      "epoch": 0.6614173228346457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2541,
      "step": 19320
    },
    {
      "epoch": 0.6614515576857241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2659,
      "step": 19321
    },
    {
      "epoch": 0.6614857925368025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2428,
      "step": 19322
    },
    {
      "epoch": 0.6615200273878808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2418,
      "step": 19323
    },
    {
      "epoch": 0.6615542622389593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2346,
      "step": 19324
    },
    {
      "epoch": 0.6615884970900376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2392,
      "step": 19325
    },
    {
      "epoch": 0.6616227319411161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2415,
      "step": 19326
    },
    {
      "epoch": 0.6616569667921944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2384,
      "step": 19327
    },
    {
      "epoch": 0.6616912016432729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2586,
      "step": 19328
    },
    {
      "epoch": 0.6617254364943512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2342,
      "step": 19329
    },
    {
      "epoch": 0.6617596713454297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2659,
      "step": 19330
    },
    {
      "epoch": 0.661793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2406,
      "step": 19331
    },
    {
      "epoch": 0.6618281410475865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2473,
      "step": 19332
    },
    {
      "epoch": 0.6618623758986648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2432,
      "step": 19333
    },
    {
      "epoch": 0.6618966107497433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2372,
      "step": 19334
    },
    {
      "epoch": 0.6619308456008216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.224,
      "step": 19335
    },
    {
      "epoch": 0.6619650804519001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.238,
      "step": 19336
    },
    {
      "epoch": 0.6619993153029784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2422,
      "step": 19337
    },
    {
      "epoch": 0.6620335501540568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2402,
      "step": 19338
    },
    {
      "epoch": 0.6620677850051352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2412,
      "step": 19339
    },
    {
      "epoch": 0.6621020198562136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2434,
      "step": 19340
    },
    {
      "epoch": 0.662136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2561,
      "step": 19341
    },
    {
      "epoch": 0.6621704895583704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2386,
      "step": 19342
    },
    {
      "epoch": 0.6622047244094488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.234,
      "step": 19343
    },
    {
      "epoch": 0.6622389592605272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2585,
      "step": 19344
    },
    {
      "epoch": 0.6622731941116056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2525,
      "step": 19345
    },
    {
      "epoch": 0.662307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2319,
      "step": 19346
    },
    {
      "epoch": 0.6623416638137624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2553,
      "step": 19347
    },
    {
      "epoch": 0.6623758986648408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.266,
      "step": 19348
    },
    {
      "epoch": 0.6624101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2509,
      "step": 19349
    },
    {
      "epoch": 0.6624443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2492,
      "step": 19350
    },
    {
      "epoch": 0.6624786032180761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2415,
      "step": 19351
    },
    {
      "epoch": 0.6625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2344,
      "step": 19352
    },
    {
      "epoch": 0.6625470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2535,
      "step": 19353
    },
    {
      "epoch": 0.6625813077713112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.232,
      "step": 19354
    },
    {
      "epoch": 0.6626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2475,
      "step": 19355
    },
    {
      "epoch": 0.662649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2299,
      "step": 19356
    },
    {
      "epoch": 0.6626840123245464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2272,
      "step": 19357
    },
    {
      "epoch": 0.6627182471756248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2592,
      "step": 19358
    },
    {
      "epoch": 0.6627524820267032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2378,
      "step": 19359
    },
    {
      "epoch": 0.6627867168777816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2345,
      "step": 19360
    },
    {
      "epoch": 0.66282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2512,
      "step": 19361
    },
    {
      "epoch": 0.6628551865799384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2106,
      "step": 19362
    },
    {
      "epoch": 0.6628894214310168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.223,
      "step": 19363
    },
    {
      "epoch": 0.6629236562820952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2459,
      "step": 19364
    },
    {
      "epoch": 0.6629578911331736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2486,
      "step": 19365
    },
    {
      "epoch": 0.662992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2404,
      "step": 19366
    },
    {
      "epoch": 0.6630263608353304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2497,
      "step": 19367
    },
    {
      "epoch": 0.6630605956864087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2433,
      "step": 19368
    },
    {
      "epoch": 0.6630948305374872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2348,
      "step": 19369
    },
    {
      "epoch": 0.6631290653885655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2442,
      "step": 19370
    },
    {
      "epoch": 0.663163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.253,
      "step": 19371
    },
    {
      "epoch": 0.6631975350907223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2513,
      "step": 19372
    },
    {
      "epoch": 0.6632317699418008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2364,
      "step": 19373
    },
    {
      "epoch": 0.6632660047928791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2364,
      "step": 19374
    },
    {
      "epoch": 0.6633002396439576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.264,
      "step": 19375
    },
    {
      "epoch": 0.6633344744950359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.227,
      "step": 19376
    },
    {
      "epoch": 0.6633687093461144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2457,
      "step": 19377
    },
    {
      "epoch": 0.6634029441971927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2296,
      "step": 19378
    },
    {
      "epoch": 0.6634371790482712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2415,
      "step": 19379
    },
    {
      "epoch": 0.6634714138993495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2485,
      "step": 19380
    },
    {
      "epoch": 0.663505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2484,
      "step": 19381
    },
    {
      "epoch": 0.6635398836015063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2384,
      "step": 19382
    },
    {
      "epoch": 0.6635741184525847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2419,
      "step": 19383
    },
    {
      "epoch": 0.6636083533036631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2482,
      "step": 19384
    },
    {
      "epoch": 0.6636425881547415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2473,
      "step": 19385
    },
    {
      "epoch": 0.6636768230058199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2459,
      "step": 19386
    },
    {
      "epoch": 0.6637110578568983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2361,
      "step": 19387
    },
    {
      "epoch": 0.6637452927079768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2388,
      "step": 19388
    },
    {
      "epoch": 0.6637795275590551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2367,
      "step": 19389
    },
    {
      "epoch": 0.6638137624101336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2271,
      "step": 19390
    },
    {
      "epoch": 0.6638479972612119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2468,
      "step": 19391
    },
    {
      "epoch": 0.6638822321122904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2448,
      "step": 19392
    },
    {
      "epoch": 0.6639164669633687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2562,
      "step": 19393
    },
    {
      "epoch": 0.6639507018144472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2198,
      "step": 19394
    },
    {
      "epoch": 0.6639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2355,
      "step": 19395
    },
    {
      "epoch": 0.664019171516604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2401,
      "step": 19396
    },
    {
      "epoch": 0.6640534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2389,
      "step": 19397
    },
    {
      "epoch": 0.6640876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2339,
      "step": 19398
    },
    {
      "epoch": 0.6641218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2538,
      "step": 19399
    },
    {
      "epoch": 0.6641561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2405,
      "step": 19400
    },
    {
      "epoch": 0.6641903457719959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2406,
      "step": 19401
    },
    {
      "epoch": 0.6642245806230743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2365,
      "step": 19402
    },
    {
      "epoch": 0.6642588154741527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2503,
      "step": 19403
    },
    {
      "epoch": 0.6642930503252311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2564,
      "step": 19404
    },
    {
      "epoch": 0.6643272851763095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2382,
      "step": 19405
    },
    {
      "epoch": 0.6643615200273879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2247,
      "step": 19406
    },
    {
      "epoch": 0.6643957548784662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2373,
      "step": 19407
    },
    {
      "epoch": 0.6644299897295447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2418,
      "step": 19408
    },
    {
      "epoch": 0.664464224580623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2487,
      "step": 19409
    },
    {
      "epoch": 0.6644984594317015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2572,
      "step": 19410
    },
    {
      "epoch": 0.6645326942827798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2298,
      "step": 19411
    },
    {
      "epoch": 0.6645669291338583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2188,
      "step": 19412
    },
    {
      "epoch": 0.6646011639849366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2423,
      "step": 19413
    },
    {
      "epoch": 0.6646353988360151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2677,
      "step": 19414
    },
    {
      "epoch": 0.6646696336870934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2433,
      "step": 19415
    },
    {
      "epoch": 0.6647038685381719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2506,
      "step": 19416
    },
    {
      "epoch": 0.6647381033892502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2307,
      "step": 19417
    },
    {
      "epoch": 0.6647723382403287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2268,
      "step": 19418
    },
    {
      "epoch": 0.664806573091407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2412,
      "step": 19419
    },
    {
      "epoch": 0.6648408079424855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2513,
      "step": 19420
    },
    {
      "epoch": 0.6648750427935638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.238,
      "step": 19421
    },
    {
      "epoch": 0.6649092776446422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2202,
      "step": 19422
    },
    {
      "epoch": 0.6649435124957206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.252,
      "step": 19423
    },
    {
      "epoch": 0.664977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2368,
      "step": 19424
    },
    {
      "epoch": 0.6650119821978775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2649,
      "step": 19425
    },
    {
      "epoch": 0.6650462170489558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2365,
      "step": 19426
    },
    {
      "epoch": 0.6650804519000343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.251,
      "step": 19427
    },
    {
      "epoch": 0.6651146867511126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2439,
      "step": 19428
    },
    {
      "epoch": 0.6651489216021911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2664,
      "step": 19429
    },
    {
      "epoch": 0.6651831564532694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2459,
      "step": 19430
    },
    {
      "epoch": 0.6652173913043479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2188,
      "step": 19431
    },
    {
      "epoch": 0.6652516261554262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2364,
      "step": 19432
    },
    {
      "epoch": 0.6652858610065047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.254,
      "step": 19433
    },
    {
      "epoch": 0.665320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2387,
      "step": 19434
    },
    {
      "epoch": 0.6653543307086615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2339,
      "step": 19435
    },
    {
      "epoch": 0.6653885655597398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2307,
      "step": 19436
    },
    {
      "epoch": 0.6654228004108182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2416,
      "step": 19437
    },
    {
      "epoch": 0.6654570352618966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2546,
      "step": 19438
    },
    {
      "epoch": 0.665491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2614,
      "step": 19439
    },
    {
      "epoch": 0.6655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2444,
      "step": 19440
    },
    {
      "epoch": 0.6655597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.231,
      "step": 19441
    },
    {
      "epoch": 0.6655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2325,
      "step": 19442
    },
    {
      "epoch": 0.6656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2364,
      "step": 19443
    },
    {
      "epoch": 0.665662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2538,
      "step": 19444
    },
    {
      "epoch": 0.6656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2377,
      "step": 19445
    },
    {
      "epoch": 0.6657309140705238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2281,
      "step": 19446
    },
    {
      "epoch": 0.6657651489216022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2325,
      "step": 19447
    },
    {
      "epoch": 0.6657993837726806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2385,
      "step": 19448
    },
    {
      "epoch": 0.665833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2327,
      "step": 19449
    },
    {
      "epoch": 0.6658678534748373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2302,
      "step": 19450
    },
    {
      "epoch": 0.6659020883259158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2124,
      "step": 19451
    },
    {
      "epoch": 0.6659363231769941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2309,
      "step": 19452
    },
    {
      "epoch": 0.6659705580280726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2468,
      "step": 19453
    },
    {
      "epoch": 0.6660047928791509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2275,
      "step": 19454
    },
    {
      "epoch": 0.6660390277302294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2316,
      "step": 19455
    },
    {
      "epoch": 0.6660732625813077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2454,
      "step": 19456
    },
    {
      "epoch": 0.6661074974323862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2324,
      "step": 19457
    },
    {
      "epoch": 0.6661417322834645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2666,
      "step": 19458
    },
    {
      "epoch": 0.666175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2553,
      "step": 19459
    },
    {
      "epoch": 0.6662102019856213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2366,
      "step": 19460
    },
    {
      "epoch": 0.6662444368366998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2319,
      "step": 19461
    },
    {
      "epoch": 0.6662786716877782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.232,
      "step": 19462
    },
    {
      "epoch": 0.6663129065388566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2515,
      "step": 19463
    },
    {
      "epoch": 0.666347141389935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2524,
      "step": 19464
    },
    {
      "epoch": 0.6663813762410133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2273,
      "step": 19465
    },
    {
      "epoch": 0.6664156110920918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2452,
      "step": 19466
    },
    {
      "epoch": 0.6664498459431701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2303,
      "step": 19467
    },
    {
      "epoch": 0.6664840807942486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2411,
      "step": 19468
    },
    {
      "epoch": 0.6665183156453269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2398,
      "step": 19469
    },
    {
      "epoch": 0.6665525504964054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2333,
      "step": 19470
    },
    {
      "epoch": 0.6665867853474837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2266,
      "step": 19471
    },
    {
      "epoch": 0.6666210201985622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2437,
      "step": 19472
    },
    {
      "epoch": 0.6666552550496405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2429,
      "step": 19473
    },
    {
      "epoch": 0.666689489900719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2444,
      "step": 19474
    },
    {
      "epoch": 0.6667237247517973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2462,
      "step": 19475
    },
    {
      "epoch": 0.6667579596028758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2314,
      "step": 19476
    },
    {
      "epoch": 0.6667921944539541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2656,
      "step": 19477
    },
    {
      "epoch": 0.6668264293050326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2276,
      "step": 19478
    },
    {
      "epoch": 0.6668606641561109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.256,
      "step": 19479
    },
    {
      "epoch": 0.6668948990071893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2106,
      "step": 19480
    },
    {
      "epoch": 0.6669291338582677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2371,
      "step": 19481
    },
    {
      "epoch": 0.6669633687093461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2394,
      "step": 19482
    },
    {
      "epoch": 0.6669976035604245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2402,
      "step": 19483
    },
    {
      "epoch": 0.6670318384115029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2343,
      "step": 19484
    },
    {
      "epoch": 0.6670660732625813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2274,
      "step": 19485
    },
    {
      "epoch": 0.6671003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2255,
      "step": 19486
    },
    {
      "epoch": 0.6671345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2218,
      "step": 19487
    },
    {
      "epoch": 0.6671687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2415,
      "step": 19488
    },
    {
      "epoch": 0.6672030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2271,
      "step": 19489
    },
    {
      "epoch": 0.6672372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.222,
      "step": 19490
    },
    {
      "epoch": 0.6672714823690516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2207,
      "step": 19491
    },
    {
      "epoch": 0.6673057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2404,
      "step": 19492
    },
    {
      "epoch": 0.6673399520712084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2486,
      "step": 19493
    },
    {
      "epoch": 0.6673741869222869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2365,
      "step": 19494
    },
    {
      "epoch": 0.6674084217733652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2381,
      "step": 19495
    },
    {
      "epoch": 0.6674426566244437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2291,
      "step": 19496
    },
    {
      "epoch": 0.667476891475522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2318,
      "step": 19497
    },
    {
      "epoch": 0.6675111263266005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2327,
      "step": 19498
    },
    {
      "epoch": 0.6675453611776789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2402,
      "step": 19499
    },
    {
      "epoch": 0.6675795960287573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.244,
      "step": 19500
    },
    {
      "epoch": 0.6676138308798357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2195,
      "step": 19501
    },
    {
      "epoch": 0.6676480657309141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2368,
      "step": 19502
    },
    {
      "epoch": 0.6676823005819925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2387,
      "step": 19503
    },
    {
      "epoch": 0.6677165354330709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2204,
      "step": 19504
    },
    {
      "epoch": 0.6677507702841493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.239,
      "step": 19505
    },
    {
      "epoch": 0.6677850051352276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2288,
      "step": 19506
    },
    {
      "epoch": 0.6678192399863061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2406,
      "step": 19507
    },
    {
      "epoch": 0.6678534748373844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2318,
      "step": 19508
    },
    {
      "epoch": 0.6678877096884629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2368,
      "step": 19509
    },
    {
      "epoch": 0.6679219445395412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2315,
      "step": 19510
    },
    {
      "epoch": 0.6679561793906197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2424,
      "step": 19511
    },
    {
      "epoch": 0.667990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2245,
      "step": 19512
    },
    {
      "epoch": 0.6680246490927765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2393,
      "step": 19513
    },
    {
      "epoch": 0.6680588839438548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2475,
      "step": 19514
    },
    {
      "epoch": 0.6680931187949333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2543,
      "step": 19515
    },
    {
      "epoch": 0.6681273536460116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.238,
      "step": 19516
    },
    {
      "epoch": 0.6681615884970901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2513,
      "step": 19517
    },
    {
      "epoch": 0.6681958233481684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2377,
      "step": 19518
    },
    {
      "epoch": 0.6682300581992469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2526,
      "step": 19519
    },
    {
      "epoch": 0.6682642930503252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2314,
      "step": 19520
    },
    {
      "epoch": 0.6682985279014036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2418,
      "step": 19521
    },
    {
      "epoch": 0.668332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2517,
      "step": 19522
    },
    {
      "epoch": 0.6683669976035604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2459,
      "step": 19523
    },
    {
      "epoch": 0.6684012324546388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2397,
      "step": 19524
    },
    {
      "epoch": 0.6684354673057172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2242,
      "step": 19525
    },
    {
      "epoch": 0.6684697021567956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2419,
      "step": 19526
    },
    {
      "epoch": 0.668503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2474,
      "step": 19527
    },
    {
      "epoch": 0.6685381718589524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2463,
      "step": 19528
    },
    {
      "epoch": 0.6685724067100308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2331,
      "step": 19529
    },
    {
      "epoch": 0.6686066415611092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2327,
      "step": 19530
    },
    {
      "epoch": 0.6686408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.224,
      "step": 19531
    },
    {
      "epoch": 0.668675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2272,
      "step": 19532
    },
    {
      "epoch": 0.6687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2451,
      "step": 19533
    },
    {
      "epoch": 0.6687435809654227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.232,
      "step": 19534
    },
    {
      "epoch": 0.6687778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2129,
      "step": 19535
    },
    {
      "epoch": 0.6688120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2369,
      "step": 19536
    },
    {
      "epoch": 0.668846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2415,
      "step": 19537
    },
    {
      "epoch": 0.6688805203697364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2406,
      "step": 19538
    },
    {
      "epoch": 0.6689147552208148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2431,
      "step": 19539
    },
    {
      "epoch": 0.6689489900718932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.243,
      "step": 19540
    },
    {
      "epoch": 0.6689832249229716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2277,
      "step": 19541
    },
    {
      "epoch": 0.66901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2394,
      "step": 19542
    },
    {
      "epoch": 0.6690516946251284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2383,
      "step": 19543
    },
    {
      "epoch": 0.6690859294762068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2339,
      "step": 19544
    },
    {
      "epoch": 0.6691201643272852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.244,
      "step": 19545
    },
    {
      "epoch": 0.6691543991783636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2114,
      "step": 19546
    },
    {
      "epoch": 0.669188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2496,
      "step": 19547
    },
    {
      "epoch": 0.6692228688805204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2355,
      "step": 19548
    },
    {
      "epoch": 0.6692571037315987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2393,
      "step": 19549
    },
    {
      "epoch": 0.6692913385826772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2456,
      "step": 19550
    },
    {
      "epoch": 0.6693255734337555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2361,
      "step": 19551
    },
    {
      "epoch": 0.669359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2358,
      "step": 19552
    },
    {
      "epoch": 0.6693940431359123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2596,
      "step": 19553
    },
    {
      "epoch": 0.6694282779869908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2274,
      "step": 19554
    },
    {
      "epoch": 0.6694625128380691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2157,
      "step": 19555
    },
    {
      "epoch": 0.6694967476891476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2406,
      "step": 19556
    },
    {
      "epoch": 0.6695309825402259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2267,
      "step": 19557
    },
    {
      "epoch": 0.6695652173913044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2395,
      "step": 19558
    },
    {
      "epoch": 0.6695994522423827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2309,
      "step": 19559
    },
    {
      "epoch": 0.6696336870934612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2214,
      "step": 19560
    },
    {
      "epoch": 0.6696679219445395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2479,
      "step": 19561
    },
    {
      "epoch": 0.669702156795618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2345,
      "step": 19562
    },
    {
      "epoch": 0.6697363916466963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2241,
      "step": 19563
    },
    {
      "epoch": 0.6697706264977747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2373,
      "step": 19564
    },
    {
      "epoch": 0.6698048613488531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2515,
      "step": 19565
    },
    {
      "epoch": 0.6698390961999315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2483,
      "step": 19566
    },
    {
      "epoch": 0.6698733310510099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2261,
      "step": 19567
    },
    {
      "epoch": 0.6699075659020883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2264,
      "step": 19568
    },
    {
      "epoch": 0.6699418007531667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.237,
      "step": 19569
    },
    {
      "epoch": 0.6699760356042451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2364,
      "step": 19570
    },
    {
      "epoch": 0.6700102704553235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2497,
      "step": 19571
    },
    {
      "epoch": 0.6700445053064019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2505,
      "step": 19572
    },
    {
      "epoch": 0.6700787401574804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2352,
      "step": 19573
    },
    {
      "epoch": 0.6701129750085587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2351,
      "step": 19574
    },
    {
      "epoch": 0.6701472098596372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2319,
      "step": 19575
    },
    {
      "epoch": 0.6701814447107155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2433,
      "step": 19576
    },
    {
      "epoch": 0.670215679561794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2416,
      "step": 19577
    },
    {
      "epoch": 0.6702499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2264,
      "step": 19578
    },
    {
      "epoch": 0.6702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2446,
      "step": 19579
    },
    {
      "epoch": 0.6703183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2234,
      "step": 19580
    },
    {
      "epoch": 0.6703526189661075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2334,
      "step": 19581
    },
    {
      "epoch": 0.6703868538171859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2311,
      "step": 19582
    },
    {
      "epoch": 0.6704210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2297,
      "step": 19583
    },
    {
      "epoch": 0.6704553235193427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2501,
      "step": 19584
    },
    {
      "epoch": 0.6704895583704211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2234,
      "step": 19585
    },
    {
      "epoch": 0.6705237932214995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2492,
      "step": 19586
    },
    {
      "epoch": 0.6705580280725779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2134,
      "step": 19587
    },
    {
      "epoch": 0.6705922629236563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2473,
      "step": 19588
    },
    {
      "epoch": 0.6706264977747347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2235,
      "step": 19589
    },
    {
      "epoch": 0.670660732625813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2324,
      "step": 19590
    },
    {
      "epoch": 0.6706949674768915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2434,
      "step": 19591
    },
    {
      "epoch": 0.6707292023279698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2562,
      "step": 19592
    },
    {
      "epoch": 0.6707634371790483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2455,
      "step": 19593
    },
    {
      "epoch": 0.6707976720301266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2319,
      "step": 19594
    },
    {
      "epoch": 0.6708319068812051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2517,
      "step": 19595
    },
    {
      "epoch": 0.6708661417322834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2311,
      "step": 19596
    },
    {
      "epoch": 0.6709003765833619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.244,
      "step": 19597
    },
    {
      "epoch": 0.6709346114344402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2464,
      "step": 19598
    },
    {
      "epoch": 0.6709688462855187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2528,
      "step": 19599
    },
    {
      "epoch": 0.671003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2629,
      "step": 19600
    },
    {
      "epoch": 0.6710373159876755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2448,
      "step": 19601
    },
    {
      "epoch": 0.6710715508387538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2376,
      "step": 19602
    },
    {
      "epoch": 0.6711057856898323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2484,
      "step": 19603
    },
    {
      "epoch": 0.6711400205409106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2296,
      "step": 19604
    },
    {
      "epoch": 0.671174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2314,
      "step": 19605
    },
    {
      "epoch": 0.6712084902430674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2415,
      "step": 19606
    },
    {
      "epoch": 0.6712427250941458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2288,
      "step": 19607
    },
    {
      "epoch": 0.6712769599452242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2275,
      "step": 19608
    },
    {
      "epoch": 0.6713111947963026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2287,
      "step": 19609
    },
    {
      "epoch": 0.6713454296473811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2343,
      "step": 19610
    },
    {
      "epoch": 0.6713796644984594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2426,
      "step": 19611
    },
    {
      "epoch": 0.6714138993495379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2326,
      "step": 19612
    },
    {
      "epoch": 0.6714481342006162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2362,
      "step": 19613
    },
    {
      "epoch": 0.6714823690516947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2187,
      "step": 19614
    },
    {
      "epoch": 0.671516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2506,
      "step": 19615
    },
    {
      "epoch": 0.6715508387538515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.23,
      "step": 19616
    },
    {
      "epoch": 0.6715850736049298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2399,
      "step": 19617
    },
    {
      "epoch": 0.6716193084560083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2356,
      "step": 19618
    },
    {
      "epoch": 0.6716535433070866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2204,
      "step": 19619
    },
    {
      "epoch": 0.671687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2136,
      "step": 19620
    },
    {
      "epoch": 0.6717220130092434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2251,
      "step": 19621
    },
    {
      "epoch": 0.6717562478603218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.223,
      "step": 19622
    },
    {
      "epoch": 0.6717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2322,
      "step": 19623
    },
    {
      "epoch": 0.6718247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2418,
      "step": 19624
    },
    {
      "epoch": 0.671858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2291,
      "step": 19625
    },
    {
      "epoch": 0.6718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2233,
      "step": 19626
    },
    {
      "epoch": 0.6719274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2307,
      "step": 19627
    },
    {
      "epoch": 0.6719616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2479,
      "step": 19628
    },
    {
      "epoch": 0.6719958918178706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2237,
      "step": 19629
    },
    {
      "epoch": 0.672030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2276,
      "step": 19630
    },
    {
      "epoch": 0.6720643615200274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2438,
      "step": 19631
    },
    {
      "epoch": 0.6720985963711058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2214,
      "step": 19632
    },
    {
      "epoch": 0.6721328312221841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2359,
      "step": 19633
    },
    {
      "epoch": 0.6721670660732626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2449,
      "step": 19634
    },
    {
      "epoch": 0.6722013009243409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2288,
      "step": 19635
    },
    {
      "epoch": 0.6722355357754194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2241,
      "step": 19636
    },
    {
      "epoch": 0.6722697706264977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2218,
      "step": 19637
    },
    {
      "epoch": 0.6723040054775762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2207,
      "step": 19638
    },
    {
      "epoch": 0.6723382403286545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2353,
      "step": 19639
    },
    {
      "epoch": 0.672372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2186,
      "step": 19640
    },
    {
      "epoch": 0.6724067100308113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2304,
      "step": 19641
    },
    {
      "epoch": 0.6724409448818898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.234,
      "step": 19642
    },
    {
      "epoch": 0.6724751797329681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2387,
      "step": 19643
    },
    {
      "epoch": 0.6725094145840466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2317,
      "step": 19644
    },
    {
      "epoch": 0.6725436494351249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.248,
      "step": 19645
    },
    {
      "epoch": 0.6725778842862034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2389,
      "step": 19646
    },
    {
      "epoch": 0.6726121191372818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.25,
      "step": 19647
    },
    {
      "epoch": 0.6726463539883601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.23,
      "step": 19648
    },
    {
      "epoch": 0.6726805888394386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2322,
      "step": 19649
    },
    {
      "epoch": 0.6727148236905169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2253,
      "step": 19650
    },
    {
      "epoch": 0.6727490585415954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2321,
      "step": 19651
    },
    {
      "epoch": 0.6727832933926737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2413,
      "step": 19652
    },
    {
      "epoch": 0.6728175282437522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2156,
      "step": 19653
    },
    {
      "epoch": 0.6728517630948305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2528,
      "step": 19654
    },
    {
      "epoch": 0.672885997945909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2219,
      "step": 19655
    },
    {
      "epoch": 0.6729202327969873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.232,
      "step": 19656
    },
    {
      "epoch": 0.6729544676480658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.241,
      "step": 19657
    },
    {
      "epoch": 0.6729887024991441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2411,
      "step": 19658
    },
    {
      "epoch": 0.6730229373502226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2334,
      "step": 19659
    },
    {
      "epoch": 0.6730571722013009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2241,
      "step": 19660
    },
    {
      "epoch": 0.6730914070523794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2172,
      "step": 19661
    },
    {
      "epoch": 0.6731256419034577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2145,
      "step": 19662
    },
    {
      "epoch": 0.6731598767545361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2243,
      "step": 19663
    },
    {
      "epoch": 0.6731941116056145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2336,
      "step": 19664
    },
    {
      "epoch": 0.6732283464566929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.235,
      "step": 19665
    },
    {
      "epoch": 0.6732625813077713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2621,
      "step": 19666
    },
    {
      "epoch": 0.6732968161588497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2373,
      "step": 19667
    },
    {
      "epoch": 0.6733310510099281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2311,
      "step": 19668
    },
    {
      "epoch": 0.6733652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.229,
      "step": 19669
    },
    {
      "epoch": 0.6733995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2336,
      "step": 19670
    },
    {
      "epoch": 0.6734337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.213,
      "step": 19671
    },
    {
      "epoch": 0.6734679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2466,
      "step": 19672
    },
    {
      "epoch": 0.6735022252653201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2212,
      "step": 19673
    },
    {
      "epoch": 0.6735364601163984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2391,
      "step": 19674
    },
    {
      "epoch": 0.6735706949674769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2307,
      "step": 19675
    },
    {
      "epoch": 0.6736049298185552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2239,
      "step": 19676
    },
    {
      "epoch": 0.6736391646696337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2365,
      "step": 19677
    },
    {
      "epoch": 0.673673399520712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2246,
      "step": 19678
    },
    {
      "epoch": 0.6737076343717905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2487,
      "step": 19679
    },
    {
      "epoch": 0.6737418692228688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2431,
      "step": 19680
    },
    {
      "epoch": 0.6737761040739473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2306,
      "step": 19681
    },
    {
      "epoch": 0.6738103389250257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2419,
      "step": 19682
    },
    {
      "epoch": 0.6738445737761041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.228,
      "step": 19683
    },
    {
      "epoch": 0.6738788086271825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2304,
      "step": 19684
    },
    {
      "epoch": 0.6739130434782609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2189,
      "step": 19685
    },
    {
      "epoch": 0.6739472783293393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2286,
      "step": 19686
    },
    {
      "epoch": 0.6739815131804177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2172,
      "step": 19687
    },
    {
      "epoch": 0.6740157480314961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2522,
      "step": 19688
    },
    {
      "epoch": 0.6740499828825744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2248,
      "step": 19689
    },
    {
      "epoch": 0.6740842177336529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2412,
      "step": 19690
    },
    {
      "epoch": 0.6741184525847312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2405,
      "step": 19691
    },
    {
      "epoch": 0.6741526874358097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2287,
      "step": 19692
    },
    {
      "epoch": 0.674186922286888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2253,
      "step": 19693
    },
    {
      "epoch": 0.6742211571379665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2165,
      "step": 19694
    },
    {
      "epoch": 0.6742553919890448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2214,
      "step": 19695
    },
    {
      "epoch": 0.6742896268401233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2458,
      "step": 19696
    },
    {
      "epoch": 0.6743238616912016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2408,
      "step": 19697
    },
    {
      "epoch": 0.6743580965422801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2311,
      "step": 19698
    },
    {
      "epoch": 0.6743923313933584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2081,
      "step": 19699
    },
    {
      "epoch": 0.6744265662444369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2316,
      "step": 19700
    },
    {
      "epoch": 0.6744608010955152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.231,
      "step": 19701
    },
    {
      "epoch": 0.6744950359465937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2459,
      "step": 19702
    },
    {
      "epoch": 0.674529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2485,
      "step": 19703
    },
    {
      "epoch": 0.6745635056487504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2219,
      "step": 19704
    },
    {
      "epoch": 0.6745977404998288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2456,
      "step": 19705
    },
    {
      "epoch": 0.6746319753509072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2388,
      "step": 19706
    },
    {
      "epoch": 0.6746662102019856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2376,
      "step": 19707
    },
    {
      "epoch": 0.674700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2321,
      "step": 19708
    },
    {
      "epoch": 0.6747346799041424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2024,
      "step": 19709
    },
    {
      "epoch": 0.6747689147552208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2347,
      "step": 19710
    },
    {
      "epoch": 0.6748031496062992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2132,
      "step": 19711
    },
    {
      "epoch": 0.6748373844573776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2248,
      "step": 19712
    },
    {
      "epoch": 0.674871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2306,
      "step": 19713
    },
    {
      "epoch": 0.6749058541595344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2376,
      "step": 19714
    },
    {
      "epoch": 0.6749400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2439,
      "step": 19715
    },
    {
      "epoch": 0.6749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2362,
      "step": 19716
    },
    {
      "epoch": 0.6750085587127695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2199,
      "step": 19717
    },
    {
      "epoch": 0.675042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2181,
      "step": 19718
    },
    {
      "epoch": 0.6750770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2371,
      "step": 19719
    },
    {
      "epoch": 0.6751112632660048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2182,
      "step": 19720
    },
    {
      "epoch": 0.6751454981170832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2375,
      "step": 19721
    },
    {
      "epoch": 0.6751797329681616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2443,
      "step": 19722
    },
    {
      "epoch": 0.67521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.249,
      "step": 19723
    },
    {
      "epoch": 0.6752482026703184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2253,
      "step": 19724
    },
    {
      "epoch": 0.6752824375213968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2262,
      "step": 19725
    },
    {
      "epoch": 0.6753166723724752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2343,
      "step": 19726
    },
    {
      "epoch": 0.6753509072235536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2484,
      "step": 19727
    },
    {
      "epoch": 0.675385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2149,
      "step": 19728
    },
    {
      "epoch": 0.6754193769257104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2154,
      "step": 19729
    },
    {
      "epoch": 0.6754536117767888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2367,
      "step": 19730
    },
    {
      "epoch": 0.6754878466278672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2066,
      "step": 19731
    },
    {
      "epoch": 0.6755220814789455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2358,
      "step": 19732
    },
    {
      "epoch": 0.675556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2373,
      "step": 19733
    },
    {
      "epoch": 0.6755905511811023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2316,
      "step": 19734
    },
    {
      "epoch": 0.6756247860321808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2254,
      "step": 19735
    },
    {
      "epoch": 0.6756590208832591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2209,
      "step": 19736
    },
    {
      "epoch": 0.6756932557343376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2206,
      "step": 19737
    },
    {
      "epoch": 0.6757274905854159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2338,
      "step": 19738
    },
    {
      "epoch": 0.6757617254364944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2285,
      "step": 19739
    },
    {
      "epoch": 0.6757959602875727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2392,
      "step": 19740
    },
    {
      "epoch": 0.6758301951386512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2168,
      "step": 19741
    },
    {
      "epoch": 0.6758644299897295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2302,
      "step": 19742
    },
    {
      "epoch": 0.675898664840808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2263,
      "step": 19743
    },
    {
      "epoch": 0.6759328996918863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2338,
      "step": 19744
    },
    {
      "epoch": 0.6759671345429648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2209,
      "step": 19745
    },
    {
      "epoch": 0.6760013693940431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.232,
      "step": 19746
    },
    {
      "epoch": 0.6760356042451215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2298,
      "step": 19747
    },
    {
      "epoch": 0.6760698390961999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2317,
      "step": 19748
    },
    {
      "epoch": 0.6761040739472783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2334,
      "step": 19749
    },
    {
      "epoch": 0.6761383087983567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.208,
      "step": 19750
    },
    {
      "epoch": 0.6761725436494351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2398,
      "step": 19751
    },
    {
      "epoch": 0.6762067785005135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2136,
      "step": 19752
    },
    {
      "epoch": 0.6762410133515919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2492,
      "step": 19753
    },
    {
      "epoch": 0.6762752482026703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2359,
      "step": 19754
    },
    {
      "epoch": 0.6763094830537487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2304,
      "step": 19755
    },
    {
      "epoch": 0.6763437179048272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2182,
      "step": 19756
    },
    {
      "epoch": 0.6763779527559055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2504,
      "step": 19757
    },
    {
      "epoch": 0.676412187606984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2207,
      "step": 19758
    },
    {
      "epoch": 0.6764464224580623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.264,
      "step": 19759
    },
    {
      "epoch": 0.6764806573091408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2316,
      "step": 19760
    },
    {
      "epoch": 0.6765148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2325,
      "step": 19761
    },
    {
      "epoch": 0.6765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2203,
      "step": 19762
    },
    {
      "epoch": 0.6765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2334,
      "step": 19763
    },
    {
      "epoch": 0.6766175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2296,
      "step": 19764
    },
    {
      "epoch": 0.6766518315645327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2282,
      "step": 19765
    },
    {
      "epoch": 0.6766860664156111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2406,
      "step": 19766
    },
    {
      "epoch": 0.6767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2519,
      "step": 19767
    },
    {
      "epoch": 0.6767545361177679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2288,
      "step": 19768
    },
    {
      "epoch": 0.6767887709688463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2225,
      "step": 19769
    },
    {
      "epoch": 0.6768230058199247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2382,
      "step": 19770
    },
    {
      "epoch": 0.6768572406710031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2307,
      "step": 19771
    },
    {
      "epoch": 0.6768914755220815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2187,
      "step": 19772
    },
    {
      "epoch": 0.6769257103731599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2302,
      "step": 19773
    },
    {
      "epoch": 0.6769599452242383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2115,
      "step": 19774
    },
    {
      "epoch": 0.6769941800753166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2105,
      "step": 19775
    },
    {
      "epoch": 0.6770284149263951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2615,
      "step": 19776
    },
    {
      "epoch": 0.6770626497774734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2421,
      "step": 19777
    },
    {
      "epoch": 0.6770968846285519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2134,
      "step": 19778
    },
    {
      "epoch": 0.6771311194796302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2118,
      "step": 19779
    },
    {
      "epoch": 0.6771653543307087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2396,
      "step": 19780
    },
    {
      "epoch": 0.677199589181787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2334,
      "step": 19781
    },
    {
      "epoch": 0.6772338240328655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.214,
      "step": 19782
    },
    {
      "epoch": 0.6772680588839438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2194,
      "step": 19783
    },
    {
      "epoch": 0.6773022937350223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.217,
      "step": 19784
    },
    {
      "epoch": 0.6773365285861006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2359,
      "step": 19785
    },
    {
      "epoch": 0.6773707634371791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2196,
      "step": 19786
    },
    {
      "epoch": 0.6774049982882574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2352,
      "step": 19787
    },
    {
      "epoch": 0.6774392331393359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2311,
      "step": 19788
    },
    {
      "epoch": 0.6774734679904142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2208,
      "step": 19789
    },
    {
      "epoch": 0.6775077028414926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2601,
      "step": 19790
    },
    {
      "epoch": 0.677541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2283,
      "step": 19791
    },
    {
      "epoch": 0.6775761725436494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2416,
      "step": 19792
    },
    {
      "epoch": 0.6776104073947279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2262,
      "step": 19793
    },
    {
      "epoch": 0.6776446422458062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2204,
      "step": 19794
    },
    {
      "epoch": 0.6776788770968847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2205,
      "step": 19795
    },
    {
      "epoch": 0.677713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2292,
      "step": 19796
    },
    {
      "epoch": 0.6777473467990415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.22,
      "step": 19797
    },
    {
      "epoch": 0.6777815816501198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2116,
      "step": 19798
    },
    {
      "epoch": 0.6778158165011983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2358,
      "step": 19799
    },
    {
      "epoch": 0.6778500513522766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2272,
      "step": 19800
    },
    {
      "epoch": 0.6778842862033551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2259,
      "step": 19801
    },
    {
      "epoch": 0.6779185210544334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2218,
      "step": 19802
    },
    {
      "epoch": 0.6779527559055119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2345,
      "step": 19803
    },
    {
      "epoch": 0.6779869907565902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2342,
      "step": 19804
    },
    {
      "epoch": 0.6780212256076686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2489,
      "step": 19805
    },
    {
      "epoch": 0.678055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2318,
      "step": 19806
    },
    {
      "epoch": 0.6780896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2347,
      "step": 19807
    },
    {
      "epoch": 0.6781239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2314,
      "step": 19808
    },
    {
      "epoch": 0.6781581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2305,
      "step": 19809
    },
    {
      "epoch": 0.6781923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2326,
      "step": 19810
    },
    {
      "epoch": 0.678226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2265,
      "step": 19811
    },
    {
      "epoch": 0.6782608695652174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.222,
      "step": 19812
    },
    {
      "epoch": 0.6782951044162958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2221,
      "step": 19813
    },
    {
      "epoch": 0.6783293392673742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2175,
      "step": 19814
    },
    {
      "epoch": 0.6783635741184526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2322,
      "step": 19815
    },
    {
      "epoch": 0.678397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2303,
      "step": 19816
    },
    {
      "epoch": 0.6784320438206094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2353,
      "step": 19817
    },
    {
      "epoch": 0.6784662786716877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2448,
      "step": 19818
    },
    {
      "epoch": 0.6785005135227662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2364,
      "step": 19819
    },
    {
      "epoch": 0.6785347483738445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.225,
      "step": 19820
    },
    {
      "epoch": 0.678568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2335,
      "step": 19821
    },
    {
      "epoch": 0.6786032180760013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2405,
      "step": 19822
    },
    {
      "epoch": 0.6786374529270798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2207,
      "step": 19823
    },
    {
      "epoch": 0.6786716877781581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2357,
      "step": 19824
    },
    {
      "epoch": 0.6787059226292366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2258,
      "step": 19825
    },
    {
      "epoch": 0.6787401574803149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2318,
      "step": 19826
    },
    {
      "epoch": 0.6787743923313934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2257,
      "step": 19827
    },
    {
      "epoch": 0.6788086271824717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2392,
      "step": 19828
    },
    {
      "epoch": 0.6788428620335502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2107,
      "step": 19829
    },
    {
      "epoch": 0.6788770968846286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2342,
      "step": 19830
    },
    {
      "epoch": 0.678911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2369,
      "step": 19831
    },
    {
      "epoch": 0.6789455665867854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2075,
      "step": 19832
    },
    {
      "epoch": 0.6789798014378637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2286,
      "step": 19833
    },
    {
      "epoch": 0.6790140362889422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2448,
      "step": 19834
    },
    {
      "epoch": 0.6790482711400205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.23,
      "step": 19835
    },
    {
      "epoch": 0.679082505991099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2032,
      "step": 19836
    },
    {
      "epoch": 0.6791167408421773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2415,
      "step": 19837
    },
    {
      "epoch": 0.6791509756932558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2383,
      "step": 19838
    },
    {
      "epoch": 0.6791852105443341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2259,
      "step": 19839
    },
    {
      "epoch": 0.6792194453954126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2293,
      "step": 19840
    },
    {
      "epoch": 0.6792536802464909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2276,
      "step": 19841
    },
    {
      "epoch": 0.6792879150975694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2223,
      "step": 19842
    },
    {
      "epoch": 0.6793221499486477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2292,
      "step": 19843
    },
    {
      "epoch": 0.6793563847997262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2304,
      "step": 19844
    },
    {
      "epoch": 0.6793906196508045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2305,
      "step": 19845
    },
    {
      "epoch": 0.679424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2153,
      "step": 19846
    },
    {
      "epoch": 0.6794590893529613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2379,
      "step": 19847
    },
    {
      "epoch": 0.6794933242040397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2485,
      "step": 19848
    },
    {
      "epoch": 0.6795275590551181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2347,
      "step": 19849
    },
    {
      "epoch": 0.6795617939061965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2225,
      "step": 19850
    },
    {
      "epoch": 0.6795960287572749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2279,
      "step": 19851
    },
    {
      "epoch": 0.6796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2304,
      "step": 19852
    },
    {
      "epoch": 0.6796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2298,
      "step": 19853
    },
    {
      "epoch": 0.6796987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2155,
      "step": 19854
    },
    {
      "epoch": 0.6797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2207,
      "step": 19855
    },
    {
      "epoch": 0.6797672030126669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2111,
      "step": 19856
    },
    {
      "epoch": 0.6798014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2284,
      "step": 19857
    },
    {
      "epoch": 0.6798356727148237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2249,
      "step": 19858
    },
    {
      "epoch": 0.679869907565902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2236,
      "step": 19859
    },
    {
      "epoch": 0.6799041424169805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2387,
      "step": 19860
    },
    {
      "epoch": 0.6799383772680588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2417,
      "step": 19861
    },
    {
      "epoch": 0.6799726121191373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2296,
      "step": 19862
    },
    {
      "epoch": 0.6800068469702156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2345,
      "step": 19863
    },
    {
      "epoch": 0.6800410818212941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2079,
      "step": 19864
    },
    {
      "epoch": 0.6800753166723724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2133,
      "step": 19865
    },
    {
      "epoch": 0.6801095515234509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2154,
      "step": 19866
    },
    {
      "epoch": 0.6801437863745293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.229,
      "step": 19867
    },
    {
      "epoch": 0.6801780212256077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.23,
      "step": 19868
    },
    {
      "epoch": 0.6802122560766861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2177,
      "step": 19869
    },
    {
      "epoch": 0.6802464909277645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2267,
      "step": 19870
    },
    {
      "epoch": 0.6802807257788429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2148,
      "step": 19871
    },
    {
      "epoch": 0.6803149606299213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.239,
      "step": 19872
    },
    {
      "epoch": 0.6803491954809997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2287,
      "step": 19873
    },
    {
      "epoch": 0.680383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2319,
      "step": 19874
    },
    {
      "epoch": 0.6804176651831565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2338,
      "step": 19875
    },
    {
      "epoch": 0.6804519000342348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2185,
      "step": 19876
    },
    {
      "epoch": 0.6804861348853133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2275,
      "step": 19877
    },
    {
      "epoch": 0.6805203697363916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2251,
      "step": 19878
    },
    {
      "epoch": 0.6805546045874701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.232,
      "step": 19879
    },
    {
      "epoch": 0.6805888394385484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2115,
      "step": 19880
    },
    {
      "epoch": 0.6806230742896269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2327,
      "step": 19881
    },
    {
      "epoch": 0.6806573091407052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2273,
      "step": 19882
    },
    {
      "epoch": 0.6806915439917837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2143,
      "step": 19883
    },
    {
      "epoch": 0.680725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.213,
      "step": 19884
    },
    {
      "epoch": 0.6807600136939405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2148,
      "step": 19885
    },
    {
      "epoch": 0.6807942485450188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2227,
      "step": 19886
    },
    {
      "epoch": 0.6808284833960973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2264,
      "step": 19887
    },
    {
      "epoch": 0.6808627182471756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2328,
      "step": 19888
    },
    {
      "epoch": 0.680896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2201,
      "step": 19889
    },
    {
      "epoch": 0.6809311879493324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.235,
      "step": 19890
    },
    {
      "epoch": 0.6809654228004108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.241,
      "step": 19891
    },
    {
      "epoch": 0.6809996576514892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2241,
      "step": 19892
    },
    {
      "epoch": 0.6810338925025676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2262,
      "step": 19893
    },
    {
      "epoch": 0.681068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2398,
      "step": 19894
    },
    {
      "epoch": 0.6811023622047244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2248,
      "step": 19895
    },
    {
      "epoch": 0.6811365970558028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.22,
      "step": 19896
    },
    {
      "epoch": 0.6811708319068812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2297,
      "step": 19897
    },
    {
      "epoch": 0.6812050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2353,
      "step": 19898
    },
    {
      "epoch": 0.681239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2311,
      "step": 19899
    },
    {
      "epoch": 0.6812735364601163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2351,
      "step": 19900
    },
    {
      "epoch": 0.6813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2407,
      "step": 19901
    },
    {
      "epoch": 0.6813420061622731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2193,
      "step": 19902
    },
    {
      "epoch": 0.6813762410133516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2101,
      "step": 19903
    },
    {
      "epoch": 0.68141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2614,
      "step": 19904
    },
    {
      "epoch": 0.6814447107155084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2364,
      "step": 19905
    },
    {
      "epoch": 0.6814789455665868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2167,
      "step": 19906
    },
    {
      "epoch": 0.6815131804176652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2276,
      "step": 19907
    },
    {
      "epoch": 0.6815474152687436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2228,
      "step": 19908
    },
    {
      "epoch": 0.681581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.229,
      "step": 19909
    },
    {
      "epoch": 0.6816158849709004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2044,
      "step": 19910
    },
    {
      "epoch": 0.6816501198219788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2232,
      "step": 19911
    },
    {
      "epoch": 0.6816843546730572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2442,
      "step": 19912
    },
    {
      "epoch": 0.6817185895241356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2152,
      "step": 19913
    },
    {
      "epoch": 0.681752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2251,
      "step": 19914
    },
    {
      "epoch": 0.6817870592262923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2236,
      "step": 19915
    },
    {
      "epoch": 0.6818212940773708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2315,
      "step": 19916
    },
    {
      "epoch": 0.6818555289284491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2239,
      "step": 19917
    },
    {
      "epoch": 0.6818897637795276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2322,
      "step": 19918
    },
    {
      "epoch": 0.6819239986306059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2268,
      "step": 19919
    },
    {
      "epoch": 0.6819582334816844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2234,
      "step": 19920
    },
    {
      "epoch": 0.6819924683327627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2302,
      "step": 19921
    },
    {
      "epoch": 0.6820267031838412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2237,
      "step": 19922
    },
    {
      "epoch": 0.6820609380349195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2305,
      "step": 19923
    },
    {
      "epoch": 0.682095172885998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2248,
      "step": 19924
    },
    {
      "epoch": 0.6821294077370763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2118,
      "step": 19925
    },
    {
      "epoch": 0.6821636425881548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2192,
      "step": 19926
    },
    {
      "epoch": 0.6821978774392331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2192,
      "step": 19927
    },
    {
      "epoch": 0.6822321122903116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2177,
      "step": 19928
    },
    {
      "epoch": 0.6822663471413899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2292,
      "step": 19929
    },
    {
      "epoch": 0.6823005819924683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2145,
      "step": 19930
    },
    {
      "epoch": 0.6823348168435467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2273,
      "step": 19931
    },
    {
      "epoch": 0.6823690516946251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2317,
      "step": 19932
    },
    {
      "epoch": 0.6824032865457035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2199,
      "step": 19933
    },
    {
      "epoch": 0.6824375213967819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2222,
      "step": 19934
    },
    {
      "epoch": 0.6824717562478603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.227,
      "step": 19935
    },
    {
      "epoch": 0.6825059910989387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2394,
      "step": 19936
    },
    {
      "epoch": 0.6825402259500171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2196,
      "step": 19937
    },
    {
      "epoch": 0.6825744608010955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2376,
      "step": 19938
    },
    {
      "epoch": 0.6826086956521739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2157,
      "step": 19939
    },
    {
      "epoch": 0.6826429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.221,
      "step": 19940
    },
    {
      "epoch": 0.6826771653543308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2538,
      "step": 19941
    },
    {
      "epoch": 0.6827114002054091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2302,
      "step": 19942
    },
    {
      "epoch": 0.6827456350564876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2267,
      "step": 19943
    },
    {
      "epoch": 0.6827798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2233,
      "step": 19944
    },
    {
      "epoch": 0.6828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2174,
      "step": 19945
    },
    {
      "epoch": 0.6828483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2322,
      "step": 19946
    },
    {
      "epoch": 0.6828825744608011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2271,
      "step": 19947
    },
    {
      "epoch": 0.6829168093118795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2561,
      "step": 19948
    },
    {
      "epoch": 0.6829510441629579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2214,
      "step": 19949
    },
    {
      "epoch": 0.6829852790140363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2276,
      "step": 19950
    },
    {
      "epoch": 0.6830195138651147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2156,
      "step": 19951
    },
    {
      "epoch": 0.6830537487161931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2134,
      "step": 19952
    },
    {
      "epoch": 0.6830879835672715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2212,
      "step": 19953
    },
    {
      "epoch": 0.6831222184183499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.226,
      "step": 19954
    },
    {
      "epoch": 0.6831564532694283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2268,
      "step": 19955
    },
    {
      "epoch": 0.6831906881205067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2324,
      "step": 19956
    },
    {
      "epoch": 0.6832249229715851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2246,
      "step": 19957
    },
    {
      "epoch": 0.6832591578226634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2241,
      "step": 19958
    },
    {
      "epoch": 0.6832933926737419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2437,
      "step": 19959
    },
    {
      "epoch": 0.6833276275248202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2331,
      "step": 19960
    },
    {
      "epoch": 0.6833618623758987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2225,
      "step": 19961
    },
    {
      "epoch": 0.683396097226977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2439,
      "step": 19962
    },
    {
      "epoch": 0.6834303320780555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2376,
      "step": 19963
    },
    {
      "epoch": 0.6834645669291338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2247,
      "step": 19964
    },
    {
      "epoch": 0.6834988017802123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2087,
      "step": 19965
    },
    {
      "epoch": 0.6835330366312906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2241,
      "step": 19966
    },
    {
      "epoch": 0.6835672714823691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2327,
      "step": 19967
    },
    {
      "epoch": 0.6836015063334474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2604,
      "step": 19968
    },
    {
      "epoch": 0.6836357411845259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2255,
      "step": 19969
    },
    {
      "epoch": 0.6836699760356042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2261,
      "step": 19970
    },
    {
      "epoch": 0.6837042108866827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2155,
      "step": 19971
    },
    {
      "epoch": 0.683738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2433,
      "step": 19972
    },
    {
      "epoch": 0.6837726805888394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2184,
      "step": 19973
    },
    {
      "epoch": 0.6838069154399178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2185,
      "step": 19974
    },
    {
      "epoch": 0.6838411502909962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.248,
      "step": 19975
    },
    {
      "epoch": 0.6838753851420746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2501,
      "step": 19976
    },
    {
      "epoch": 0.683909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2173,
      "step": 19977
    },
    {
      "epoch": 0.6839438548442315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2313,
      "step": 19978
    },
    {
      "epoch": 0.6839780896953098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2324,
      "step": 19979
    },
    {
      "epoch": 0.6840123245463883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2344,
      "step": 19980
    },
    {
      "epoch": 0.6840465593974666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2303,
      "step": 19981
    },
    {
      "epoch": 0.6840807942485451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2264,
      "step": 19982
    },
    {
      "epoch": 0.6841150290996234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2301,
      "step": 19983
    },
    {
      "epoch": 0.6841492639507019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2308,
      "step": 19984
    },
    {
      "epoch": 0.6841834988017802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2295,
      "step": 19985
    },
    {
      "epoch": 0.6842177336528587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2302,
      "step": 19986
    },
    {
      "epoch": 0.684251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2276,
      "step": 19987
    },
    {
      "epoch": 0.6842862033550154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2277,
      "step": 19988
    },
    {
      "epoch": 0.6843204382060938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2267,
      "step": 19989
    },
    {
      "epoch": 0.6843546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2191,
      "step": 19990
    },
    {
      "epoch": 0.6843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2164,
      "step": 19991
    },
    {
      "epoch": 0.684423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2454,
      "step": 19992
    },
    {
      "epoch": 0.6844573776104074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2267,
      "step": 19993
    },
    {
      "epoch": 0.6844916124614858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2224,
      "step": 19994
    },
    {
      "epoch": 0.6845258473125642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2243,
      "step": 19995
    },
    {
      "epoch": 0.6845600821636426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2332,
      "step": 19996
    },
    {
      "epoch": 0.684594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2154,
      "step": 19997
    },
    {
      "epoch": 0.6846285518657994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2377,
      "step": 19998
    },
    {
      "epoch": 0.6846627867168777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2424,
      "step": 19999
    },
    {
      "epoch": 0.6846970215679562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2244,
      "step": 20000
    },
    {
      "epoch": 0.6847312564190345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2184,
      "step": 20001
    },
    {
      "epoch": 0.684765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2202,
      "step": 20002
    },
    {
      "epoch": 0.6847997261211913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2158,
      "step": 20003
    },
    {
      "epoch": 0.6848339609722698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2245,
      "step": 20004
    },
    {
      "epoch": 0.6848681958233481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2203,
      "step": 20005
    },
    {
      "epoch": 0.6849024306744266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2232,
      "step": 20006
    },
    {
      "epoch": 0.6849366655255049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2251,
      "step": 20007
    },
    {
      "epoch": 0.6849709003765834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2276,
      "step": 20008
    },
    {
      "epoch": 0.6850051352276617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2012,
      "step": 20009
    },
    {
      "epoch": 0.6850393700787402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2251,
      "step": 20010
    },
    {
      "epoch": 0.6850736049298185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2163,
      "step": 20011
    },
    {
      "epoch": 0.685107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2119,
      "step": 20012
    },
    {
      "epoch": 0.6851420746319754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2265,
      "step": 20013
    },
    {
      "epoch": 0.6851763094830537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2179,
      "step": 20014
    },
    {
      "epoch": 0.6852105443341322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2216,
      "step": 20015
    },
    {
      "epoch": 0.6852447791852105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2264,
      "step": 20016
    },
    {
      "epoch": 0.685279014036289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2239,
      "step": 20017
    },
    {
      "epoch": 0.6853132488873673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2379,
      "step": 20018
    },
    {
      "epoch": 0.6853474837384458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2136,
      "step": 20019
    },
    {
      "epoch": 0.6853817185895241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2209,
      "step": 20020
    },
    {
      "epoch": 0.6854159534406026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2104,
      "step": 20021
    },
    {
      "epoch": 0.6854501882916809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2113,
      "step": 20022
    },
    {
      "epoch": 0.6854844231427594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2436,
      "step": 20023
    },
    {
      "epoch": 0.6855186579938377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2267,
      "step": 20024
    },
    {
      "epoch": 0.6855528928449162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.23,
      "step": 20025
    },
    {
      "epoch": 0.6855871276959945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2273,
      "step": 20026
    },
    {
      "epoch": 0.685621362547073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2305,
      "step": 20027
    },
    {
      "epoch": 0.6856555973981513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2392,
      "step": 20028
    },
    {
      "epoch": 0.6856898322492297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2171,
      "step": 20029
    },
    {
      "epoch": 0.6857240671003081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2398,
      "step": 20030
    },
    {
      "epoch": 0.6857583019513865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2142,
      "step": 20031
    },
    {
      "epoch": 0.6857925368024649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.231,
      "step": 20032
    },
    {
      "epoch": 0.6858267716535433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2198,
      "step": 20033
    },
    {
      "epoch": 0.6858610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2321,
      "step": 20034
    },
    {
      "epoch": 0.6858952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2343,
      "step": 20035
    },
    {
      "epoch": 0.6859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.21,
      "step": 20036
    },
    {
      "epoch": 0.6859637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2258,
      "step": 20037
    },
    {
      "epoch": 0.6859979459089353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2161,
      "step": 20038
    },
    {
      "epoch": 0.6860321807600137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2307,
      "step": 20039
    },
    {
      "epoch": 0.686066415611092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2131,
      "step": 20040
    },
    {
      "epoch": 0.6861006504621705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2163,
      "step": 20041
    },
    {
      "epoch": 0.6861348853132488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2035,
      "step": 20042
    },
    {
      "epoch": 0.6861691201643273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2256,
      "step": 20043
    },
    {
      "epoch": 0.6862033550154056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.243,
      "step": 20044
    },
    {
      "epoch": 0.6862375898664841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2129,
      "step": 20045
    },
    {
      "epoch": 0.6862718247175624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2206,
      "step": 20046
    },
    {
      "epoch": 0.6863060595686409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2245,
      "step": 20047
    },
    {
      "epoch": 0.6863402944197192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2271,
      "step": 20048
    },
    {
      "epoch": 0.6863745292707977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2179,
      "step": 20049
    },
    {
      "epoch": 0.6864087641218761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2024,
      "step": 20050
    },
    {
      "epoch": 0.6864429989729545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2245,
      "step": 20051
    },
    {
      "epoch": 0.6864772338240329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.219,
      "step": 20052
    },
    {
      "epoch": 0.6865114686751113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2191,
      "step": 20053
    },
    {
      "epoch": 0.6865457035261897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2341,
      "step": 20054
    },
    {
      "epoch": 0.686579938377268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2157,
      "step": 20055
    },
    {
      "epoch": 0.6866141732283465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2332,
      "step": 20056
    },
    {
      "epoch": 0.6866484080794248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2195,
      "step": 20057
    },
    {
      "epoch": 0.6866826429305033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.204,
      "step": 20058
    },
    {
      "epoch": 0.6867168777815816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2406,
      "step": 20059
    },
    {
      "epoch": 0.6867511126326601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2184,
      "step": 20060
    },
    {
      "epoch": 0.6867853474837384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2271,
      "step": 20061
    },
    {
      "epoch": 0.6868195823348169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1988,
      "step": 20062
    },
    {
      "epoch": 0.6868538171858952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2144,
      "step": 20063
    },
    {
      "epoch": 0.6868880520369737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.224,
      "step": 20064
    },
    {
      "epoch": 0.686922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1949,
      "step": 20065
    },
    {
      "epoch": 0.6869565217391305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2046,
      "step": 20066
    },
    {
      "epoch": 0.6869907565902088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2372,
      "step": 20067
    },
    {
      "epoch": 0.6870249914412873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2241,
      "step": 20068
    },
    {
      "epoch": 0.6870592262923656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2154,
      "step": 20069
    },
    {
      "epoch": 0.687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2153,
      "step": 20070
    },
    {
      "epoch": 0.6871276959945224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2152,
      "step": 20071
    },
    {
      "epoch": 0.6871619308456008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2144,
      "step": 20072
    },
    {
      "epoch": 0.6871961656966792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2284,
      "step": 20073
    },
    {
      "epoch": 0.6872304005477576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2236,
      "step": 20074
    },
    {
      "epoch": 0.687264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2159,
      "step": 20075
    },
    {
      "epoch": 0.6872988702499144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2212,
      "step": 20076
    },
    {
      "epoch": 0.6873331051009928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2108,
      "step": 20077
    },
    {
      "epoch": 0.6873673399520712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2263,
      "step": 20078
    },
    {
      "epoch": 0.6874015748031496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2279,
      "step": 20079
    },
    {
      "epoch": 0.687435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2332,
      "step": 20080
    },
    {
      "epoch": 0.6874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2258,
      "step": 20081
    },
    {
      "epoch": 0.6875042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.213,
      "step": 20082
    },
    {
      "epoch": 0.6875385142074631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2205,
      "step": 20083
    },
    {
      "epoch": 0.6875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2128,
      "step": 20084
    },
    {
      "epoch": 0.6876069839096199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2429,
      "step": 20085
    },
    {
      "epoch": 0.6876412187606984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2324,
      "step": 20086
    },
    {
      "epoch": 0.6876754536117768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2166,
      "step": 20087
    },
    {
      "epoch": 0.6877096884628552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2138,
      "step": 20088
    },
    {
      "epoch": 0.6877439233139336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2049,
      "step": 20089
    },
    {
      "epoch": 0.687778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2158,
      "step": 20090
    },
    {
      "epoch": 0.6878123930160904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2484,
      "step": 20091
    },
    {
      "epoch": 0.6878466278671688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2286,
      "step": 20092
    },
    {
      "epoch": 0.6878808627182472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2243,
      "step": 20093
    },
    {
      "epoch": 0.6879150975693256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2124,
      "step": 20094
    },
    {
      "epoch": 0.687949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2078,
      "step": 20095
    },
    {
      "epoch": 0.6879835672714824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2336,
      "step": 20096
    },
    {
      "epoch": 0.6880178021225608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2259,
      "step": 20097
    },
    {
      "epoch": 0.6880520369736391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2336,
      "step": 20098
    },
    {
      "epoch": 0.6880862718247176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2377,
      "step": 20099
    },
    {
      "epoch": 0.6881205066757959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2189,
      "step": 20100
    },
    {
      "epoch": 0.6881547415268744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2271,
      "step": 20101
    },
    {
      "epoch": 0.6881889763779527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2004,
      "step": 20102
    },
    {
      "epoch": 0.6882232112290312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.218,
      "step": 20103
    },
    {
      "epoch": 0.6882574460801095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2216,
      "step": 20104
    },
    {
      "epoch": 0.688291680931188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2144,
      "step": 20105
    },
    {
      "epoch": 0.6883259157822663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2194,
      "step": 20106
    },
    {
      "epoch": 0.6883601506333448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2083,
      "step": 20107
    },
    {
      "epoch": 0.6883943854844231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2123,
      "step": 20108
    },
    {
      "epoch": 0.6884286203355016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2087,
      "step": 20109
    },
    {
      "epoch": 0.6884628551865799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2192,
      "step": 20110
    },
    {
      "epoch": 0.6884970900376584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2164,
      "step": 20111
    },
    {
      "epoch": 0.6885313248887367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2329,
      "step": 20112
    },
    {
      "epoch": 0.6885655597398151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2248,
      "step": 20113
    },
    {
      "epoch": 0.6885997945908935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2406,
      "step": 20114
    },
    {
      "epoch": 0.6886340294419719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2199,
      "step": 20115
    },
    {
      "epoch": 0.6886682642930503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2008,
      "step": 20116
    },
    {
      "epoch": 0.6887024991441287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2296,
      "step": 20117
    },
    {
      "epoch": 0.6887367339952071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2227,
      "step": 20118
    },
    {
      "epoch": 0.6887709688462855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.237,
      "step": 20119
    },
    {
      "epoch": 0.6888052036973639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2272,
      "step": 20120
    },
    {
      "epoch": 0.6888394385484423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2061,
      "step": 20121
    },
    {
      "epoch": 0.6888736733995207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2178,
      "step": 20122
    },
    {
      "epoch": 0.6889079082505991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2138,
      "step": 20123
    },
    {
      "epoch": 0.6889421431016776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2315,
      "step": 20124
    },
    {
      "epoch": 0.6889763779527559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2265,
      "step": 20125
    },
    {
      "epoch": 0.6890106128038344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2254,
      "step": 20126
    },
    {
      "epoch": 0.6890448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2126,
      "step": 20127
    },
    {
      "epoch": 0.6890790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2139,
      "step": 20128
    },
    {
      "epoch": 0.6891133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2181,
      "step": 20129
    },
    {
      "epoch": 0.6891475522081479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2514,
      "step": 20130
    },
    {
      "epoch": 0.6891817870592263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2166,
      "step": 20131
    },
    {
      "epoch": 0.6892160219103047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.233,
      "step": 20132
    },
    {
      "epoch": 0.6892502567613831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.208,
      "step": 20133
    },
    {
      "epoch": 0.6892844916124615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2211,
      "step": 20134
    },
    {
      "epoch": 0.6893187264635399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2216,
      "step": 20135
    },
    {
      "epoch": 0.6893529613146183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2083,
      "step": 20136
    },
    {
      "epoch": 0.6893871961656967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2281,
      "step": 20137
    },
    {
      "epoch": 0.6894214310167751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2153,
      "step": 20138
    },
    {
      "epoch": 0.6894556658678535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.235,
      "step": 20139
    },
    {
      "epoch": 0.6894899007189319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2269,
      "step": 20140
    },
    {
      "epoch": 0.6895241355700102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2161,
      "step": 20141
    },
    {
      "epoch": 0.6895583704210887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.217,
      "step": 20142
    },
    {
      "epoch": 0.689592605272167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2288,
      "step": 20143
    },
    {
      "epoch": 0.6896268401232455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.204,
      "step": 20144
    },
    {
      "epoch": 0.6896610749743238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2175,
      "step": 20145
    },
    {
      "epoch": 0.6896953098254023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2377,
      "step": 20146
    },
    {
      "epoch": 0.6897295446764806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2169,
      "step": 20147
    },
    {
      "epoch": 0.6897637795275591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2133,
      "step": 20148
    },
    {
      "epoch": 0.6897980143786374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2293,
      "step": 20149
    },
    {
      "epoch": 0.6898322492297159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2274,
      "step": 20150
    },
    {
      "epoch": 0.6898664840807942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2182,
      "step": 20151
    },
    {
      "epoch": 0.6899007189318727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1957,
      "step": 20152
    },
    {
      "epoch": 0.689934953782951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2336,
      "step": 20153
    },
    {
      "epoch": 0.6899691886340295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2115,
      "step": 20154
    },
    {
      "epoch": 0.6900034234851078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2092,
      "step": 20155
    },
    {
      "epoch": 0.6900376583361862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2235,
      "step": 20156
    },
    {
      "epoch": 0.6900718931872646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2246,
      "step": 20157
    },
    {
      "epoch": 0.690106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2242,
      "step": 20158
    },
    {
      "epoch": 0.6901403628894214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2534,
      "step": 20159
    },
    {
      "epoch": 0.6901745977404998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2447,
      "step": 20160
    },
    {
      "epoch": 0.6902088325915783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2342,
      "step": 20161
    },
    {
      "epoch": 0.6902430674426566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2039,
      "step": 20162
    },
    {
      "epoch": 0.6902773022937351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2259,
      "step": 20163
    },
    {
      "epoch": 0.6903115371448134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2236,
      "step": 20164
    },
    {
      "epoch": 0.6903457719958919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2176,
      "step": 20165
    },
    {
      "epoch": 0.6903800068469702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2209,
      "step": 20166
    },
    {
      "epoch": 0.6904142416980487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2138,
      "step": 20167
    },
    {
      "epoch": 0.690448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2174,
      "step": 20168
    },
    {
      "epoch": 0.6904827114002055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2214,
      "step": 20169
    },
    {
      "epoch": 0.6905169462512838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2263,
      "step": 20170
    },
    {
      "epoch": 0.6905511811023622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2269,
      "step": 20171
    },
    {
      "epoch": 0.6905854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2081,
      "step": 20172
    },
    {
      "epoch": 0.690619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2287,
      "step": 20173
    },
    {
      "epoch": 0.6906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2219,
      "step": 20174
    },
    {
      "epoch": 0.6906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.235,
      "step": 20175
    },
    {
      "epoch": 0.6907223553577542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2294,
      "step": 20176
    },
    {
      "epoch": 0.6907565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.24,
      "step": 20177
    },
    {
      "epoch": 0.690790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2388,
      "step": 20178
    },
    {
      "epoch": 0.6908250599109894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.234,
      "step": 20179
    },
    {
      "epoch": 0.6908592947620678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.226,
      "step": 20180
    },
    {
      "epoch": 0.6908935296131462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2258,
      "step": 20181
    },
    {
      "epoch": 0.6909277644642245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2328,
      "step": 20182
    },
    {
      "epoch": 0.690961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2365,
      "step": 20183
    },
    {
      "epoch": 0.6909962341663813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2247,
      "step": 20184
    },
    {
      "epoch": 0.6910304690174598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1998,
      "step": 20185
    },
    {
      "epoch": 0.6910647038685381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2077,
      "step": 20186
    },
    {
      "epoch": 0.6910989387196166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2097,
      "step": 20187
    },
    {
      "epoch": 0.6911331735706949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2257,
      "step": 20188
    },
    {
      "epoch": 0.6911674084217734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2219,
      "step": 20189
    },
    {
      "epoch": 0.6912016432728517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2206,
      "step": 20190
    },
    {
      "epoch": 0.6912358781239302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2284,
      "step": 20191
    },
    {
      "epoch": 0.6912701129750085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2333,
      "step": 20192
    },
    {
      "epoch": 0.691304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2055,
      "step": 20193
    },
    {
      "epoch": 0.6913385826771653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2266,
      "step": 20194
    },
    {
      "epoch": 0.6913728175282438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2253,
      "step": 20195
    },
    {
      "epoch": 0.6914070523793221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2212,
      "step": 20196
    },
    {
      "epoch": 0.6914412872304005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2356,
      "step": 20197
    },
    {
      "epoch": 0.691475522081479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2159,
      "step": 20198
    },
    {
      "epoch": 0.6915097569325573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2273,
      "step": 20199
    },
    {
      "epoch": 0.6915439917836358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2425,
      "step": 20200
    },
    {
      "epoch": 0.6915782266347141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2252,
      "step": 20201
    },
    {
      "epoch": 0.6916124614857926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2192,
      "step": 20202
    },
    {
      "epoch": 0.6916466963368709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2394,
      "step": 20203
    },
    {
      "epoch": 0.6916809311879494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.232,
      "step": 20204
    },
    {
      "epoch": 0.6917151660390277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2094,
      "step": 20205
    },
    {
      "epoch": 0.6917494008901062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2186,
      "step": 20206
    },
    {
      "epoch": 0.6917836357411845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2002,
      "step": 20207
    },
    {
      "epoch": 0.691817870592263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2237,
      "step": 20208
    },
    {
      "epoch": 0.6918521054433413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2393,
      "step": 20209
    },
    {
      "epoch": 0.6918863402944198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2258,
      "step": 20210
    },
    {
      "epoch": 0.6919205751454981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2318,
      "step": 20211
    },
    {
      "epoch": 0.6919548099965765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2235,
      "step": 20212
    },
    {
      "epoch": 0.6919890448476549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2455,
      "step": 20213
    },
    {
      "epoch": 0.6920232796987333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2311,
      "step": 20214
    },
    {
      "epoch": 0.6920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2221,
      "step": 20215
    },
    {
      "epoch": 0.6920917494008901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2131,
      "step": 20216
    },
    {
      "epoch": 0.6921259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2218,
      "step": 20217
    },
    {
      "epoch": 0.6921602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2194,
      "step": 20218
    },
    {
      "epoch": 0.6921944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.228,
      "step": 20219
    },
    {
      "epoch": 0.6922286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2297,
      "step": 20220
    },
    {
      "epoch": 0.6922629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2011,
      "step": 20221
    },
    {
      "epoch": 0.6922971585073605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2311,
      "step": 20222
    },
    {
      "epoch": 0.6923313933584389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2276,
      "step": 20223
    },
    {
      "epoch": 0.6923656282095173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2214,
      "step": 20224
    },
    {
      "epoch": 0.6923998630605956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2275,
      "step": 20225
    },
    {
      "epoch": 0.6924340979116741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2294,
      "step": 20226
    },
    {
      "epoch": 0.6924683327627524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2309,
      "step": 20227
    },
    {
      "epoch": 0.6925025676138309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2135,
      "step": 20228
    },
    {
      "epoch": 0.6925368024649092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2178,
      "step": 20229
    },
    {
      "epoch": 0.6925710373159877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2091,
      "step": 20230
    },
    {
      "epoch": 0.692605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2275,
      "step": 20231
    },
    {
      "epoch": 0.6926395070181445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2015,
      "step": 20232
    },
    {
      "epoch": 0.6926737418692228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2191,
      "step": 20233
    },
    {
      "epoch": 0.6927079767203013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2162,
      "step": 20234
    },
    {
      "epoch": 0.6927422115713797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2232,
      "step": 20235
    },
    {
      "epoch": 0.6927764464224581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.217,
      "step": 20236
    },
    {
      "epoch": 0.6928106812735365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2151,
      "step": 20237
    },
    {
      "epoch": 0.6928449161246149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2111,
      "step": 20238
    },
    {
      "epoch": 0.6928791509756933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2177,
      "step": 20239
    },
    {
      "epoch": 0.6929133858267716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2126,
      "step": 20240
    },
    {
      "epoch": 0.6929476206778501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2096,
      "step": 20241
    },
    {
      "epoch": 0.6929818555289284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2113,
      "step": 20242
    },
    {
      "epoch": 0.6930160903800069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2126,
      "step": 20243
    },
    {
      "epoch": 0.6930503252310852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2251,
      "step": 20244
    },
    {
      "epoch": 0.6930845600821637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2082,
      "step": 20245
    },
    {
      "epoch": 0.693118794933242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2375,
      "step": 20246
    },
    {
      "epoch": 0.6931530297843205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2181,
      "step": 20247
    },
    {
      "epoch": 0.6931872646353988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2283,
      "step": 20248
    },
    {
      "epoch": 0.6932214994864773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2211,
      "step": 20249
    },
    {
      "epoch": 0.6932557343375556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2284,
      "step": 20250
    },
    {
      "epoch": 0.6932899691886341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2223,
      "step": 20251
    },
    {
      "epoch": 0.6933242040397124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2321,
      "step": 20252
    },
    {
      "epoch": 0.6933584388907909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2331,
      "step": 20253
    },
    {
      "epoch": 0.6933926737418692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.204,
      "step": 20254
    },
    {
      "epoch": 0.6934269085929476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1977,
      "step": 20255
    },
    {
      "epoch": 0.693461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2149,
      "step": 20256
    },
    {
      "epoch": 0.6934953782951044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2143,
      "step": 20257
    },
    {
      "epoch": 0.6935296131461828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2166,
      "step": 20258
    },
    {
      "epoch": 0.6935638479972612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2212,
      "step": 20259
    },
    {
      "epoch": 0.6935980828483396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2228,
      "step": 20260
    },
    {
      "epoch": 0.693632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2017,
      "step": 20261
    },
    {
      "epoch": 0.6936665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2133,
      "step": 20262
    },
    {
      "epoch": 0.6937007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.218,
      "step": 20263
    },
    {
      "epoch": 0.6937350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2251,
      "step": 20264
    },
    {
      "epoch": 0.6937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2237,
      "step": 20265
    },
    {
      "epoch": 0.69380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2271,
      "step": 20266
    },
    {
      "epoch": 0.6938377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2169,
      "step": 20267
    },
    {
      "epoch": 0.6938719616569667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2137,
      "step": 20268
    },
    {
      "epoch": 0.6939061965080452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2149,
      "step": 20269
    },
    {
      "epoch": 0.6939404313591235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2243,
      "step": 20270
    },
    {
      "epoch": 0.693974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2147,
      "step": 20271
    },
    {
      "epoch": 0.6940089010612804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2155,
      "step": 20272
    },
    {
      "epoch": 0.6940431359123588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2014,
      "step": 20273
    },
    {
      "epoch": 0.6940773707634372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2267,
      "step": 20274
    },
    {
      "epoch": 0.6941116056145156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2159,
      "step": 20275
    },
    {
      "epoch": 0.694145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2089,
      "step": 20276
    },
    {
      "epoch": 0.6941800753166724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2112,
      "step": 20277
    },
    {
      "epoch": 0.6942143101677508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2239,
      "step": 20278
    },
    {
      "epoch": 0.6942485450188292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2234,
      "step": 20279
    },
    {
      "epoch": 0.6942827798699076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.224,
      "step": 20280
    },
    {
      "epoch": 0.694317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2111,
      "step": 20281
    },
    {
      "epoch": 0.6943512495720644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2157,
      "step": 20282
    },
    {
      "epoch": 0.6943854844231427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2235,
      "step": 20283
    },
    {
      "epoch": 0.6944197192742212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2239,
      "step": 20284
    },
    {
      "epoch": 0.6944539541252995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1923,
      "step": 20285
    },
    {
      "epoch": 0.694488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2388,
      "step": 20286
    },
    {
      "epoch": 0.6945224238274563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2281,
      "step": 20287
    },
    {
      "epoch": 0.6945566586785348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.218,
      "step": 20288
    },
    {
      "epoch": 0.6945908935296131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2173,
      "step": 20289
    },
    {
      "epoch": 0.6946251283806916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2144,
      "step": 20290
    },
    {
      "epoch": 0.6946593632317699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2302,
      "step": 20291
    },
    {
      "epoch": 0.6946935980828484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2263,
      "step": 20292
    },
    {
      "epoch": 0.6947278329339267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2277,
      "step": 20293
    },
    {
      "epoch": 0.6947620677850052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2291,
      "step": 20294
    },
    {
      "epoch": 0.6947963026360835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2107,
      "step": 20295
    },
    {
      "epoch": 0.694830537487162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2053,
      "step": 20296
    },
    {
      "epoch": 0.6948647723382403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2099,
      "step": 20297
    },
    {
      "epoch": 0.6948990071893187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.201,
      "step": 20298
    },
    {
      "epoch": 0.6949332420403971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1918,
      "step": 20299
    },
    {
      "epoch": 0.6949674768914755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2227,
      "step": 20300
    },
    {
      "epoch": 0.6950017117425539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2227,
      "step": 20301
    },
    {
      "epoch": 0.6950359465936323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2242,
      "step": 20302
    },
    {
      "epoch": 0.6950701814447107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2352,
      "step": 20303
    },
    {
      "epoch": 0.6951044162957891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2227,
      "step": 20304
    },
    {
      "epoch": 0.6951386511468675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2159,
      "step": 20305
    },
    {
      "epoch": 0.6951728859979459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2244,
      "step": 20306
    },
    {
      "epoch": 0.6952071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2397,
      "step": 20307
    },
    {
      "epoch": 0.6952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2189,
      "step": 20308
    },
    {
      "epoch": 0.6952755905511812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2266,
      "step": 20309
    },
    {
      "epoch": 0.6953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2053,
      "step": 20310
    },
    {
      "epoch": 0.695344060253338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2128,
      "step": 20311
    },
    {
      "epoch": 0.6953782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2277,
      "step": 20312
    },
    {
      "epoch": 0.6954125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2141,
      "step": 20313
    },
    {
      "epoch": 0.6954467648065731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2191,
      "step": 20314
    },
    {
      "epoch": 0.6954809996576515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2194,
      "step": 20315
    },
    {
      "epoch": 0.6955152345087299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.227,
      "step": 20316
    },
    {
      "epoch": 0.6955494693598083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2104,
      "step": 20317
    },
    {
      "epoch": 0.6955837042108867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2104,
      "step": 20318
    },
    {
      "epoch": 0.6956179390619651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2326,
      "step": 20319
    },
    {
      "epoch": 0.6956521739130435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.219,
      "step": 20320
    },
    {
      "epoch": 0.6956864087641219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2084,
      "step": 20321
    },
    {
      "epoch": 0.6957206436152003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2173,
      "step": 20322
    },
    {
      "epoch": 0.6957548784662787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.21,
      "step": 20323
    },
    {
      "epoch": 0.695789113317357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2188,
      "step": 20324
    },
    {
      "epoch": 0.6958233481684355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2042,
      "step": 20325
    },
    {
      "epoch": 0.6958575830195138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2169,
      "step": 20326
    },
    {
      "epoch": 0.6958918178705923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2175,
      "step": 20327
    },
    {
      "epoch": 0.6959260527216706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2061,
      "step": 20328
    },
    {
      "epoch": 0.6959602875727491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2158,
      "step": 20329
    },
    {
      "epoch": 0.6959945224238274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2102,
      "step": 20330
    },
    {
      "epoch": 0.6960287572749059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2006,
      "step": 20331
    },
    {
      "epoch": 0.6960629921259842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2266,
      "step": 20332
    },
    {
      "epoch": 0.6960972269770627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1983,
      "step": 20333
    },
    {
      "epoch": 0.696131461828141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2285,
      "step": 20334
    },
    {
      "epoch": 0.6961656966792195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.217,
      "step": 20335
    },
    {
      "epoch": 0.6961999315302978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2059,
      "step": 20336
    },
    {
      "epoch": 0.6962341663813763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2173,
      "step": 20337
    },
    {
      "epoch": 0.6962684012324546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2135,
      "step": 20338
    },
    {
      "epoch": 0.696302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2202,
      "step": 20339
    },
    {
      "epoch": 0.6963368709346114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2053,
      "step": 20340
    },
    {
      "epoch": 0.6963711057856898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2287,
      "step": 20341
    },
    {
      "epoch": 0.6964053406367682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2091,
      "step": 20342
    },
    {
      "epoch": 0.6964395754878466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2214,
      "step": 20343
    },
    {
      "epoch": 0.6964738103389251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2372,
      "step": 20344
    },
    {
      "epoch": 0.6965080451900034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1942,
      "step": 20345
    },
    {
      "epoch": 0.6965422800410819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.206,
      "step": 20346
    },
    {
      "epoch": 0.6965765148921602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2114,
      "step": 20347
    },
    {
      "epoch": 0.6966107497432387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2095,
      "step": 20348
    },
    {
      "epoch": 0.696644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2108,
      "step": 20349
    },
    {
      "epoch": 0.6966792194453955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2186,
      "step": 20350
    },
    {
      "epoch": 0.6967134542964738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2123,
      "step": 20351
    },
    {
      "epoch": 0.6967476891475523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2123,
      "step": 20352
    },
    {
      "epoch": 0.6967819239986306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2175,
      "step": 20353
    },
    {
      "epoch": 0.696816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2145,
      "step": 20354
    },
    {
      "epoch": 0.6968503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2275,
      "step": 20355
    },
    {
      "epoch": 0.6968846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2189,
      "step": 20356
    },
    {
      "epoch": 0.6969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2119,
      "step": 20357
    },
    {
      "epoch": 0.6969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2073,
      "step": 20358
    },
    {
      "epoch": 0.696987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2232,
      "step": 20359
    },
    {
      "epoch": 0.6970215679561794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2057,
      "step": 20360
    },
    {
      "epoch": 0.6970558028072578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2158,
      "step": 20361
    },
    {
      "epoch": 0.6970900376583362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1992,
      "step": 20362
    },
    {
      "epoch": 0.6971242725094146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2064,
      "step": 20363
    },
    {
      "epoch": 0.697158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.21,
      "step": 20364
    },
    {
      "epoch": 0.6971927422115713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2141,
      "step": 20365
    },
    {
      "epoch": 0.6972269770626498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2096,
      "step": 20366
    },
    {
      "epoch": 0.6972612119137281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2053,
      "step": 20367
    },
    {
      "epoch": 0.6972954467648066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2209,
      "step": 20368
    },
    {
      "epoch": 0.6973296816158849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2133,
      "step": 20369
    },
    {
      "epoch": 0.6973639164669634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2137,
      "step": 20370
    },
    {
      "epoch": 0.6973981513180417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2217,
      "step": 20371
    },
    {
      "epoch": 0.6974323861691202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2085,
      "step": 20372
    },
    {
      "epoch": 0.6974666210201985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2202,
      "step": 20373
    },
    {
      "epoch": 0.697500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2225,
      "step": 20374
    },
    {
      "epoch": 0.6975350907223553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1932,
      "step": 20375
    },
    {
      "epoch": 0.6975693255734338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2251,
      "step": 20376
    },
    {
      "epoch": 0.6976035604245121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2232,
      "step": 20377
    },
    {
      "epoch": 0.6976377952755906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2191,
      "step": 20378
    },
    {
      "epoch": 0.6976720301266689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2021,
      "step": 20379
    },
    {
      "epoch": 0.6977062649777473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2282,
      "step": 20380
    },
    {
      "epoch": 0.6977404998288258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2251,
      "step": 20381
    },
    {
      "epoch": 0.6977747346799041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2022,
      "step": 20382
    },
    {
      "epoch": 0.6978089695309826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2134,
      "step": 20383
    },
    {
      "epoch": 0.6978432043820609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2178,
      "step": 20384
    },
    {
      "epoch": 0.6978774392331394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2089,
      "step": 20385
    },
    {
      "epoch": 0.6979116740842177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.203,
      "step": 20386
    },
    {
      "epoch": 0.6979459089352962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2077,
      "step": 20387
    },
    {
      "epoch": 0.6979801437863745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.212,
      "step": 20388
    },
    {
      "epoch": 0.698014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1937,
      "step": 20389
    },
    {
      "epoch": 0.6980486134885313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2237,
      "step": 20390
    },
    {
      "epoch": 0.6980828483396098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1935,
      "step": 20391
    },
    {
      "epoch": 0.6981170831906881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.21,
      "step": 20392
    },
    {
      "epoch": 0.6981513180417666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2111,
      "step": 20393
    },
    {
      "epoch": 0.6981855528928449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2136,
      "step": 20394
    },
    {
      "epoch": 0.6982197877439233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2146,
      "step": 20395
    },
    {
      "epoch": 0.6982540225950017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2171,
      "step": 20396
    },
    {
      "epoch": 0.6982882574460801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.214,
      "step": 20397
    },
    {
      "epoch": 0.6983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2177,
      "step": 20398
    },
    {
      "epoch": 0.6983567271482369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2009,
      "step": 20399
    },
    {
      "epoch": 0.6983909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2127,
      "step": 20400
    },
    {
      "epoch": 0.6984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2154,
      "step": 20401
    },
    {
      "epoch": 0.6984594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2278,
      "step": 20402
    },
    {
      "epoch": 0.6984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2123,
      "step": 20403
    },
    {
      "epoch": 0.6985279014036289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2204,
      "step": 20404
    },
    {
      "epoch": 0.6985621362547073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2269,
      "step": 20405
    },
    {
      "epoch": 0.6985963711057857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2303,
      "step": 20406
    },
    {
      "epoch": 0.6986306059568641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2076,
      "step": 20407
    },
    {
      "epoch": 0.6986648408079424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2279,
      "step": 20408
    },
    {
      "epoch": 0.6986990756590209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2255,
      "step": 20409
    },
    {
      "epoch": 0.6987333105100992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1967,
      "step": 20410
    },
    {
      "epoch": 0.6987675453611777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2072,
      "step": 20411
    },
    {
      "epoch": 0.698801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.211,
      "step": 20412
    },
    {
      "epoch": 0.6988360150633345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2073,
      "step": 20413
    },
    {
      "epoch": 0.6988702499144128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2011,
      "step": 20414
    },
    {
      "epoch": 0.6989044847654913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1949,
      "step": 20415
    },
    {
      "epoch": 0.6989387196165696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2273,
      "step": 20416
    },
    {
      "epoch": 0.6989729544676481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2089,
      "step": 20417
    },
    {
      "epoch": 0.6990071893187265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2167,
      "step": 20418
    },
    {
      "epoch": 0.6990414241698049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2204,
      "step": 20419
    },
    {
      "epoch": 0.6990756590208833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.193,
      "step": 20420
    },
    {
      "epoch": 0.6991098938719617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2118,
      "step": 20421
    },
    {
      "epoch": 0.6991441287230401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2099,
      "step": 20422
    },
    {
      "epoch": 0.6991783635741184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2078,
      "step": 20423
    },
    {
      "epoch": 0.6992125984251969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2015,
      "step": 20424
    },
    {
      "epoch": 0.6992468332762752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1941,
      "step": 20425
    },
    {
      "epoch": 0.6992810681273537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2142,
      "step": 20426
    },
    {
      "epoch": 0.699315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2001,
      "step": 20427
    },
    {
      "epoch": 0.6993495378295105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1986,
      "step": 20428
    },
    {
      "epoch": 0.6993837726805888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2024,
      "step": 20429
    },
    {
      "epoch": 0.6994180075316673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2127,
      "step": 20430
    },
    {
      "epoch": 0.6994522423827456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2175,
      "step": 20431
    },
    {
      "epoch": 0.6994864772338241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2074,
      "step": 20432
    },
    {
      "epoch": 0.6995207120849024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2125,
      "step": 20433
    },
    {
      "epoch": 0.6995549469359809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2277,
      "step": 20434
    },
    {
      "epoch": 0.6995891817870592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2184,
      "step": 20435
    },
    {
      "epoch": 0.6996234166381377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2061,
      "step": 20436
    },
    {
      "epoch": 0.699657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.213,
      "step": 20437
    },
    {
      "epoch": 0.6996918863402944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2413,
      "step": 20438
    },
    {
      "epoch": 0.6997261211913728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2125,
      "step": 20439
    },
    {
      "epoch": 0.6997603560424512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2044,
      "step": 20440
    },
    {
      "epoch": 0.6997945908935296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2245,
      "step": 20441
    },
    {
      "epoch": 0.699828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2149,
      "step": 20442
    },
    {
      "epoch": 0.6998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2166,
      "step": 20443
    },
    {
      "epoch": 0.6998972954467648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2029,
      "step": 20444
    },
    {
      "epoch": 0.6999315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2004,
      "step": 20445
    },
    {
      "epoch": 0.6999657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2019,
      "step": 20446
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2083,
      "step": 20447
    },
    {
      "epoch": 0.7000342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2093,
      "step": 20448
    },
    {
      "epoch": 0.7000684697021567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2158,
      "step": 20449
    },
    {
      "epoch": 0.7001027045532352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2206,
      "step": 20450
    },
    {
      "epoch": 0.7001369394043135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1932,
      "step": 20451
    },
    {
      "epoch": 0.700171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2191,
      "step": 20452
    },
    {
      "epoch": 0.7002054091064703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2082,
      "step": 20453
    },
    {
      "epoch": 0.7002396439575488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2056,
      "step": 20454
    },
    {
      "epoch": 0.7002738788086272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2204,
      "step": 20455
    },
    {
      "epoch": 0.7003081136597056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1976,
      "step": 20456
    },
    {
      "epoch": 0.700342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2271,
      "step": 20457
    },
    {
      "epoch": 0.7003765833618624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2221,
      "step": 20458
    },
    {
      "epoch": 0.7004108182129408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2063,
      "step": 20459
    },
    {
      "epoch": 0.7004450530640192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2233,
      "step": 20460
    },
    {
      "epoch": 0.7004792879150976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2251,
      "step": 20461
    },
    {
      "epoch": 0.700513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2296,
      "step": 20462
    },
    {
      "epoch": 0.7005477576172544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2162,
      "step": 20463
    },
    {
      "epoch": 0.7005819924683327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2264,
      "step": 20464
    },
    {
      "epoch": 0.7006162273194112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2061,
      "step": 20465
    },
    {
      "epoch": 0.7006504621704895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2117,
      "step": 20466
    },
    {
      "epoch": 0.700684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1854,
      "step": 20467
    },
    {
      "epoch": 0.7007189318726463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.204,
      "step": 20468
    },
    {
      "epoch": 0.7007531667237248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2237,
      "step": 20469
    },
    {
      "epoch": 0.7007874015748031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2048,
      "step": 20470
    },
    {
      "epoch": 0.7008216364258816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2234,
      "step": 20471
    },
    {
      "epoch": 0.7008558712769599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2201,
      "step": 20472
    },
    {
      "epoch": 0.7008901061280384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2148,
      "step": 20473
    },
    {
      "epoch": 0.7009243409791167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2241,
      "step": 20474
    },
    {
      "epoch": 0.7009585758301952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2116,
      "step": 20475
    },
    {
      "epoch": 0.7009928106812735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2307,
      "step": 20476
    },
    {
      "epoch": 0.701027045532352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2161,
      "step": 20477
    },
    {
      "epoch": 0.7010612803834303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2241,
      "step": 20478
    },
    {
      "epoch": 0.7010955152345087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1994,
      "step": 20479
    },
    {
      "epoch": 0.7011297500855871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2263,
      "step": 20480
    },
    {
      "epoch": 0.7011639849366655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2052,
      "step": 20481
    },
    {
      "epoch": 0.7011982197877439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2127,
      "step": 20482
    },
    {
      "epoch": 0.7012324546388223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1911,
      "step": 20483
    },
    {
      "epoch": 0.7012666894899007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2078,
      "step": 20484
    },
    {
      "epoch": 0.7013009243409791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2297,
      "step": 20485
    },
    {
      "epoch": 0.7013351591920575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2059,
      "step": 20486
    },
    {
      "epoch": 0.7013693940431359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1976,
      "step": 20487
    },
    {
      "epoch": 0.7014036288942143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2137,
      "step": 20488
    },
    {
      "epoch": 0.7014378637452927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2097,
      "step": 20489
    },
    {
      "epoch": 0.701472098596371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2022,
      "step": 20490
    },
    {
      "epoch": 0.7015063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2093,
      "step": 20491
    },
    {
      "epoch": 0.701540568298528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.206,
      "step": 20492
    },
    {
      "epoch": 0.7015748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.213,
      "step": 20493
    },
    {
      "epoch": 0.7016090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2021,
      "step": 20494
    },
    {
      "epoch": 0.7016432728517631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2205,
      "step": 20495
    },
    {
      "epoch": 0.7016775077028415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2099,
      "step": 20496
    },
    {
      "epoch": 0.7017117425539199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.222,
      "step": 20497
    },
    {
      "epoch": 0.7017459774049983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1993,
      "step": 20498
    },
    {
      "epoch": 0.7017802122560767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2165,
      "step": 20499
    },
    {
      "epoch": 0.7018144471071551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2029,
      "step": 20500
    },
    {
      "epoch": 0.7018486819582335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2058,
      "step": 20501
    },
    {
      "epoch": 0.7018829168093119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2043,
      "step": 20502
    },
    {
      "epoch": 0.7019171516603903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2058,
      "step": 20503
    },
    {
      "epoch": 0.7019513865114687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2112,
      "step": 20504
    },
    {
      "epoch": 0.701985621362547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2066,
      "step": 20505
    },
    {
      "epoch": 0.7020198562136255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2144,
      "step": 20506
    },
    {
      "epoch": 0.7020540910647038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.21,
      "step": 20507
    },
    {
      "epoch": 0.7020883259157823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2114,
      "step": 20508
    },
    {
      "epoch": 0.7021225607668606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1998,
      "step": 20509
    },
    {
      "epoch": 0.7021567956179391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.224,
      "step": 20510
    },
    {
      "epoch": 0.7021910304690174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2119,
      "step": 20511
    },
    {
      "epoch": 0.7022252653200959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2019,
      "step": 20512
    },
    {
      "epoch": 0.7022595001711742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2107,
      "step": 20513
    },
    {
      "epoch": 0.7022937350222527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2288,
      "step": 20514
    },
    {
      "epoch": 0.702327969873331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1983,
      "step": 20515
    },
    {
      "epoch": 0.7023622047244095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2169,
      "step": 20516
    },
    {
      "epoch": 0.7023964395754878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.208,
      "step": 20517
    },
    {
      "epoch": 0.7024306744265663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2297,
      "step": 20518
    },
    {
      "epoch": 0.7024649092776446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2123,
      "step": 20519
    },
    {
      "epoch": 0.702499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1908,
      "step": 20520
    },
    {
      "epoch": 0.7025333789798014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2083,
      "step": 20521
    },
    {
      "epoch": 0.7025676138308798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2045,
      "step": 20522
    },
    {
      "epoch": 0.7026018486819582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.205,
      "step": 20523
    },
    {
      "epoch": 0.7026360835330366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2153,
      "step": 20524
    },
    {
      "epoch": 0.702670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2134,
      "step": 20525
    },
    {
      "epoch": 0.7027045532351934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2233,
      "step": 20526
    },
    {
      "epoch": 0.7027387880862718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2246,
      "step": 20527
    },
    {
      "epoch": 0.7027730229373502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2102,
      "step": 20528
    },
    {
      "epoch": 0.7028072577884287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2195,
      "step": 20529
    },
    {
      "epoch": 0.702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2047,
      "step": 20530
    },
    {
      "epoch": 0.7028757274905855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2088,
      "step": 20531
    },
    {
      "epoch": 0.7029099623416638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2148,
      "step": 20532
    },
    {
      "epoch": 0.7029441971927423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2175,
      "step": 20533
    },
    {
      "epoch": 0.7029784320438206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2044,
      "step": 20534
    },
    {
      "epoch": 0.703012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.214,
      "step": 20535
    },
    {
      "epoch": 0.7030469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2089,
      "step": 20536
    },
    {
      "epoch": 0.7030811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2157,
      "step": 20537
    },
    {
      "epoch": 0.7031153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2101,
      "step": 20538
    },
    {
      "epoch": 0.7031496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2179,
      "step": 20539
    },
    {
      "epoch": 0.703183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2073,
      "step": 20540
    },
    {
      "epoch": 0.7032180760013694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2081,
      "step": 20541
    },
    {
      "epoch": 0.7032523108524478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1905,
      "step": 20542
    },
    {
      "epoch": 0.7032865457035262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2113,
      "step": 20543
    },
    {
      "epoch": 0.7033207805546046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1963,
      "step": 20544
    },
    {
      "epoch": 0.703355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.199,
      "step": 20545
    },
    {
      "epoch": 0.7033892502567614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2135,
      "step": 20546
    },
    {
      "epoch": 0.7034234851078398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2046,
      "step": 20547
    },
    {
      "epoch": 0.7034577199589181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2211,
      "step": 20548
    },
    {
      "epoch": 0.7034919548099966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2086,
      "step": 20549
    },
    {
      "epoch": 0.7035261896610749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2022,
      "step": 20550
    },
    {
      "epoch": 0.7035604245121534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2089,
      "step": 20551
    },
    {
      "epoch": 0.7035946593632317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2224,
      "step": 20552
    },
    {
      "epoch": 0.7036288942143102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.197,
      "step": 20553
    },
    {
      "epoch": 0.7036631290653885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2002,
      "step": 20554
    },
    {
      "epoch": 0.703697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2196,
      "step": 20555
    },
    {
      "epoch": 0.7037315987675453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2243,
      "step": 20556
    },
    {
      "epoch": 0.7037658336186238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2,
      "step": 20557
    },
    {
      "epoch": 0.7038000684697021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1924,
      "step": 20558
    },
    {
      "epoch": 0.7038343033207806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1974,
      "step": 20559
    },
    {
      "epoch": 0.7038685381718589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2194,
      "step": 20560
    },
    {
      "epoch": 0.7039027730229374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1965,
      "step": 20561
    },
    {
      "epoch": 0.7039370078740157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2073,
      "step": 20562
    },
    {
      "epoch": 0.7039712427250941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2043,
      "step": 20563
    },
    {
      "epoch": 0.7040054775761725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2113,
      "step": 20564
    },
    {
      "epoch": 0.7040397124272509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.216,
      "step": 20565
    },
    {
      "epoch": 0.7040739472783294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2136,
      "step": 20566
    },
    {
      "epoch": 0.7041081821294077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2213,
      "step": 20567
    },
    {
      "epoch": 0.7041424169804862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2271,
      "step": 20568
    },
    {
      "epoch": 0.7041766518315645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2215,
      "step": 20569
    },
    {
      "epoch": 0.704210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2242,
      "step": 20570
    },
    {
      "epoch": 0.7042451215337213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2141,
      "step": 20571
    },
    {
      "epoch": 0.7042793563847998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.202,
      "step": 20572
    },
    {
      "epoch": 0.7043135912358781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1947,
      "step": 20573
    },
    {
      "epoch": 0.7043478260869566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2118,
      "step": 20574
    },
    {
      "epoch": 0.7043820609380349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2184,
      "step": 20575
    },
    {
      "epoch": 0.7044162957891134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2157,
      "step": 20576
    },
    {
      "epoch": 0.7044505306401917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2271,
      "step": 20577
    },
    {
      "epoch": 0.7044847654912701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2118,
      "step": 20578
    },
    {
      "epoch": 0.7045190003423485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2102,
      "step": 20579
    },
    {
      "epoch": 0.7045532351934269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2016,
      "step": 20580
    },
    {
      "epoch": 0.7045874700445053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2116,
      "step": 20581
    },
    {
      "epoch": 0.7046217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2064,
      "step": 20582
    },
    {
      "epoch": 0.7046559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2166,
      "step": 20583
    },
    {
      "epoch": 0.7046901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2049,
      "step": 20584
    },
    {
      "epoch": 0.7047244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2173,
      "step": 20585
    },
    {
      "epoch": 0.7047586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1977,
      "step": 20586
    },
    {
      "epoch": 0.7047928791509757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1946,
      "step": 20587
    },
    {
      "epoch": 0.7048271140020541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2017,
      "step": 20588
    },
    {
      "epoch": 0.7048613488531325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2066,
      "step": 20589
    },
    {
      "epoch": 0.7048955837042109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2114,
      "step": 20590
    },
    {
      "epoch": 0.7049298185552892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.212,
      "step": 20591
    },
    {
      "epoch": 0.7049640534063677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2116,
      "step": 20592
    },
    {
      "epoch": 0.704998288257446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2142,
      "step": 20593
    },
    {
      "epoch": 0.7050325231085245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2225,
      "step": 20594
    },
    {
      "epoch": 0.7050667579596028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1996,
      "step": 20595
    },
    {
      "epoch": 0.7051009928106813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2049,
      "step": 20596
    },
    {
      "epoch": 0.7051352276617596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2089,
      "step": 20597
    },
    {
      "epoch": 0.7051694625128381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.196,
      "step": 20598
    },
    {
      "epoch": 0.7052036973639164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2101,
      "step": 20599
    },
    {
      "epoch": 0.7052379322149949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.195,
      "step": 20600
    },
    {
      "epoch": 0.7052721670660732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2261,
      "step": 20601
    },
    {
      "epoch": 0.7053064019171517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.199,
      "step": 20602
    },
    {
      "epoch": 0.7053406367682301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2206,
      "step": 20603
    },
    {
      "epoch": 0.7053748716193085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2029,
      "step": 20604
    },
    {
      "epoch": 0.7054091064703869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2031,
      "step": 20605
    },
    {
      "epoch": 0.7054433413214652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2189,
      "step": 20606
    },
    {
      "epoch": 0.7054775761725437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2118,
      "step": 20607
    },
    {
      "epoch": 0.705511811023622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2154,
      "step": 20608
    },
    {
      "epoch": 0.7055460458747005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2164,
      "step": 20609
    },
    {
      "epoch": 0.7055802807257788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2037,
      "step": 20610
    },
    {
      "epoch": 0.7056145155768573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1993,
      "step": 20611
    },
    {
      "epoch": 0.7056487504279356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2257,
      "step": 20612
    },
    {
      "epoch": 0.7056829852790141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2188,
      "step": 20613
    },
    {
      "epoch": 0.7057172201300924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.197,
      "step": 20614
    },
    {
      "epoch": 0.7057514549811709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2075,
      "step": 20615
    },
    {
      "epoch": 0.7057856898322492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2174,
      "step": 20616
    },
    {
      "epoch": 0.7058199246833277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2153,
      "step": 20617
    },
    {
      "epoch": 0.705854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2134,
      "step": 20618
    },
    {
      "epoch": 0.7058883943854845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2125,
      "step": 20619
    },
    {
      "epoch": 0.7059226292365628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1871,
      "step": 20620
    },
    {
      "epoch": 0.7059568640876412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2057,
      "step": 20621
    },
    {
      "epoch": 0.7059910989387196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2068,
      "step": 20622
    },
    {
      "epoch": 0.706025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2154,
      "step": 20623
    },
    {
      "epoch": 0.7060595686408764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1938,
      "step": 20624
    },
    {
      "epoch": 0.7060938034919548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1905,
      "step": 20625
    },
    {
      "epoch": 0.7061280383430332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2225,
      "step": 20626
    },
    {
      "epoch": 0.7061622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.196,
      "step": 20627
    },
    {
      "epoch": 0.70619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2096,
      "step": 20628
    },
    {
      "epoch": 0.7062307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2112,
      "step": 20629
    },
    {
      "epoch": 0.7062649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.21,
      "step": 20630
    },
    {
      "epoch": 0.7062992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2067,
      "step": 20631
    },
    {
      "epoch": 0.7063334474495035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1937,
      "step": 20632
    },
    {
      "epoch": 0.706367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2049,
      "step": 20633
    },
    {
      "epoch": 0.7064019171516603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1965,
      "step": 20634
    },
    {
      "epoch": 0.7064361520027388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2064,
      "step": 20635
    },
    {
      "epoch": 0.7064703868538171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2049,
      "step": 20636
    },
    {
      "epoch": 0.7065046217048956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.213,
      "step": 20637
    },
    {
      "epoch": 0.7065388565559739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.223,
      "step": 20638
    },
    {
      "epoch": 0.7065730914070524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1977,
      "step": 20639
    },
    {
      "epoch": 0.7066073262581308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2268,
      "step": 20640
    },
    {
      "epoch": 0.7066415611092092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2121,
      "step": 20641
    },
    {
      "epoch": 0.7066757959602876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2066,
      "step": 20642
    },
    {
      "epoch": 0.706710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1983,
      "step": 20643
    },
    {
      "epoch": 0.7067442656624444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2181,
      "step": 20644
    },
    {
      "epoch": 0.7067785005135228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2128,
      "step": 20645
    },
    {
      "epoch": 0.7068127353646012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2198,
      "step": 20646
    },
    {
      "epoch": 0.7068469702156795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1972,
      "step": 20647
    },
    {
      "epoch": 0.706881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1993,
      "step": 20648
    },
    {
      "epoch": 0.7069154399178363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1996,
      "step": 20649
    },
    {
      "epoch": 0.7069496747689148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2101,
      "step": 20650
    },
    {
      "epoch": 0.7069839096199931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2242,
      "step": 20651
    },
    {
      "epoch": 0.7070181444710716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2178,
      "step": 20652
    },
    {
      "epoch": 0.7070523793221499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2222,
      "step": 20653
    },
    {
      "epoch": 0.7070866141732284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2131,
      "step": 20654
    },
    {
      "epoch": 0.7071208490243067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1947,
      "step": 20655
    },
    {
      "epoch": 0.7071550838753852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2228,
      "step": 20656
    },
    {
      "epoch": 0.7071893187264635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2097,
      "step": 20657
    },
    {
      "epoch": 0.707223553577542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2073,
      "step": 20658
    },
    {
      "epoch": 0.7072577884286203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2103,
      "step": 20659
    },
    {
      "epoch": 0.7072920232796988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2084,
      "step": 20660
    },
    {
      "epoch": 0.7073262581307771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2021,
      "step": 20661
    },
    {
      "epoch": 0.7073604929818555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2123,
      "step": 20662
    },
    {
      "epoch": 0.7073947278329339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2032,
      "step": 20663
    },
    {
      "epoch": 0.7074289626840123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2161,
      "step": 20664
    },
    {
      "epoch": 0.7074631975350907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1964,
      "step": 20665
    },
    {
      "epoch": 0.7074974323861691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.212,
      "step": 20666
    },
    {
      "epoch": 0.7075316672372475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2078,
      "step": 20667
    },
    {
      "epoch": 0.7075659020883259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2038,
      "step": 20668
    },
    {
      "epoch": 0.7076001369394043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2132,
      "step": 20669
    },
    {
      "epoch": 0.7076343717904827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2131,
      "step": 20670
    },
    {
      "epoch": 0.7076686066415611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2062,
      "step": 20671
    },
    {
      "epoch": 0.7077028414926395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2072,
      "step": 20672
    },
    {
      "epoch": 0.7077370763437179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2056,
      "step": 20673
    },
    {
      "epoch": 0.7077713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.193,
      "step": 20674
    },
    {
      "epoch": 0.7078055460458746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2192,
      "step": 20675
    },
    {
      "epoch": 0.7078397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1877,
      "step": 20676
    },
    {
      "epoch": 0.7078740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1986,
      "step": 20677
    },
    {
      "epoch": 0.7079082505991099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2061,
      "step": 20678
    },
    {
      "epoch": 0.7079424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.213,
      "step": 20679
    },
    {
      "epoch": 0.7079767203012667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2203,
      "step": 20680
    },
    {
      "epoch": 0.7080109551523451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2124,
      "step": 20681
    },
    {
      "epoch": 0.7080451900034235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2279,
      "step": 20682
    },
    {
      "epoch": 0.7080794248545019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2103,
      "step": 20683
    },
    {
      "epoch": 0.7081136597055803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2244,
      "step": 20684
    },
    {
      "epoch": 0.7081478945566587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2105,
      "step": 20685
    },
    {
      "epoch": 0.7081821294077371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2161,
      "step": 20686
    },
    {
      "epoch": 0.7082163642588155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1955,
      "step": 20687
    },
    {
      "epoch": 0.7082505991098939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1925,
      "step": 20688
    },
    {
      "epoch": 0.7082848339609723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2112,
      "step": 20689
    },
    {
      "epoch": 0.7083190688120506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1937,
      "step": 20690
    },
    {
      "epoch": 0.7083533036631291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2287,
      "step": 20691
    },
    {
      "epoch": 0.7083875385142074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2047,
      "step": 20692
    },
    {
      "epoch": 0.7084217733652859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1977,
      "step": 20693
    },
    {
      "epoch": 0.7084560082163642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2144,
      "step": 20694
    },
    {
      "epoch": 0.7084902430674427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.208,
      "step": 20695
    },
    {
      "epoch": 0.708524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2194,
      "step": 20696
    },
    {
      "epoch": 0.7085587127695995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.203,
      "step": 20697
    },
    {
      "epoch": 0.7085929476206778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2169,
      "step": 20698
    },
    {
      "epoch": 0.7086271824717563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1968,
      "step": 20699
    },
    {
      "epoch": 0.7086614173228346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2019,
      "step": 20700
    },
    {
      "epoch": 0.7086956521739131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2138,
      "step": 20701
    },
    {
      "epoch": 0.7087298870249914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2162,
      "step": 20702
    },
    {
      "epoch": 0.7087641218760699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.209,
      "step": 20703
    },
    {
      "epoch": 0.7087983567271482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2207,
      "step": 20704
    },
    {
      "epoch": 0.7088325915782266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2162,
      "step": 20705
    },
    {
      "epoch": 0.708866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2119,
      "step": 20706
    },
    {
      "epoch": 0.7089010612803834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2104,
      "step": 20707
    },
    {
      "epoch": 0.7089352961314618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.204,
      "step": 20708
    },
    {
      "epoch": 0.7089695309825402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1899,
      "step": 20709
    },
    {
      "epoch": 0.7090037658336186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2076,
      "step": 20710
    },
    {
      "epoch": 0.709038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1854,
      "step": 20711
    },
    {
      "epoch": 0.7090722355357755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2096,
      "step": 20712
    },
    {
      "epoch": 0.7091064703868538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2176,
      "step": 20713
    },
    {
      "epoch": 0.7091407052379323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2291,
      "step": 20714
    },
    {
      "epoch": 0.7091749400890106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2102,
      "step": 20715
    },
    {
      "epoch": 0.7092091749400891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2145,
      "step": 20716
    },
    {
      "epoch": 0.7092434097911674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2058,
      "step": 20717
    },
    {
      "epoch": 0.7092776446422459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2027,
      "step": 20718
    },
    {
      "epoch": 0.7093118794933242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2033,
      "step": 20719
    },
    {
      "epoch": 0.7093461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2194,
      "step": 20720
    },
    {
      "epoch": 0.709380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2172,
      "step": 20721
    },
    {
      "epoch": 0.7094145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.197,
      "step": 20722
    },
    {
      "epoch": 0.7094488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2316,
      "step": 20723
    },
    {
      "epoch": 0.7094830537487162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2049,
      "step": 20724
    },
    {
      "epoch": 0.7095172885997946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2112,
      "step": 20725
    },
    {
      "epoch": 0.709551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1954,
      "step": 20726
    },
    {
      "epoch": 0.7095857583019514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1928,
      "step": 20727
    },
    {
      "epoch": 0.7096199931530298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2065,
      "step": 20728
    },
    {
      "epoch": 0.7096542280041082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2073,
      "step": 20729
    },
    {
      "epoch": 0.7096884628551866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2056,
      "step": 20730
    },
    {
      "epoch": 0.709722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2052,
      "step": 20731
    },
    {
      "epoch": 0.7097569325573434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2021,
      "step": 20732
    },
    {
      "epoch": 0.7097911674084217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2083,
      "step": 20733
    },
    {
      "epoch": 0.7098254022595002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1993,
      "step": 20734
    },
    {
      "epoch": 0.7098596371105785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2168,
      "step": 20735
    },
    {
      "epoch": 0.709893871961657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2069,
      "step": 20736
    },
    {
      "epoch": 0.7099281068127353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.223,
      "step": 20737
    },
    {
      "epoch": 0.7099623416638138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2126,
      "step": 20738
    },
    {
      "epoch": 0.7099965765148921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2081,
      "step": 20739
    },
    {
      "epoch": 0.7100308113659706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1906,
      "step": 20740
    },
    {
      "epoch": 0.7100650462170489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2016,
      "step": 20741
    },
    {
      "epoch": 0.7100992810681274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.211,
      "step": 20742
    },
    {
      "epoch": 0.7101335159192057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2075,
      "step": 20743
    },
    {
      "epoch": 0.7101677507702842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2219,
      "step": 20744
    },
    {
      "epoch": 0.7102019856213625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2245,
      "step": 20745
    },
    {
      "epoch": 0.710236220472441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2188,
      "step": 20746
    },
    {
      "epoch": 0.7102704553235193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.196,
      "step": 20747
    },
    {
      "epoch": 0.7103046901745977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.211,
      "step": 20748
    },
    {
      "epoch": 0.7103389250256762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2,
      "step": 20749
    },
    {
      "epoch": 0.7103731598767545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2328,
      "step": 20750
    },
    {
      "epoch": 0.710407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2007,
      "step": 20751
    },
    {
      "epoch": 0.7104416295789113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2172,
      "step": 20752
    },
    {
      "epoch": 0.7104758644299898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2071,
      "step": 20753
    },
    {
      "epoch": 0.7105100992810681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2118,
      "step": 20754
    },
    {
      "epoch": 0.7105443341321466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1993,
      "step": 20755
    },
    {
      "epoch": 0.7105785689832249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.22,
      "step": 20756
    },
    {
      "epoch": 0.7106128038343034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2222,
      "step": 20757
    },
    {
      "epoch": 0.7106470386853817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2055,
      "step": 20758
    },
    {
      "epoch": 0.7106812735364602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2053,
      "step": 20759
    },
    {
      "epoch": 0.7107155083875385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.212,
      "step": 20760
    },
    {
      "epoch": 0.710749743238617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2068,
      "step": 20761
    },
    {
      "epoch": 0.7107839780896953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2013,
      "step": 20762
    },
    {
      "epoch": 0.7108182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2006,
      "step": 20763
    },
    {
      "epoch": 0.7108524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2063,
      "step": 20764
    },
    {
      "epoch": 0.7108866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2137,
      "step": 20765
    },
    {
      "epoch": 0.7109209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1989,
      "step": 20766
    },
    {
      "epoch": 0.7109551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1933,
      "step": 20767
    },
    {
      "epoch": 0.7109893871961657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2286,
      "step": 20768
    },
    {
      "epoch": 0.7110236220472441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2131,
      "step": 20769
    },
    {
      "epoch": 0.7110578568983225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2145,
      "step": 20770
    },
    {
      "epoch": 0.7110920917494009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1965,
      "step": 20771
    },
    {
      "epoch": 0.7111263266004793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2237,
      "step": 20772
    },
    {
      "epoch": 0.7111605614515577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2165,
      "step": 20773
    },
    {
      "epoch": 0.711194796302636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1959,
      "step": 20774
    },
    {
      "epoch": 0.7112290311537145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2048,
      "step": 20775
    },
    {
      "epoch": 0.7112632660047928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2169,
      "step": 20776
    },
    {
      "epoch": 0.7112975008558713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2101,
      "step": 20777
    },
    {
      "epoch": 0.7113317357069496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2193,
      "step": 20778
    },
    {
      "epoch": 0.7113659705580281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1953,
      "step": 20779
    },
    {
      "epoch": 0.7114002054091064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2112,
      "step": 20780
    },
    {
      "epoch": 0.7114344402601849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.223,
      "step": 20781
    },
    {
      "epoch": 0.7114686751112632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1999,
      "step": 20782
    },
    {
      "epoch": 0.7115029099623417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2108,
      "step": 20783
    },
    {
      "epoch": 0.71153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2123,
      "step": 20784
    },
    {
      "epoch": 0.7115713796644985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2241,
      "step": 20785
    },
    {
      "epoch": 0.7116056145155769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2059,
      "step": 20786
    },
    {
      "epoch": 0.7116398493666553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1996,
      "step": 20787
    },
    {
      "epoch": 0.7116740842177337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1941,
      "step": 20788
    },
    {
      "epoch": 0.711708319068812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1817,
      "step": 20789
    },
    {
      "epoch": 0.7117425539198905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2123,
      "step": 20790
    },
    {
      "epoch": 0.7117767887709688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2094,
      "step": 20791
    },
    {
      "epoch": 0.7118110236220473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2133,
      "step": 20792
    },
    {
      "epoch": 0.7118452584731256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1992,
      "step": 20793
    },
    {
      "epoch": 0.7118794933242041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.204,
      "step": 20794
    },
    {
      "epoch": 0.7119137281752824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2042,
      "step": 20795
    },
    {
      "epoch": 0.7119479630263609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1934,
      "step": 20796
    },
    {
      "epoch": 0.7119821978774392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2068,
      "step": 20797
    },
    {
      "epoch": 0.7120164327285177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2116,
      "step": 20798
    },
    {
      "epoch": 0.712050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2162,
      "step": 20799
    },
    {
      "epoch": 0.7120849024306745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1997,
      "step": 20800
    },
    {
      "epoch": 0.7121191372817528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2002,
      "step": 20801
    },
    {
      "epoch": 0.7121533721328313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1965,
      "step": 20802
    },
    {
      "epoch": 0.7121876069839096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.214,
      "step": 20803
    },
    {
      "epoch": 0.712221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1949,
      "step": 20804
    },
    {
      "epoch": 0.7122560766860664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1866,
      "step": 20805
    },
    {
      "epoch": 0.7122903115371448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.21,
      "step": 20806
    },
    {
      "epoch": 0.7123245463882232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1992,
      "step": 20807
    },
    {
      "epoch": 0.7123587812393016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1975,
      "step": 20808
    },
    {
      "epoch": 0.71239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1867,
      "step": 20809
    },
    {
      "epoch": 0.7124272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1913,
      "step": 20810
    },
    {
      "epoch": 0.7124614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2081,
      "step": 20811
    },
    {
      "epoch": 0.7124957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1952,
      "step": 20812
    },
    {
      "epoch": 0.7125299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1926,
      "step": 20813
    },
    {
      "epoch": 0.712564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2129,
      "step": 20814
    },
    {
      "epoch": 0.7125984251968503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2013,
      "step": 20815
    },
    {
      "epoch": 0.7126326600479288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2076,
      "step": 20816
    },
    {
      "epoch": 0.7126668948990071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1909,
      "step": 20817
    },
    {
      "epoch": 0.7127011297500856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2075,
      "step": 20818
    },
    {
      "epoch": 0.7127353646011639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1817,
      "step": 20819
    },
    {
      "epoch": 0.7127695994522424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2056,
      "step": 20820
    },
    {
      "epoch": 0.7128038343033207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1996,
      "step": 20821
    },
    {
      "epoch": 0.7128380691543992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1966,
      "step": 20822
    },
    {
      "epoch": 0.7128723040054776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1935,
      "step": 20823
    },
    {
      "epoch": 0.712906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1956,
      "step": 20824
    },
    {
      "epoch": 0.7129407737076344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2108,
      "step": 20825
    },
    {
      "epoch": 0.7129750085587128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2092,
      "step": 20826
    },
    {
      "epoch": 0.7130092434097912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2076,
      "step": 20827
    },
    {
      "epoch": 0.7130434782608696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2125,
      "step": 20828
    },
    {
      "epoch": 0.713077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2035,
      "step": 20829
    },
    {
      "epoch": 0.7131119479630263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2172,
      "step": 20830
    },
    {
      "epoch": 0.7131461828141048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2025,
      "step": 20831
    },
    {
      "epoch": 0.7131804176651831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1984,
      "step": 20832
    },
    {
      "epoch": 0.7132146525162616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1993,
      "step": 20833
    },
    {
      "epoch": 0.7132488873673399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2009,
      "step": 20834
    },
    {
      "epoch": 0.7132831222184184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.213,
      "step": 20835
    },
    {
      "epoch": 0.7133173570694967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2201,
      "step": 20836
    },
    {
      "epoch": 0.7133515919205752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1894,
      "step": 20837
    },
    {
      "epoch": 0.7133858267716535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2064,
      "step": 20838
    },
    {
      "epoch": 0.713420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2031,
      "step": 20839
    },
    {
      "epoch": 0.7134542964738103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2032,
      "step": 20840
    },
    {
      "epoch": 0.7134885313248888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2113,
      "step": 20841
    },
    {
      "epoch": 0.7135227661759671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2066,
      "step": 20842
    },
    {
      "epoch": 0.7135570010270456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2068,
      "step": 20843
    },
    {
      "epoch": 0.7135912358781239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2061,
      "step": 20844
    },
    {
      "epoch": 0.7136254707292023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2004,
      "step": 20845
    },
    {
      "epoch": 0.7136597055802807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1761,
      "step": 20846
    },
    {
      "epoch": 0.7136939404313591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1984,
      "step": 20847
    },
    {
      "epoch": 0.7137281752824375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2025,
      "step": 20848
    },
    {
      "epoch": 0.7137624101335159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2121,
      "step": 20849
    },
    {
      "epoch": 0.7137966449845943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2108,
      "step": 20850
    },
    {
      "epoch": 0.7138308798356727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1873,
      "step": 20851
    },
    {
      "epoch": 0.7138651146867511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2038,
      "step": 20852
    },
    {
      "epoch": 0.7138993495378295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2044,
      "step": 20853
    },
    {
      "epoch": 0.7139335843889079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2113,
      "step": 20854
    },
    {
      "epoch": 0.7139678192399863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1944,
      "step": 20855
    },
    {
      "epoch": 0.7140020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2056,
      "step": 20856
    },
    {
      "epoch": 0.7140362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1977,
      "step": 20857
    },
    {
      "epoch": 0.7140705237932214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2127,
      "step": 20858
    },
    {
      "epoch": 0.7141047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1922,
      "step": 20859
    },
    {
      "epoch": 0.7141389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2075,
      "step": 20860
    },
    {
      "epoch": 0.7141732283464567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1999,
      "step": 20861
    },
    {
      "epoch": 0.7142074631975351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1972,
      "step": 20862
    },
    {
      "epoch": 0.7142416980486135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2027,
      "step": 20863
    },
    {
      "epoch": 0.7142759328996919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1924,
      "step": 20864
    },
    {
      "epoch": 0.7143101677507703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1989,
      "step": 20865
    },
    {
      "epoch": 0.7143444026018487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.211,
      "step": 20866
    },
    {
      "epoch": 0.7143786374529271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2124,
      "step": 20867
    },
    {
      "epoch": 0.7144128723040055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2105,
      "step": 20868
    },
    {
      "epoch": 0.7144471071550839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1973,
      "step": 20869
    },
    {
      "epoch": 0.7144813420061623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1982,
      "step": 20870
    },
    {
      "epoch": 0.7145155768572407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1829,
      "step": 20871
    },
    {
      "epoch": 0.7145498117083191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2027,
      "step": 20872
    },
    {
      "epoch": 0.7145840465593974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2168,
      "step": 20873
    },
    {
      "epoch": 0.7146182814104759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2013,
      "step": 20874
    },
    {
      "epoch": 0.7146525162615542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2106,
      "step": 20875
    },
    {
      "epoch": 0.7146867511126327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.202,
      "step": 20876
    },
    {
      "epoch": 0.714720985963711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.193,
      "step": 20877
    },
    {
      "epoch": 0.7147552208147895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2111,
      "step": 20878
    },
    {
      "epoch": 0.7147894556658678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1995,
      "step": 20879
    },
    {
      "epoch": 0.7148236905169463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2233,
      "step": 20880
    },
    {
      "epoch": 0.7148579253680246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1986,
      "step": 20881
    },
    {
      "epoch": 0.7148921602191031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2124,
      "step": 20882
    },
    {
      "epoch": 0.7149263950701814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2072,
      "step": 20883
    },
    {
      "epoch": 0.7149606299212599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2191,
      "step": 20884
    },
    {
      "epoch": 0.7149948647723382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1934,
      "step": 20885
    },
    {
      "epoch": 0.7150290996234167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2076,
      "step": 20886
    },
    {
      "epoch": 0.715063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1901,
      "step": 20887
    },
    {
      "epoch": 0.7150975693255734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2124,
      "step": 20888
    },
    {
      "epoch": 0.7151318041766518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2028,
      "step": 20889
    },
    {
      "epoch": 0.7151660390277302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.206,
      "step": 20890
    },
    {
      "epoch": 0.7152002738788086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2045,
      "step": 20891
    },
    {
      "epoch": 0.715234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2087,
      "step": 20892
    },
    {
      "epoch": 0.7152687435809654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2071,
      "step": 20893
    },
    {
      "epoch": 0.7153029784320438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2057,
      "step": 20894
    },
    {
      "epoch": 0.7153372132831222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2097,
      "step": 20895
    },
    {
      "epoch": 0.7153714481342006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2026,
      "step": 20896
    },
    {
      "epoch": 0.7154056829852791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1915,
      "step": 20897
    },
    {
      "epoch": 0.7154399178363574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1943,
      "step": 20898
    },
    {
      "epoch": 0.7154741526874359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2184,
      "step": 20899
    },
    {
      "epoch": 0.7155083875385142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2004,
      "step": 20900
    },
    {
      "epoch": 0.7155426223895927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2017,
      "step": 20901
    },
    {
      "epoch": 0.715576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2021,
      "step": 20902
    },
    {
      "epoch": 0.7156110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.188,
      "step": 20903
    },
    {
      "epoch": 0.7156453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2032,
      "step": 20904
    },
    {
      "epoch": 0.7156795617939062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.191,
      "step": 20905
    },
    {
      "epoch": 0.7157137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1901,
      "step": 20906
    },
    {
      "epoch": 0.715748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.191,
      "step": 20907
    },
    {
      "epoch": 0.7157822663471414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2193,
      "step": 20908
    },
    {
      "epoch": 0.7158165011982198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.222,
      "step": 20909
    },
    {
      "epoch": 0.7158507360492982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2156,
      "step": 20910
    },
    {
      "epoch": 0.7158849709003766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2059,
      "step": 20911
    },
    {
      "epoch": 0.715919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1982,
      "step": 20912
    },
    {
      "epoch": 0.7159534406025334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2063,
      "step": 20913
    },
    {
      "epoch": 0.7159876754536117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2205,
      "step": 20914
    },
    {
      "epoch": 0.7160219103046902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1987,
      "step": 20915
    },
    {
      "epoch": 0.7160561451557685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2036,
      "step": 20916
    },
    {
      "epoch": 0.716090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2018,
      "step": 20917
    },
    {
      "epoch": 0.7161246148579253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2138,
      "step": 20918
    },
    {
      "epoch": 0.7161588497090038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2002,
      "step": 20919
    },
    {
      "epoch": 0.7161930845600821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2088,
      "step": 20920
    },
    {
      "epoch": 0.7162273194111606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.202,
      "step": 20921
    },
    {
      "epoch": 0.7162615542622389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2,
      "step": 20922
    },
    {
      "epoch": 0.7162957891133174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1871,
      "step": 20923
    },
    {
      "epoch": 0.7163300239643957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2108,
      "step": 20924
    },
    {
      "epoch": 0.7163642588154742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2069,
      "step": 20925
    },
    {
      "epoch": 0.7163984936665525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1963,
      "step": 20926
    },
    {
      "epoch": 0.716432728517631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2022,
      "step": 20927
    },
    {
      "epoch": 0.7164669633687093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1925,
      "step": 20928
    },
    {
      "epoch": 0.7165011982197877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2091,
      "step": 20929
    },
    {
      "epoch": 0.7165354330708661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2004,
      "step": 20930
    },
    {
      "epoch": 0.7165696679219445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1995,
      "step": 20931
    },
    {
      "epoch": 0.7166039027730229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1899,
      "step": 20932
    },
    {
      "epoch": 0.7166381376241013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2102,
      "step": 20933
    },
    {
      "epoch": 0.7166723724751798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1865,
      "step": 20934
    },
    {
      "epoch": 0.7167066073262581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2072,
      "step": 20935
    },
    {
      "epoch": 0.7167408421773366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2004,
      "step": 20936
    },
    {
      "epoch": 0.7167750770284149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1956,
      "step": 20937
    },
    {
      "epoch": 0.7168093118794934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1876,
      "step": 20938
    },
    {
      "epoch": 0.7168435467305717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1899,
      "step": 20939
    },
    {
      "epoch": 0.7168777815816502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1928,
      "step": 20940
    },
    {
      "epoch": 0.7169120164327285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1874,
      "step": 20941
    },
    {
      "epoch": 0.716946251283807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2098,
      "step": 20942
    },
    {
      "epoch": 0.7169804861348853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.196,
      "step": 20943
    },
    {
      "epoch": 0.7170147209859637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1895,
      "step": 20944
    },
    {
      "epoch": 0.7170489558370421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2029,
      "step": 20945
    },
    {
      "epoch": 0.7170831906881205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2094,
      "step": 20946
    },
    {
      "epoch": 0.7171174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.219,
      "step": 20947
    },
    {
      "epoch": 0.7171516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1889,
      "step": 20948
    },
    {
      "epoch": 0.7171858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1879,
      "step": 20949
    },
    {
      "epoch": 0.7172201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2058,
      "step": 20950
    },
    {
      "epoch": 0.7172543649435125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2009,
      "step": 20951
    },
    {
      "epoch": 0.7172885997945909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1897,
      "step": 20952
    },
    {
      "epoch": 0.7173228346456693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.199,
      "step": 20953
    },
    {
      "epoch": 0.7173570694967477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2139,
      "step": 20954
    },
    {
      "epoch": 0.717391304347826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2107,
      "step": 20955
    },
    {
      "epoch": 0.7174255391989045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2133,
      "step": 20956
    },
    {
      "epoch": 0.7174597740499828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.208,
      "step": 20957
    },
    {
      "epoch": 0.7174940089010613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1957,
      "step": 20958
    },
    {
      "epoch": 0.7175282437521396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1963,
      "step": 20959
    },
    {
      "epoch": 0.7175624786032181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1907,
      "step": 20960
    },
    {
      "epoch": 0.7175967134542964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2139,
      "step": 20961
    },
    {
      "epoch": 0.7176309483053749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2118,
      "step": 20962
    },
    {
      "epoch": 0.7176651831564532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2131,
      "step": 20963
    },
    {
      "epoch": 0.7176994180075317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1998,
      "step": 20964
    },
    {
      "epoch": 0.71773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1948,
      "step": 20965
    },
    {
      "epoch": 0.7177678877096885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1855,
      "step": 20966
    },
    {
      "epoch": 0.7178021225607668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1923,
      "step": 20967
    },
    {
      "epoch": 0.7178363574118453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1812,
      "step": 20968
    },
    {
      "epoch": 0.7178705922629236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1838,
      "step": 20969
    },
    {
      "epoch": 0.717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1995,
      "step": 20970
    },
    {
      "epoch": 0.7179390619650805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2103,
      "step": 20971
    },
    {
      "epoch": 0.7179732968161588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1986,
      "step": 20972
    },
    {
      "epoch": 0.7180075316672373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2059,
      "step": 20973
    },
    {
      "epoch": 0.7180417665183156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2077,
      "step": 20974
    },
    {
      "epoch": 0.7180760013693941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.187,
      "step": 20975
    },
    {
      "epoch": 0.7181102362204724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2045,
      "step": 20976
    },
    {
      "epoch": 0.7181444710715509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2031,
      "step": 20977
    },
    {
      "epoch": 0.7181787059226292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1994,
      "step": 20978
    },
    {
      "epoch": 0.7182129407737077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2023,
      "step": 20979
    },
    {
      "epoch": 0.718247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1956,
      "step": 20980
    },
    {
      "epoch": 0.7182814104758645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1928,
      "step": 20981
    },
    {
      "epoch": 0.7183156453269428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1908,
      "step": 20982
    },
    {
      "epoch": 0.7183498801780213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2115,
      "step": 20983
    },
    {
      "epoch": 0.7183841150290996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1775,
      "step": 20984
    },
    {
      "epoch": 0.718418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1919,
      "step": 20985
    },
    {
      "epoch": 0.7184525847312564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1996,
      "step": 20986
    },
    {
      "epoch": 0.7184868195823348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2089,
      "step": 20987
    },
    {
      "epoch": 0.7185210544334132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1984,
      "step": 20988
    },
    {
      "epoch": 0.7185552892844916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1943,
      "step": 20989
    },
    {
      "epoch": 0.71858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1912,
      "step": 20990
    },
    {
      "epoch": 0.7186237589866484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2027,
      "step": 20991
    },
    {
      "epoch": 0.7186579938377268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1885,
      "step": 20992
    },
    {
      "epoch": 0.7186922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1919,
      "step": 20993
    },
    {
      "epoch": 0.7187264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1978,
      "step": 20994
    },
    {
      "epoch": 0.718760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2048,
      "step": 20995
    },
    {
      "epoch": 0.7187949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.21,
      "step": 20996
    },
    {
      "epoch": 0.7188291680931188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.222,
      "step": 20997
    },
    {
      "epoch": 0.7188634029441971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1925,
      "step": 20998
    },
    {
      "epoch": 0.7188976377952756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1897,
      "step": 20999
    },
    {
      "epoch": 0.7189318726463539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2043,
      "step": 21000
    },
    {
      "epoch": 0.7189661074974324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.206,
      "step": 21001
    },
    {
      "epoch": 0.7190003423485107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2229,
      "step": 21002
    },
    {
      "epoch": 0.7190345771995892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1676,
      "step": 21003
    },
    {
      "epoch": 0.7190688120506675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2019,
      "step": 21004
    },
    {
      "epoch": 0.719103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2165,
      "step": 21005
    },
    {
      "epoch": 0.7191372817528243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1954,
      "step": 21006
    },
    {
      "epoch": 0.7191715166039028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1962,
      "step": 21007
    },
    {
      "epoch": 0.7192057514549812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2048,
      "step": 21008
    },
    {
      "epoch": 0.7192399863060596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2045,
      "step": 21009
    },
    {
      "epoch": 0.719274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2081,
      "step": 21010
    },
    {
      "epoch": 0.7193084560082164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.207,
      "step": 21011
    },
    {
      "epoch": 0.7193426908592948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2092,
      "step": 21012
    },
    {
      "epoch": 0.7193769257103731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2039,
      "step": 21013
    },
    {
      "epoch": 0.7194111605614516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2016,
      "step": 21014
    },
    {
      "epoch": 0.7194453954125299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2049,
      "step": 21015
    },
    {
      "epoch": 0.7194796302636084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1969,
      "step": 21016
    },
    {
      "epoch": 0.7195138651146867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.182,
      "step": 21017
    },
    {
      "epoch": 0.7195480999657652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1946,
      "step": 21018
    },
    {
      "epoch": 0.7195823348168435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2099,
      "step": 21019
    },
    {
      "epoch": 0.719616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2091,
      "step": 21020
    },
    {
      "epoch": 0.7196508045190003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1952,
      "step": 21021
    },
    {
      "epoch": 0.7196850393700788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2028,
      "step": 21022
    },
    {
      "epoch": 0.7197192742211571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.205,
      "step": 21023
    },
    {
      "epoch": 0.7197535090722356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2114,
      "step": 21024
    },
    {
      "epoch": 0.7197877439233139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2015,
      "step": 21025
    },
    {
      "epoch": 0.7198219787743924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2046,
      "step": 21026
    },
    {
      "epoch": 0.7198562136254707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1806,
      "step": 21027
    },
    {
      "epoch": 0.7198904484765491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1925,
      "step": 21028
    },
    {
      "epoch": 0.7199246833276275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1927,
      "step": 21029
    },
    {
      "epoch": 0.7199589181787059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2112,
      "step": 21030
    },
    {
      "epoch": 0.7199931530297843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1974,
      "step": 21031
    },
    {
      "epoch": 0.7200273878808627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1861,
      "step": 21032
    },
    {
      "epoch": 0.7200616227319411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2115,
      "step": 21033
    },
    {
      "epoch": 0.7200958575830195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2044,
      "step": 21034
    },
    {
      "epoch": 0.7201300924340979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1922,
      "step": 21035
    },
    {
      "epoch": 0.7201643272851763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2014,
      "step": 21036
    },
    {
      "epoch": 0.7201985621362547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2057,
      "step": 21037
    },
    {
      "epoch": 0.7202327969873331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2019,
      "step": 21038
    },
    {
      "epoch": 0.7202670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1809,
      "step": 21039
    },
    {
      "epoch": 0.7203012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2123,
      "step": 21040
    },
    {
      "epoch": 0.7203355015405682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2231,
      "step": 21041
    },
    {
      "epoch": 0.7203697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1935,
      "step": 21042
    },
    {
      "epoch": 0.7204039712427251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1801,
      "step": 21043
    },
    {
      "epoch": 0.7204382060938035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2061,
      "step": 21044
    },
    {
      "epoch": 0.7204724409448819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1873,
      "step": 21045
    },
    {
      "epoch": 0.7205066757959603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2049,
      "step": 21046
    },
    {
      "epoch": 0.7205409106470387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1917,
      "step": 21047
    },
    {
      "epoch": 0.7205751454981171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1945,
      "step": 21048
    },
    {
      "epoch": 0.7206093803491955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1853,
      "step": 21049
    },
    {
      "epoch": 0.7206436152002739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1947,
      "step": 21050
    },
    {
      "epoch": 0.7206778500513523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1983,
      "step": 21051
    },
    {
      "epoch": 0.7207120849024307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1939,
      "step": 21052
    },
    {
      "epoch": 0.7207463197535091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2047,
      "step": 21053
    },
    {
      "epoch": 0.7207805546045875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2072,
      "step": 21054
    },
    {
      "epoch": 0.7208147894556659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2033,
      "step": 21055
    },
    {
      "epoch": 0.7208490243067442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2091,
      "step": 21056
    },
    {
      "epoch": 0.7208832591578227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1924,
      "step": 21057
    },
    {
      "epoch": 0.720917494008901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1979,
      "step": 21058
    },
    {
      "epoch": 0.7209517288599795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1888,
      "step": 21059
    },
    {
      "epoch": 0.7209859637110578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1856,
      "step": 21060
    },
    {
      "epoch": 0.7210201985621363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2037,
      "step": 21061
    },
    {
      "epoch": 0.7210544334132146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1942,
      "step": 21062
    },
    {
      "epoch": 0.7210886682642931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1866,
      "step": 21063
    },
    {
      "epoch": 0.7211229031153714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1985,
      "step": 21064
    },
    {
      "epoch": 0.7211571379664499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1863,
      "step": 21065
    },
    {
      "epoch": 0.7211913728175282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2031,
      "step": 21066
    },
    {
      "epoch": 0.7212256076686067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1879,
      "step": 21067
    },
    {
      "epoch": 0.721259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1983,
      "step": 21068
    },
    {
      "epoch": 0.7212940773707635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2096,
      "step": 21069
    },
    {
      "epoch": 0.7213283122218418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1918,
      "step": 21070
    },
    {
      "epoch": 0.7213625470729202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2089,
      "step": 21071
    },
    {
      "epoch": 0.7213967819239986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2096,
      "step": 21072
    },
    {
      "epoch": 0.721431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2011,
      "step": 21073
    },
    {
      "epoch": 0.7214652516261554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2052,
      "step": 21074
    },
    {
      "epoch": 0.7214994864772338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1942,
      "step": 21075
    },
    {
      "epoch": 0.7215337213283122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1939,
      "step": 21076
    },
    {
      "epoch": 0.7215679561793906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2109,
      "step": 21077
    },
    {
      "epoch": 0.721602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1884,
      "step": 21078
    },
    {
      "epoch": 0.7216364258815474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1982,
      "step": 21079
    },
    {
      "epoch": 0.7216706607326259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2011,
      "step": 21080
    },
    {
      "epoch": 0.7217048955837042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2044,
      "step": 21081
    },
    {
      "epoch": 0.7217391304347827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.201,
      "step": 21082
    },
    {
      "epoch": 0.721773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1976,
      "step": 21083
    },
    {
      "epoch": 0.7218076001369395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1896,
      "step": 21084
    },
    {
      "epoch": 0.7218418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.198,
      "step": 21085
    },
    {
      "epoch": 0.7218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2138,
      "step": 21086
    },
    {
      "epoch": 0.7219103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.188,
      "step": 21087
    },
    {
      "epoch": 0.721944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2046,
      "step": 21088
    },
    {
      "epoch": 0.7219787743923314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1998,
      "step": 21089
    },
    {
      "epoch": 0.7220130092434098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2092,
      "step": 21090
    },
    {
      "epoch": 0.7220472440944882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.184,
      "step": 21091
    },
    {
      "epoch": 0.7220814789455666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.178,
      "step": 21092
    },
    {
      "epoch": 0.722115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1939,
      "step": 21093
    },
    {
      "epoch": 0.7221499486477234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2078,
      "step": 21094
    },
    {
      "epoch": 0.7221841834988018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1946,
      "step": 21095
    },
    {
      "epoch": 0.7222184183498802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2085,
      "step": 21096
    },
    {
      "epoch": 0.7222526532009585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1874,
      "step": 21097
    },
    {
      "epoch": 0.722286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2024,
      "step": 21098
    },
    {
      "epoch": 0.7223211229031153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2012,
      "step": 21099
    },
    {
      "epoch": 0.7223553577541938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1975,
      "step": 21100
    },
    {
      "epoch": 0.7223895926052721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2149,
      "step": 21101
    },
    {
      "epoch": 0.7224238274563506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1892,
      "step": 21102
    },
    {
      "epoch": 0.7224580623074289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1942,
      "step": 21103
    },
    {
      "epoch": 0.7224922971585074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2035,
      "step": 21104
    },
    {
      "epoch": 0.7225265320095857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2004,
      "step": 21105
    },
    {
      "epoch": 0.7225607668606642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1941,
      "step": 21106
    },
    {
      "epoch": 0.7225950017117425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2067,
      "step": 21107
    },
    {
      "epoch": 0.722629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1792,
      "step": 21108
    },
    {
      "epoch": 0.7226634714138993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.201,
      "step": 21109
    },
    {
      "epoch": 0.7226977062649778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2029,
      "step": 21110
    },
    {
      "epoch": 0.7227319411160561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1983,
      "step": 21111
    },
    {
      "epoch": 0.7227661759671345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2022,
      "step": 21112
    },
    {
      "epoch": 0.7228004108182129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1894,
      "step": 21113
    },
    {
      "epoch": 0.7228346456692913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1964,
      "step": 21114
    },
    {
      "epoch": 0.7228688805203697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2098,
      "step": 21115
    },
    {
      "epoch": 0.7229031153714481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2074,
      "step": 21116
    },
    {
      "epoch": 0.7229373502225266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2069,
      "step": 21117
    },
    {
      "epoch": 0.7229715850736049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2078,
      "step": 21118
    },
    {
      "epoch": 0.7230058199246834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.185,
      "step": 21119
    },
    {
      "epoch": 0.7230400547757617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1989,
      "step": 21120
    },
    {
      "epoch": 0.7230742896268402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2062,
      "step": 21121
    },
    {
      "epoch": 0.7231085244779185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2264,
      "step": 21122
    },
    {
      "epoch": 0.723142759328997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2049,
      "step": 21123
    },
    {
      "epoch": 0.7231769941800753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1831,
      "step": 21124
    },
    {
      "epoch": 0.7232112290311538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1877,
      "step": 21125
    },
    {
      "epoch": 0.7232454638822321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1914,
      "step": 21126
    },
    {
      "epoch": 0.7232796987333105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2021,
      "step": 21127
    },
    {
      "epoch": 0.7233139335843889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2004,
      "step": 21128
    },
    {
      "epoch": 0.7233481684354673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1924,
      "step": 21129
    },
    {
      "epoch": 0.7233824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2215,
      "step": 21130
    },
    {
      "epoch": 0.7234166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2067,
      "step": 21131
    },
    {
      "epoch": 0.7234508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1898,
      "step": 21132
    },
    {
      "epoch": 0.7234851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1917,
      "step": 21133
    },
    {
      "epoch": 0.7235193426908593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1922,
      "step": 21134
    },
    {
      "epoch": 0.7235535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1883,
      "step": 21135
    },
    {
      "epoch": 0.7235878123930161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1816,
      "step": 21136
    },
    {
      "epoch": 0.7236220472440945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1998,
      "step": 21137
    },
    {
      "epoch": 0.7236562820951729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2032,
      "step": 21138
    },
    {
      "epoch": 0.7236905169462513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1893,
      "step": 21139
    },
    {
      "epoch": 0.7237247517973296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1914,
      "step": 21140
    },
    {
      "epoch": 0.7237589866484081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.194,
      "step": 21141
    },
    {
      "epoch": 0.7237932214994864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2194,
      "step": 21142
    },
    {
      "epoch": 0.7238274563505649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1955,
      "step": 21143
    },
    {
      "epoch": 0.7238616912016432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1887,
      "step": 21144
    },
    {
      "epoch": 0.7238959260527217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2056,
      "step": 21145
    },
    {
      "epoch": 0.7239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1938,
      "step": 21146
    },
    {
      "epoch": 0.7239643957548785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1833,
      "step": 21147
    },
    {
      "epoch": 0.7239986306059568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2142,
      "step": 21148
    },
    {
      "epoch": 0.7240328654570353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2059,
      "step": 21149
    },
    {
      "epoch": 0.7240671003081136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1852,
      "step": 21150
    },
    {
      "epoch": 0.7241013351591921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2143,
      "step": 21151
    },
    {
      "epoch": 0.7241355700102704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1939,
      "step": 21152
    },
    {
      "epoch": 0.7241698048613489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2,
      "step": 21153
    },
    {
      "epoch": 0.7242040397124273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.191,
      "step": 21154
    },
    {
      "epoch": 0.7242382745635056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2105,
      "step": 21155
    },
    {
      "epoch": 0.7242725094145841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1976,
      "step": 21156
    },
    {
      "epoch": 0.7243067442656624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1824,
      "step": 21157
    },
    {
      "epoch": 0.7243409791167409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1954,
      "step": 21158
    },
    {
      "epoch": 0.7243752139678192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2085,
      "step": 21159
    },
    {
      "epoch": 0.7244094488188977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1928,
      "step": 21160
    },
    {
      "epoch": 0.724443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2034,
      "step": 21161
    },
    {
      "epoch": 0.7244779185210545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1963,
      "step": 21162
    },
    {
      "epoch": 0.7245121533721328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2103,
      "step": 21163
    },
    {
      "epoch": 0.7245463882232113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1915,
      "step": 21164
    },
    {
      "epoch": 0.7245806230742896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1749,
      "step": 21165
    },
    {
      "epoch": 0.7246148579253681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2169,
      "step": 21166
    },
    {
      "epoch": 0.7246490927764464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1887,
      "step": 21167
    },
    {
      "epoch": 0.7246833276275249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1949,
      "step": 21168
    },
    {
      "epoch": 0.7247175624786032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1915,
      "step": 21169
    },
    {
      "epoch": 0.7247517973296816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1964,
      "step": 21170
    },
    {
      "epoch": 0.72478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1953,
      "step": 21171
    },
    {
      "epoch": 0.7248202670318384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1922,
      "step": 21172
    },
    {
      "epoch": 0.7248545018829168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2072,
      "step": 21173
    },
    {
      "epoch": 0.7248887367339952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2051,
      "step": 21174
    },
    {
      "epoch": 0.7249229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1777,
      "step": 21175
    },
    {
      "epoch": 0.724957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1893,
      "step": 21176
    },
    {
      "epoch": 0.7249914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1949,
      "step": 21177
    },
    {
      "epoch": 0.7250256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1948,
      "step": 21178
    },
    {
      "epoch": 0.7250599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2136,
      "step": 21179
    },
    {
      "epoch": 0.7250941458404656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2006,
      "step": 21180
    },
    {
      "epoch": 0.725128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2101,
      "step": 21181
    },
    {
      "epoch": 0.7251626155426224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2037,
      "step": 21182
    },
    {
      "epoch": 0.7251968503937007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2156,
      "step": 21183
    },
    {
      "epoch": 0.7252310852447792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.198,
      "step": 21184
    },
    {
      "epoch": 0.7252653200958575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1847,
      "step": 21185
    },
    {
      "epoch": 0.725299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1954,
      "step": 21186
    },
    {
      "epoch": 0.7253337897980143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1888,
      "step": 21187
    },
    {
      "epoch": 0.7253680246490928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2046,
      "step": 21188
    },
    {
      "epoch": 0.7254022595001711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1809,
      "step": 21189
    },
    {
      "epoch": 0.7254364943512496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1866,
      "step": 21190
    },
    {
      "epoch": 0.725470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1918,
      "step": 21191
    },
    {
      "epoch": 0.7255049640534064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1862,
      "step": 21192
    },
    {
      "epoch": 0.7255391989044848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1931,
      "step": 21193
    },
    {
      "epoch": 0.7255734337555632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2042,
      "step": 21194
    },
    {
      "epoch": 0.7256076686066416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1927,
      "step": 21195
    },
    {
      "epoch": 0.72564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1879,
      "step": 21196
    },
    {
      "epoch": 0.7256761383087984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1916,
      "step": 21197
    },
    {
      "epoch": 0.7257103731598767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1996,
      "step": 21198
    },
    {
      "epoch": 0.7257446080109552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1975,
      "step": 21199
    },
    {
      "epoch": 0.7257788428620335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1947,
      "step": 21200
    },
    {
      "epoch": 0.725813077713112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1724,
      "step": 21201
    },
    {
      "epoch": 0.7258473125641903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2111,
      "step": 21202
    },
    {
      "epoch": 0.7258815474152688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2081,
      "step": 21203
    },
    {
      "epoch": 0.7259157822663471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1843,
      "step": 21204
    },
    {
      "epoch": 0.7259500171174256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2015,
      "step": 21205
    },
    {
      "epoch": 0.7259842519685039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2074,
      "step": 21206
    },
    {
      "epoch": 0.7260184868195824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2019,
      "step": 21207
    },
    {
      "epoch": 0.7260527216706607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1813,
      "step": 21208
    },
    {
      "epoch": 0.7260869565217392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2087,
      "step": 21209
    },
    {
      "epoch": 0.7261211913728175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1991,
      "step": 21210
    },
    {
      "epoch": 0.726155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1903,
      "step": 21211
    },
    {
      "epoch": 0.7261896610749743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1934,
      "step": 21212
    },
    {
      "epoch": 0.7262238959260527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.198,
      "step": 21213
    },
    {
      "epoch": 0.7262581307771311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1877,
      "step": 21214
    },
    {
      "epoch": 0.7262923656282095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1931,
      "step": 21215
    },
    {
      "epoch": 0.7263266004792879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2021,
      "step": 21216
    },
    {
      "epoch": 0.7263608353303663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2039,
      "step": 21217
    },
    {
      "epoch": 0.7263950701814447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1797,
      "step": 21218
    },
    {
      "epoch": 0.7264293050325231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1866,
      "step": 21219
    },
    {
      "epoch": 0.7264635398836015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1926,
      "step": 21220
    },
    {
      "epoch": 0.7264977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2138,
      "step": 21221
    },
    {
      "epoch": 0.7265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2037,
      "step": 21222
    },
    {
      "epoch": 0.7265662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2064,
      "step": 21223
    },
    {
      "epoch": 0.726600479287915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1869,
      "step": 21224
    },
    {
      "epoch": 0.7266347141389935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.194,
      "step": 21225
    },
    {
      "epoch": 0.7266689489900718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1964,
      "step": 21226
    },
    {
      "epoch": 0.7267031838411503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.186,
      "step": 21227
    },
    {
      "epoch": 0.7267374186922287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1984,
      "step": 21228
    },
    {
      "epoch": 0.7267716535433071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1981,
      "step": 21229
    },
    {
      "epoch": 0.7268058883943855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1916,
      "step": 21230
    },
    {
      "epoch": 0.7268401232454639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2043,
      "step": 21231
    },
    {
      "epoch": 0.7268743580965423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2094,
      "step": 21232
    },
    {
      "epoch": 0.7269085929476207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.195,
      "step": 21233
    },
    {
      "epoch": 0.7269428277986991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1897,
      "step": 21234
    },
    {
      "epoch": 0.7269770626497775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2102,
      "step": 21235
    },
    {
      "epoch": 0.7270112975008559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1946,
      "step": 21236
    },
    {
      "epoch": 0.7270455323519343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1995,
      "step": 21237
    },
    {
      "epoch": 0.7270797672030127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.194,
      "step": 21238
    },
    {
      "epoch": 0.727114002054091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1864,
      "step": 21239
    },
    {
      "epoch": 0.7271482369051695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1944,
      "step": 21240
    },
    {
      "epoch": 0.7271824717562478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.194,
      "step": 21241
    },
    {
      "epoch": 0.7272167066073263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2066,
      "step": 21242
    },
    {
      "epoch": 0.7272509414584046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1906,
      "step": 21243
    },
    {
      "epoch": 0.7272851763094831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1963,
      "step": 21244
    },
    {
      "epoch": 0.7273194111605614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1795,
      "step": 21245
    },
    {
      "epoch": 0.7273536460116399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1957,
      "step": 21246
    },
    {
      "epoch": 0.7273878808627182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1931,
      "step": 21247
    },
    {
      "epoch": 0.7274221157137967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2,
      "step": 21248
    },
    {
      "epoch": 0.727456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1927,
      "step": 21249
    },
    {
      "epoch": 0.7274905854159535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1937,
      "step": 21250
    },
    {
      "epoch": 0.7275248202670318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2006,
      "step": 21251
    },
    {
      "epoch": 0.7275590551181103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1726,
      "step": 21252
    },
    {
      "epoch": 0.7275932899691886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2114,
      "step": 21253
    },
    {
      "epoch": 0.727627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1837,
      "step": 21254
    },
    {
      "epoch": 0.7276617596713454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1938,
      "step": 21255
    },
    {
      "epoch": 0.7276959945224238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1829,
      "step": 21256
    },
    {
      "epoch": 0.7277302293735022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1934,
      "step": 21257
    },
    {
      "epoch": 0.7277644642245806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1878,
      "step": 21258
    },
    {
      "epoch": 0.727798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1994,
      "step": 21259
    },
    {
      "epoch": 0.7278329339267374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1956,
      "step": 21260
    },
    {
      "epoch": 0.7278671687778158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1801,
      "step": 21261
    },
    {
      "epoch": 0.7279014036288942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1903,
      "step": 21262
    },
    {
      "epoch": 0.7279356384799726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2043,
      "step": 21263
    },
    {
      "epoch": 0.727969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1877,
      "step": 21264
    },
    {
      "epoch": 0.7280041081821295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1889,
      "step": 21265
    },
    {
      "epoch": 0.7280383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1899,
      "step": 21266
    },
    {
      "epoch": 0.7280725778842863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1827,
      "step": 21267
    },
    {
      "epoch": 0.7281068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.197,
      "step": 21268
    },
    {
      "epoch": 0.728141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.205,
      "step": 21269
    },
    {
      "epoch": 0.7281752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.189,
      "step": 21270
    },
    {
      "epoch": 0.7282095172885998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2036,
      "step": 21271
    },
    {
      "epoch": 0.7282437521396782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1899,
      "step": 21272
    },
    {
      "epoch": 0.7282779869907566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1988,
      "step": 21273
    },
    {
      "epoch": 0.728312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1838,
      "step": 21274
    },
    {
      "epoch": 0.7283464566929134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1796,
      "step": 21275
    },
    {
      "epoch": 0.7283806915439918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1884,
      "step": 21276
    },
    {
      "epoch": 0.7284149263950702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1847,
      "step": 21277
    },
    {
      "epoch": 0.7284491612461486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1819,
      "step": 21278
    },
    {
      "epoch": 0.728483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1825,
      "step": 21279
    },
    {
      "epoch": 0.7285176309483054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1967,
      "step": 21280
    },
    {
      "epoch": 0.7285518657993838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1976,
      "step": 21281
    },
    {
      "epoch": 0.7285861006504621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1805,
      "step": 21282
    },
    {
      "epoch": 0.7286203355015406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1903,
      "step": 21283
    },
    {
      "epoch": 0.7286545703526189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1913,
      "step": 21284
    },
    {
      "epoch": 0.7286888052036974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1982,
      "step": 21285
    },
    {
      "epoch": 0.7287230400547757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1711,
      "step": 21286
    },
    {
      "epoch": 0.7287572749058542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1935,
      "step": 21287
    },
    {
      "epoch": 0.7287915097569325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2163,
      "step": 21288
    },
    {
      "epoch": 0.728825744608011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2011,
      "step": 21289
    },
    {
      "epoch": 0.7288599794590893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1876,
      "step": 21290
    },
    {
      "epoch": 0.7288942143101678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1918,
      "step": 21291
    },
    {
      "epoch": 0.7289284491612461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1915,
      "step": 21292
    },
    {
      "epoch": 0.7289626840123246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2094,
      "step": 21293
    },
    {
      "epoch": 0.7289969188634029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.201,
      "step": 21294
    },
    {
      "epoch": 0.7290311537144814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.207,
      "step": 21295
    },
    {
      "epoch": 0.7290653885655597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2081,
      "step": 21296
    },
    {
      "epoch": 0.7290996234166381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1878,
      "step": 21297
    },
    {
      "epoch": 0.7291338582677165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2006,
      "step": 21298
    },
    {
      "epoch": 0.7291680931187949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2138,
      "step": 21299
    },
    {
      "epoch": 0.7292023279698733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1715,
      "step": 21300
    },
    {
      "epoch": 0.7292365628209517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1953,
      "step": 21301
    },
    {
      "epoch": 0.7292707976720302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2053,
      "step": 21302
    },
    {
      "epoch": 0.7293050325231085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1841,
      "step": 21303
    },
    {
      "epoch": 0.729339267374187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1822,
      "step": 21304
    },
    {
      "epoch": 0.7293735022252653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1963,
      "step": 21305
    },
    {
      "epoch": 0.7294077370763438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1978,
      "step": 21306
    },
    {
      "epoch": 0.7294419719274221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1985,
      "step": 21307
    },
    {
      "epoch": 0.7294762067785006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1781,
      "step": 21308
    },
    {
      "epoch": 0.7295104416295789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1997,
      "step": 21309
    },
    {
      "epoch": 0.7295446764806574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1995,
      "step": 21310
    },
    {
      "epoch": 0.7295789113317357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1732,
      "step": 21311
    },
    {
      "epoch": 0.7296131461828141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1826,
      "step": 21312
    },
    {
      "epoch": 0.7296473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1875,
      "step": 21313
    },
    {
      "epoch": 0.7296816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1952,
      "step": 21314
    },
    {
      "epoch": 0.7297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1809,
      "step": 21315
    },
    {
      "epoch": 0.7297500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.198,
      "step": 21316
    },
    {
      "epoch": 0.7297843204382061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1617,
      "step": 21317
    },
    {
      "epoch": 0.7298185552892845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1821,
      "step": 21318
    },
    {
      "epoch": 0.7298527901403629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1982,
      "step": 21319
    },
    {
      "epoch": 0.7298870249914413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.186,
      "step": 21320
    },
    {
      "epoch": 0.7299212598425197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1974,
      "step": 21321
    },
    {
      "epoch": 0.7299554946935981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1839,
      "step": 21322
    },
    {
      "epoch": 0.7299897295446764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.179,
      "step": 21323
    },
    {
      "epoch": 0.7300239643957549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1936,
      "step": 21324
    },
    {
      "epoch": 0.7300581992468332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1964,
      "step": 21325
    },
    {
      "epoch": 0.7300924340979117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1946,
      "step": 21326
    },
    {
      "epoch": 0.73012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1968,
      "step": 21327
    },
    {
      "epoch": 0.7301609038000685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1905,
      "step": 21328
    },
    {
      "epoch": 0.7301951386511468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2145,
      "step": 21329
    },
    {
      "epoch": 0.7302293735022253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1841,
      "step": 21330
    },
    {
      "epoch": 0.7302636083533036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2128,
      "step": 21331
    },
    {
      "epoch": 0.7302978432043821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2027,
      "step": 21332
    },
    {
      "epoch": 0.7303320780554604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1897,
      "step": 21333
    },
    {
      "epoch": 0.7303663129065389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1903,
      "step": 21334
    },
    {
      "epoch": 0.7304005477576172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2057,
      "step": 21335
    },
    {
      "epoch": 0.7304347826086957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.186,
      "step": 21336
    },
    {
      "epoch": 0.730469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1831,
      "step": 21337
    },
    {
      "epoch": 0.7305032523108524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1993,
      "step": 21338
    },
    {
      "epoch": 0.7305374871619309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1907,
      "step": 21339
    },
    {
      "epoch": 0.7305717220130092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2023,
      "step": 21340
    },
    {
      "epoch": 0.7306059568640877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2048,
      "step": 21341
    },
    {
      "epoch": 0.730640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1899,
      "step": 21342
    },
    {
      "epoch": 0.7306744265662445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.212,
      "step": 21343
    },
    {
      "epoch": 0.7307086614173228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1972,
      "step": 21344
    },
    {
      "epoch": 0.7307428962684013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1836,
      "step": 21345
    },
    {
      "epoch": 0.7307771311194796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2072,
      "step": 21346
    },
    {
      "epoch": 0.7308113659705581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2232,
      "step": 21347
    },
    {
      "epoch": 0.7308456008216364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.194,
      "step": 21348
    },
    {
      "epoch": 0.7308798356727149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2202,
      "step": 21349
    },
    {
      "epoch": 0.7309140705237932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1758,
      "step": 21350
    },
    {
      "epoch": 0.7309483053748717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1823,
      "step": 21351
    },
    {
      "epoch": 0.73098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1816,
      "step": 21352
    },
    {
      "epoch": 0.7310167750770284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2035,
      "step": 21353
    },
    {
      "epoch": 0.7310510099281068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1857,
      "step": 21354
    },
    {
      "epoch": 0.7310852447791852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2034,
      "step": 21355
    },
    {
      "epoch": 0.7311194796302636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2188,
      "step": 21356
    },
    {
      "epoch": 0.731153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1947,
      "step": 21357
    },
    {
      "epoch": 0.7311879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1754,
      "step": 21358
    },
    {
      "epoch": 0.7312221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1921,
      "step": 21359
    },
    {
      "epoch": 0.7312564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1903,
      "step": 21360
    },
    {
      "epoch": 0.7312906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1996,
      "step": 21361
    },
    {
      "epoch": 0.731324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1842,
      "step": 21362
    },
    {
      "epoch": 0.7313591235878124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1752,
      "step": 21363
    },
    {
      "epoch": 0.7313933584388908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1767,
      "step": 21364
    },
    {
      "epoch": 0.7314275932899692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2016,
      "step": 21365
    },
    {
      "epoch": 0.7314618281410475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1822,
      "step": 21366
    },
    {
      "epoch": 0.731496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.194,
      "step": 21367
    },
    {
      "epoch": 0.7315302978432043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1776,
      "step": 21368
    },
    {
      "epoch": 0.7315645326942828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1833,
      "step": 21369
    },
    {
      "epoch": 0.7315987675453611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1891,
      "step": 21370
    },
    {
      "epoch": 0.7316330023964396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1803,
      "step": 21371
    },
    {
      "epoch": 0.7316672372475179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1927,
      "step": 21372
    },
    {
      "epoch": 0.7317014720985964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1906,
      "step": 21373
    },
    {
      "epoch": 0.7317357069496747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1919,
      "step": 21374
    },
    {
      "epoch": 0.7317699418007532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1911,
      "step": 21375
    },
    {
      "epoch": 0.7318041766518316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1862,
      "step": 21376
    },
    {
      "epoch": 0.73183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1925,
      "step": 21377
    },
    {
      "epoch": 0.7318726463539884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1775,
      "step": 21378
    },
    {
      "epoch": 0.7319068812050668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1834,
      "step": 21379
    },
    {
      "epoch": 0.7319411160561452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1787,
      "step": 21380
    },
    {
      "epoch": 0.7319753509072235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.191,
      "step": 21381
    },
    {
      "epoch": 0.732009585758302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1719,
      "step": 21382
    },
    {
      "epoch": 0.7320438206093803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1815,
      "step": 21383
    },
    {
      "epoch": 0.7320780554604588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1892,
      "step": 21384
    },
    {
      "epoch": 0.7321122903115371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1841,
      "step": 21385
    },
    {
      "epoch": 0.7321465251626156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1826,
      "step": 21386
    },
    {
      "epoch": 0.7321807600136939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.184,
      "step": 21387
    },
    {
      "epoch": 0.7322149948647724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1968,
      "step": 21388
    },
    {
      "epoch": 0.7322492297158507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1813,
      "step": 21389
    },
    {
      "epoch": 0.7322834645669292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1869,
      "step": 21390
    },
    {
      "epoch": 0.7323176994180075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2069,
      "step": 21391
    },
    {
      "epoch": 0.732351934269086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1962,
      "step": 21392
    },
    {
      "epoch": 0.7323861691201643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1938,
      "step": 21393
    },
    {
      "epoch": 0.7324204039712428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1835,
      "step": 21394
    },
    {
      "epoch": 0.7324546388223211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1816,
      "step": 21395
    },
    {
      "epoch": 0.7324888736733995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1875,
      "step": 21396
    },
    {
      "epoch": 0.7325231085244779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1895,
      "step": 21397
    },
    {
      "epoch": 0.7325573433755563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1737,
      "step": 21398
    },
    {
      "epoch": 0.7325915782266347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1962,
      "step": 21399
    },
    {
      "epoch": 0.7326258130777131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1718,
      "step": 21400
    },
    {
      "epoch": 0.7326600479287915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1802,
      "step": 21401
    },
    {
      "epoch": 0.7326942827798699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1888,
      "step": 21402
    },
    {
      "epoch": 0.7327285176309483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2077,
      "step": 21403
    },
    {
      "epoch": 0.7327627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1832,
      "step": 21404
    },
    {
      "epoch": 0.732796987333105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1792,
      "step": 21405
    },
    {
      "epoch": 0.7328312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2116,
      "step": 21406
    },
    {
      "epoch": 0.7328654570352618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2011,
      "step": 21407
    },
    {
      "epoch": 0.7328996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1892,
      "step": 21408
    },
    {
      "epoch": 0.7329339267374186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1843,
      "step": 21409
    },
    {
      "epoch": 0.7329681615884971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1808,
      "step": 21410
    },
    {
      "epoch": 0.7330023964395755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1614,
      "step": 21411
    },
    {
      "epoch": 0.7330366312906539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2099,
      "step": 21412
    },
    {
      "epoch": 0.7330708661417323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1916,
      "step": 21413
    },
    {
      "epoch": 0.7331051009928107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1992,
      "step": 21414
    },
    {
      "epoch": 0.7331393358438891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1598,
      "step": 21415
    },
    {
      "epoch": 0.7331735706949675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.168,
      "step": 21416
    },
    {
      "epoch": 0.7332078055460459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1862,
      "step": 21417
    },
    {
      "epoch": 0.7332420403971243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1852,
      "step": 21418
    },
    {
      "epoch": 0.7332762752482027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2062,
      "step": 21419
    },
    {
      "epoch": 0.733310510099281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1897,
      "step": 21420
    },
    {
      "epoch": 0.7333447449503595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2021,
      "step": 21421
    },
    {
      "epoch": 0.7333789798014378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1904,
      "step": 21422
    },
    {
      "epoch": 0.7334132146525163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2074,
      "step": 21423
    },
    {
      "epoch": 0.7334474495035946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1892,
      "step": 21424
    },
    {
      "epoch": 0.7334816843546731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2014,
      "step": 21425
    },
    {
      "epoch": 0.7335159192057514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.182,
      "step": 21426
    },
    {
      "epoch": 0.7335501540568299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1923,
      "step": 21427
    },
    {
      "epoch": 0.7335843889079082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1902,
      "step": 21428
    },
    {
      "epoch": 0.7336186237589867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1912,
      "step": 21429
    },
    {
      "epoch": 0.733652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1908,
      "step": 21430
    },
    {
      "epoch": 0.7336870934611435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2048,
      "step": 21431
    },
    {
      "epoch": 0.7337213283122218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1815,
      "step": 21432
    },
    {
      "epoch": 0.7337555631633003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1898,
      "step": 21433
    },
    {
      "epoch": 0.7337897980143786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1988,
      "step": 21434
    },
    {
      "epoch": 0.733824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1818,
      "step": 21435
    },
    {
      "epoch": 0.7338582677165354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1856,
      "step": 21436
    },
    {
      "epoch": 0.7338925025676138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1945,
      "step": 21437
    },
    {
      "epoch": 0.7339267374186922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1929,
      "step": 21438
    },
    {
      "epoch": 0.7339609722697706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2136,
      "step": 21439
    },
    {
      "epoch": 0.733995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1779,
      "step": 21440
    },
    {
      "epoch": 0.7340294419719274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1821,
      "step": 21441
    },
    {
      "epoch": 0.7340636768230058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1841,
      "step": 21442
    },
    {
      "epoch": 0.7340979116740842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1827,
      "step": 21443
    },
    {
      "epoch": 0.7341321465251626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1969,
      "step": 21444
    },
    {
      "epoch": 0.734166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1772,
      "step": 21445
    },
    {
      "epoch": 0.7342006162273194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1822,
      "step": 21446
    },
    {
      "epoch": 0.7342348510783978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.175,
      "step": 21447
    },
    {
      "epoch": 0.7342690859294763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1849,
      "step": 21448
    },
    {
      "epoch": 0.7343033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1699,
      "step": 21449
    },
    {
      "epoch": 0.734337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1818,
      "step": 21450
    },
    {
      "epoch": 0.7343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1861,
      "step": 21451
    },
    {
      "epoch": 0.7344060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1948,
      "step": 21452
    },
    {
      "epoch": 0.7344402601848682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.184,
      "step": 21453
    },
    {
      "epoch": 0.7344744950359466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2024,
      "step": 21454
    },
    {
      "epoch": 0.734508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1958,
      "step": 21455
    },
    {
      "epoch": 0.7345429647381034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1836,
      "step": 21456
    },
    {
      "epoch": 0.7345771995891818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1822,
      "step": 21457
    },
    {
      "epoch": 0.7346114344402602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1795,
      "step": 21458
    },
    {
      "epoch": 0.7346456692913386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2007,
      "step": 21459
    },
    {
      "epoch": 0.734679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.218,
      "step": 21460
    },
    {
      "epoch": 0.7347141389934954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1921,
      "step": 21461
    },
    {
      "epoch": 0.7347483738445738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1929,
      "step": 21462
    },
    {
      "epoch": 0.7347826086956522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2043,
      "step": 21463
    },
    {
      "epoch": 0.7348168435467306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1911,
      "step": 21464
    },
    {
      "epoch": 0.7348510783978089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1778,
      "step": 21465
    },
    {
      "epoch": 0.7348853132488874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1716,
      "step": 21466
    },
    {
      "epoch": 0.7349195480999657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1954,
      "step": 21467
    },
    {
      "epoch": 0.7349537829510442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1934,
      "step": 21468
    },
    {
      "epoch": 0.7349880178021225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1797,
      "step": 21469
    },
    {
      "epoch": 0.735022252653201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1911,
      "step": 21470
    },
    {
      "epoch": 0.7350564875042793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1837,
      "step": 21471
    },
    {
      "epoch": 0.7350907223553578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1791,
      "step": 21472
    },
    {
      "epoch": 0.7351249572064361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1786,
      "step": 21473
    },
    {
      "epoch": 0.7351591920575146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1883,
      "step": 21474
    },
    {
      "epoch": 0.7351934269085929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1863,
      "step": 21475
    },
    {
      "epoch": 0.7352276617596714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1739,
      "step": 21476
    },
    {
      "epoch": 0.7352618966107497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2149,
      "step": 21477
    },
    {
      "epoch": 0.7352961314618282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1818,
      "step": 21478
    },
    {
      "epoch": 0.7353303663129065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.194,
      "step": 21479
    },
    {
      "epoch": 0.7353646011639849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1919,
      "step": 21480
    },
    {
      "epoch": 0.7353988360150633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1758,
      "step": 21481
    },
    {
      "epoch": 0.7354330708661417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1969,
      "step": 21482
    },
    {
      "epoch": 0.7354673057172201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1933,
      "step": 21483
    },
    {
      "epoch": 0.7355015405682985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1827,
      "step": 21484
    },
    {
      "epoch": 0.735535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1899,
      "step": 21485
    },
    {
      "epoch": 0.7355700102704553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1736,
      "step": 21486
    },
    {
      "epoch": 0.7356042451215338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1869,
      "step": 21487
    },
    {
      "epoch": 0.7356384799726121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2059,
      "step": 21488
    },
    {
      "epoch": 0.7356727148236906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1786,
      "step": 21489
    },
    {
      "epoch": 0.7357069496747689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1808,
      "step": 21490
    },
    {
      "epoch": 0.7357411845258474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1822,
      "step": 21491
    },
    {
      "epoch": 0.7357754193769257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1812,
      "step": 21492
    },
    {
      "epoch": 0.7358096542280042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1653,
      "step": 21493
    },
    {
      "epoch": 0.7358438890790825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1753,
      "step": 21494
    },
    {
      "epoch": 0.7358781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.166,
      "step": 21495
    },
    {
      "epoch": 0.7359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1742,
      "step": 21496
    },
    {
      "epoch": 0.7359465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2132,
      "step": 21497
    },
    {
      "epoch": 0.7359808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1916,
      "step": 21498
    },
    {
      "epoch": 0.7360150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1871,
      "step": 21499
    },
    {
      "epoch": 0.7360492981855529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1947,
      "step": 21500
    },
    {
      "epoch": 0.7360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2029,
      "step": 21501
    },
    {
      "epoch": 0.7361177678877097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1971,
      "step": 21502
    },
    {
      "epoch": 0.7361520027387881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1905,
      "step": 21503
    },
    {
      "epoch": 0.7361862375898665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1762,
      "step": 21504
    },
    {
      "epoch": 0.7362204724409449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1916,
      "step": 21505
    },
    {
      "epoch": 0.7362547072920232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1947,
      "step": 21506
    },
    {
      "epoch": 0.7362889421431017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1897,
      "step": 21507
    },
    {
      "epoch": 0.73632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2014,
      "step": 21508
    },
    {
      "epoch": 0.7363574118452585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1913,
      "step": 21509
    },
    {
      "epoch": 0.7363916466963368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2108,
      "step": 21510
    },
    {
      "epoch": 0.7364258815474153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1904,
      "step": 21511
    },
    {
      "epoch": 0.7364601163984936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1895,
      "step": 21512
    },
    {
      "epoch": 0.7364943512495721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.174,
      "step": 21513
    },
    {
      "epoch": 0.7365285861006504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1993,
      "step": 21514
    },
    {
      "epoch": 0.7365628209517289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1973,
      "step": 21515
    },
    {
      "epoch": 0.7365970558028072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1879,
      "step": 21516
    },
    {
      "epoch": 0.7366312906538857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.183,
      "step": 21517
    },
    {
      "epoch": 0.736665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1988,
      "step": 21518
    },
    {
      "epoch": 0.7366997603560425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2082,
      "step": 21519
    },
    {
      "epoch": 0.7367339952071208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1722,
      "step": 21520
    },
    {
      "epoch": 0.7367682300581992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2013,
      "step": 21521
    },
    {
      "epoch": 0.7368024649092777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1811,
      "step": 21522
    },
    {
      "epoch": 0.736836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1939,
      "step": 21523
    },
    {
      "epoch": 0.7368709346114345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.193,
      "step": 21524
    },
    {
      "epoch": 0.7369051694625128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1835,
      "step": 21525
    },
    {
      "epoch": 0.7369394043135913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1782,
      "step": 21526
    },
    {
      "epoch": 0.7369736391646696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1843,
      "step": 21527
    },
    {
      "epoch": 0.7370078740157481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1871,
      "step": 21528
    },
    {
      "epoch": 0.7370421088668264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1871,
      "step": 21529
    },
    {
      "epoch": 0.7370763437179049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1844,
      "step": 21530
    },
    {
      "epoch": 0.7371105785689832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1829,
      "step": 21531
    },
    {
      "epoch": 0.7371448134200617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1891,
      "step": 21532
    },
    {
      "epoch": 0.73717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.211,
      "step": 21533
    },
    {
      "epoch": 0.7372132831222185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.187,
      "step": 21534
    },
    {
      "epoch": 0.7372475179732968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1812,
      "step": 21535
    },
    {
      "epoch": 0.7372817528243752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1946,
      "step": 21536
    },
    {
      "epoch": 0.7373159876754536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1789,
      "step": 21537
    },
    {
      "epoch": 0.737350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1837,
      "step": 21538
    },
    {
      "epoch": 0.7373844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1934,
      "step": 21539
    },
    {
      "epoch": 0.7374186922286888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1851,
      "step": 21540
    },
    {
      "epoch": 0.7374529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1732,
      "step": 21541
    },
    {
      "epoch": 0.7374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1982,
      "step": 21542
    },
    {
      "epoch": 0.737521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1906,
      "step": 21543
    },
    {
      "epoch": 0.7375556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1858,
      "step": 21544
    },
    {
      "epoch": 0.7375898664840808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1958,
      "step": 21545
    },
    {
      "epoch": 0.7376241013351592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1778,
      "step": 21546
    },
    {
      "epoch": 0.7376583361862376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1996,
      "step": 21547
    },
    {
      "epoch": 0.737692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1886,
      "step": 21548
    },
    {
      "epoch": 0.7377268058883943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1837,
      "step": 21549
    },
    {
      "epoch": 0.7377610407394728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1834,
      "step": 21550
    },
    {
      "epoch": 0.7377952755905511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.192,
      "step": 21551
    },
    {
      "epoch": 0.7378295104416296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1719,
      "step": 21552
    },
    {
      "epoch": 0.7378637452927079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1882,
      "step": 21553
    },
    {
      "epoch": 0.7378979801437864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1723,
      "step": 21554
    },
    {
      "epoch": 0.7379322149948647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1777,
      "step": 21555
    },
    {
      "epoch": 0.7379664498459432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1923,
      "step": 21556
    },
    {
      "epoch": 0.7380006846970215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1976,
      "step": 21557
    },
    {
      "epoch": 0.7380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2011,
      "step": 21558
    },
    {
      "epoch": 0.7380691543991784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1844,
      "step": 21559
    },
    {
      "epoch": 0.7381033892502568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.181,
      "step": 21560
    },
    {
      "epoch": 0.7381376241013352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1797,
      "step": 21561
    },
    {
      "epoch": 0.7381718589524136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1796,
      "step": 21562
    },
    {
      "epoch": 0.738206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1871,
      "step": 21563
    },
    {
      "epoch": 0.7382403286545703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.196,
      "step": 21564
    },
    {
      "epoch": 0.7382745635056488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1939,
      "step": 21565
    },
    {
      "epoch": 0.7383087983567271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2061,
      "step": 21566
    },
    {
      "epoch": 0.7383430332078056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1917,
      "step": 21567
    },
    {
      "epoch": 0.7383772680588839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1797,
      "step": 21568
    },
    {
      "epoch": 0.7384115029099624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1944,
      "step": 21569
    },
    {
      "epoch": 0.7384457377610407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1846,
      "step": 21570
    },
    {
      "epoch": 0.7384799726121192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1855,
      "step": 21571
    },
    {
      "epoch": 0.7385142074631975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.171,
      "step": 21572
    },
    {
      "epoch": 0.738548442314276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1904,
      "step": 21573
    },
    {
      "epoch": 0.7385826771653543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1751,
      "step": 21574
    },
    {
      "epoch": 0.7386169120164328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1982,
      "step": 21575
    },
    {
      "epoch": 0.7386511468675111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1935,
      "step": 21576
    },
    {
      "epoch": 0.7386853817185896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2023,
      "step": 21577
    },
    {
      "epoch": 0.7387196165696679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1841,
      "step": 21578
    },
    {
      "epoch": 0.7387538514207463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1729,
      "step": 21579
    },
    {
      "epoch": 0.7387880862718247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1649,
      "step": 21580
    },
    {
      "epoch": 0.7388223211229031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1874,
      "step": 21581
    },
    {
      "epoch": 0.7388565559739815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.186,
      "step": 21582
    },
    {
      "epoch": 0.7388907908250599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1894,
      "step": 21583
    },
    {
      "epoch": 0.7389250256761383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1732,
      "step": 21584
    },
    {
      "epoch": 0.7389592605272167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1684,
      "step": 21585
    },
    {
      "epoch": 0.7389934953782951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1988,
      "step": 21586
    },
    {
      "epoch": 0.7390277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1679,
      "step": 21587
    },
    {
      "epoch": 0.7390619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.199,
      "step": 21588
    },
    {
      "epoch": 0.7390961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1915,
      "step": 21589
    },
    {
      "epoch": 0.7391304347826086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.187,
      "step": 21590
    },
    {
      "epoch": 0.7391646696336871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1885,
      "step": 21591
    },
    {
      "epoch": 0.7391989044847654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.202,
      "step": 21592
    },
    {
      "epoch": 0.7392331393358439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1782,
      "step": 21593
    },
    {
      "epoch": 0.7392673741869222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1956,
      "step": 21594
    },
    {
      "epoch": 0.7393016090380007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1879,
      "step": 21595
    },
    {
      "epoch": 0.7393358438890791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1967,
      "step": 21596
    },
    {
      "epoch": 0.7393700787401575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1971,
      "step": 21597
    },
    {
      "epoch": 0.7394043135912359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1846,
      "step": 21598
    },
    {
      "epoch": 0.7394385484423143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1997,
      "step": 21599
    },
    {
      "epoch": 0.7394727832933927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1952,
      "step": 21600
    },
    {
      "epoch": 0.7395070181444711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1743,
      "step": 21601
    },
    {
      "epoch": 0.7395412529955495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1938,
      "step": 21602
    },
    {
      "epoch": 0.7395754878466279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1913,
      "step": 21603
    },
    {
      "epoch": 0.7396097226977063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1732,
      "step": 21604
    },
    {
      "epoch": 0.7396439575487846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1941,
      "step": 21605
    },
    {
      "epoch": 0.7396781923998631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1818,
      "step": 21606
    },
    {
      "epoch": 0.7397124272509414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1854,
      "step": 21607
    },
    {
      "epoch": 0.7397466621020199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1969,
      "step": 21608
    },
    {
      "epoch": 0.7397808969530982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1748,
      "step": 21609
    },
    {
      "epoch": 0.7398151318041767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.196,
      "step": 21610
    },
    {
      "epoch": 0.739849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2016,
      "step": 21611
    },
    {
      "epoch": 0.7398836015063335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1817,
      "step": 21612
    },
    {
      "epoch": 0.7399178363574118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1864,
      "step": 21613
    },
    {
      "epoch": 0.7399520712084903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1962,
      "step": 21614
    },
    {
      "epoch": 0.7399863060595686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2026,
      "step": 21615
    },
    {
      "epoch": 0.7400205409106471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1846,
      "step": 21616
    },
    {
      "epoch": 0.7400547757617254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2031,
      "step": 21617
    },
    {
      "epoch": 0.7400890106128039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1857,
      "step": 21618
    },
    {
      "epoch": 0.7401232454638822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1796,
      "step": 21619
    },
    {
      "epoch": 0.7401574803149606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1958,
      "step": 21620
    },
    {
      "epoch": 0.740191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.197,
      "step": 21621
    },
    {
      "epoch": 0.7402259500171174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1688,
      "step": 21622
    },
    {
      "epoch": 0.7402601848681958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1821,
      "step": 21623
    },
    {
      "epoch": 0.7402944197192742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1746,
      "step": 21624
    },
    {
      "epoch": 0.7403286545703526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.186,
      "step": 21625
    },
    {
      "epoch": 0.740362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1755,
      "step": 21626
    },
    {
      "epoch": 0.7403971242725094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.193,
      "step": 21627
    },
    {
      "epoch": 0.7404313591235878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1854,
      "step": 21628
    },
    {
      "epoch": 0.7404655939746662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1888,
      "step": 21629
    },
    {
      "epoch": 0.7404998288257446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1788,
      "step": 21630
    },
    {
      "epoch": 0.740534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2067,
      "step": 21631
    },
    {
      "epoch": 0.7405682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1815,
      "step": 21632
    },
    {
      "epoch": 0.7406025333789799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1786,
      "step": 21633
    },
    {
      "epoch": 0.7406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1644,
      "step": 21634
    },
    {
      "epoch": 0.7406710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1751,
      "step": 21635
    },
    {
      "epoch": 0.740705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.197,
      "step": 21636
    },
    {
      "epoch": 0.7407394727832934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1835,
      "step": 21637
    },
    {
      "epoch": 0.7407737076343718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1816,
      "step": 21638
    },
    {
      "epoch": 0.7408079424854502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1914,
      "step": 21639
    },
    {
      "epoch": 0.7408421773365286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1743,
      "step": 21640
    },
    {
      "epoch": 0.740876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1808,
      "step": 21641
    },
    {
      "epoch": 0.7409106470386854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1823,
      "step": 21642
    },
    {
      "epoch": 0.7409448818897638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1752,
      "step": 21643
    },
    {
      "epoch": 0.7409791167408422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1868,
      "step": 21644
    },
    {
      "epoch": 0.7410133515919206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1816,
      "step": 21645
    },
    {
      "epoch": 0.741047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1915,
      "step": 21646
    },
    {
      "epoch": 0.7410818212940774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1926,
      "step": 21647
    },
    {
      "epoch": 0.7411160561451557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1712,
      "step": 21648
    },
    {
      "epoch": 0.7411502909962342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1774,
      "step": 21649
    },
    {
      "epoch": 0.7411845258473125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1935,
      "step": 21650
    },
    {
      "epoch": 0.741218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1993,
      "step": 21651
    },
    {
      "epoch": 0.7412529955494693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1752,
      "step": 21652
    },
    {
      "epoch": 0.7412872304005478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1938,
      "step": 21653
    },
    {
      "epoch": 0.7413214652516261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.174,
      "step": 21654
    },
    {
      "epoch": 0.7413557001027046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1936,
      "step": 21655
    },
    {
      "epoch": 0.7413899349537829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1688,
      "step": 21656
    },
    {
      "epoch": 0.7414241698048614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1906,
      "step": 21657
    },
    {
      "epoch": 0.7414584046559397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1853,
      "step": 21658
    },
    {
      "epoch": 0.7414926395070182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1889,
      "step": 21659
    },
    {
      "epoch": 0.7415268743580965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1807,
      "step": 21660
    },
    {
      "epoch": 0.741561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1795,
      "step": 21661
    },
    {
      "epoch": 0.7415953440602533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1738,
      "step": 21662
    },
    {
      "epoch": 0.7416295789113317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1979,
      "step": 21663
    },
    {
      "epoch": 0.7416638137624101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1881,
      "step": 21664
    },
    {
      "epoch": 0.7416980486134885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1846,
      "step": 21665
    },
    {
      "epoch": 0.7417322834645669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1732,
      "step": 21666
    },
    {
      "epoch": 0.7417665183156453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.179,
      "step": 21667
    },
    {
      "epoch": 0.7418007531667237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1725,
      "step": 21668
    },
    {
      "epoch": 0.7418349880178021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2076,
      "step": 21669
    },
    {
      "epoch": 0.7418692228688806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1765,
      "step": 21670
    },
    {
      "epoch": 0.7419034577199589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1971,
      "step": 21671
    },
    {
      "epoch": 0.7419376925710374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1944,
      "step": 21672
    },
    {
      "epoch": 0.7419719274221157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1865,
      "step": 21673
    },
    {
      "epoch": 0.7420061622731942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2087,
      "step": 21674
    },
    {
      "epoch": 0.7420403971242725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1948,
      "step": 21675
    },
    {
      "epoch": 0.742074631975351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2074,
      "step": 21676
    },
    {
      "epoch": 0.7421088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1791,
      "step": 21677
    },
    {
      "epoch": 0.7421431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1836,
      "step": 21678
    },
    {
      "epoch": 0.7421773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1897,
      "step": 21679
    },
    {
      "epoch": 0.7422115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1726,
      "step": 21680
    },
    {
      "epoch": 0.7422458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2031,
      "step": 21681
    },
    {
      "epoch": 0.7422800410818213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1695,
      "step": 21682
    },
    {
      "epoch": 0.7423142759328997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1863,
      "step": 21683
    },
    {
      "epoch": 0.7423485107839781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1729,
      "step": 21684
    },
    {
      "epoch": 0.7423827456350565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1828,
      "step": 21685
    },
    {
      "epoch": 0.7424169804861349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1845,
      "step": 21686
    },
    {
      "epoch": 0.7424512153372133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1826,
      "step": 21687
    },
    {
      "epoch": 0.7424854501882917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1888,
      "step": 21688
    },
    {
      "epoch": 0.74251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1907,
      "step": 21689
    },
    {
      "epoch": 0.7425539198904485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1817,
      "step": 21690
    },
    {
      "epoch": 0.7425881547415268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1943,
      "step": 21691
    },
    {
      "epoch": 0.7426223895926053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1731,
      "step": 21692
    },
    {
      "epoch": 0.7426566244436836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1825,
      "step": 21693
    },
    {
      "epoch": 0.7426908592947621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1804,
      "step": 21694
    },
    {
      "epoch": 0.7427250941458404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1596,
      "step": 21695
    },
    {
      "epoch": 0.7427593289969189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1739,
      "step": 21696
    },
    {
      "epoch": 0.7427935638479972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1729,
      "step": 21697
    },
    {
      "epoch": 0.7428277986990757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1954,
      "step": 21698
    },
    {
      "epoch": 0.742862033550154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.194,
      "step": 21699
    },
    {
      "epoch": 0.7428962684012325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1846,
      "step": 21700
    },
    {
      "epoch": 0.7429305032523108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.188,
      "step": 21701
    },
    {
      "epoch": 0.7429647381033893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.199,
      "step": 21702
    },
    {
      "epoch": 0.7429989729544676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1888,
      "step": 21703
    },
    {
      "epoch": 0.743033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1705,
      "step": 21704
    },
    {
      "epoch": 0.7430674426566244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1775,
      "step": 21705
    },
    {
      "epoch": 0.7431016775077028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1847,
      "step": 21706
    },
    {
      "epoch": 0.7431359123587813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1756,
      "step": 21707
    },
    {
      "epoch": 0.7431701472098596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1918,
      "step": 21708
    },
    {
      "epoch": 0.7432043820609381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.182,
      "step": 21709
    },
    {
      "epoch": 0.7432386169120164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1705,
      "step": 21710
    },
    {
      "epoch": 0.7432728517630949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2005,
      "step": 21711
    },
    {
      "epoch": 0.7433070866141732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1676,
      "step": 21712
    },
    {
      "epoch": 0.7433413214652517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1772,
      "step": 21713
    },
    {
      "epoch": 0.74337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1868,
      "step": 21714
    },
    {
      "epoch": 0.7434097911674085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1895,
      "step": 21715
    },
    {
      "epoch": 0.7434440260184868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1794,
      "step": 21716
    },
    {
      "epoch": 0.7434782608695653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1689,
      "step": 21717
    },
    {
      "epoch": 0.7435124957206436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1794,
      "step": 21718
    },
    {
      "epoch": 0.743546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1847,
      "step": 21719
    },
    {
      "epoch": 0.7435809654228004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1737,
      "step": 21720
    },
    {
      "epoch": 0.7436152002738788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1757,
      "step": 21721
    },
    {
      "epoch": 0.7436494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1821,
      "step": 21722
    },
    {
      "epoch": 0.7436836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1823,
      "step": 21723
    },
    {
      "epoch": 0.743717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1885,
      "step": 21724
    },
    {
      "epoch": 0.7437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1772,
      "step": 21725
    },
    {
      "epoch": 0.7437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1716,
      "step": 21726
    },
    {
      "epoch": 0.7438206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2134,
      "step": 21727
    },
    {
      "epoch": 0.7438548442314276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1747,
      "step": 21728
    },
    {
      "epoch": 0.743889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1849,
      "step": 21729
    },
    {
      "epoch": 0.7439233139335844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1989,
      "step": 21730
    },
    {
      "epoch": 0.7439575487846628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.177,
      "step": 21731
    },
    {
      "epoch": 0.7439917836357411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1875,
      "step": 21732
    },
    {
      "epoch": 0.7440260184868196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1781,
      "step": 21733
    },
    {
      "epoch": 0.7440602533378979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1757,
      "step": 21734
    },
    {
      "epoch": 0.7440944881889764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1702,
      "step": 21735
    },
    {
      "epoch": 0.7441287230400547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1772,
      "step": 21736
    },
    {
      "epoch": 0.7441629578911332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1893,
      "step": 21737
    },
    {
      "epoch": 0.7441971927422115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1861,
      "step": 21738
    },
    {
      "epoch": 0.74423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1775,
      "step": 21739
    },
    {
      "epoch": 0.7442656624443683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1924,
      "step": 21740
    },
    {
      "epoch": 0.7442998972954468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1891,
      "step": 21741
    },
    {
      "epoch": 0.7443341321465252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1838,
      "step": 21742
    },
    {
      "epoch": 0.7443683669976036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.18,
      "step": 21743
    },
    {
      "epoch": 0.744402601848682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1847,
      "step": 21744
    },
    {
      "epoch": 0.7444368366997604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2057,
      "step": 21745
    },
    {
      "epoch": 0.7444710715508388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.187,
      "step": 21746
    },
    {
      "epoch": 0.7445053064019171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1892,
      "step": 21747
    },
    {
      "epoch": 0.7445395412529956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2067,
      "step": 21748
    },
    {
      "epoch": 0.7445737761040739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1893,
      "step": 21749
    },
    {
      "epoch": 0.7446080109551524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1911,
      "step": 21750
    },
    {
      "epoch": 0.7446422458062307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1797,
      "step": 21751
    },
    {
      "epoch": 0.7446764806573092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1836,
      "step": 21752
    },
    {
      "epoch": 0.7447107155083875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1814,
      "step": 21753
    },
    {
      "epoch": 0.744744950359466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1695,
      "step": 21754
    },
    {
      "epoch": 0.7447791852105443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1781,
      "step": 21755
    },
    {
      "epoch": 0.7448134200616228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1937,
      "step": 21756
    },
    {
      "epoch": 0.7448476549127011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1805,
      "step": 21757
    },
    {
      "epoch": 0.7448818897637796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1883,
      "step": 21758
    },
    {
      "epoch": 0.7449161246148579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1816,
      "step": 21759
    },
    {
      "epoch": 0.7449503594659364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1827,
      "step": 21760
    },
    {
      "epoch": 0.7449845943170147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1953,
      "step": 21761
    },
    {
      "epoch": 0.7450188291680931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1711,
      "step": 21762
    },
    {
      "epoch": 0.7450530640191715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1933,
      "step": 21763
    },
    {
      "epoch": 0.7450872988702499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1707,
      "step": 21764
    },
    {
      "epoch": 0.7451215337213283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1893,
      "step": 21765
    },
    {
      "epoch": 0.7451557685724067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1997,
      "step": 21766
    },
    {
      "epoch": 0.7451900034234851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.198,
      "step": 21767
    },
    {
      "epoch": 0.7452242382745635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1917,
      "step": 21768
    },
    {
      "epoch": 0.7452584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.183,
      "step": 21769
    },
    {
      "epoch": 0.7452927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.185,
      "step": 21770
    },
    {
      "epoch": 0.7453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1646,
      "step": 21771
    },
    {
      "epoch": 0.7453611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1815,
      "step": 21772
    },
    {
      "epoch": 0.7453954125299554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1872,
      "step": 21773
    },
    {
      "epoch": 0.7454296473810339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1982,
      "step": 21774
    },
    {
      "epoch": 0.7454638822321122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1934,
      "step": 21775
    },
    {
      "epoch": 0.7454981170831907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1791,
      "step": 21776
    },
    {
      "epoch": 0.745532351934269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.179,
      "step": 21777
    },
    {
      "epoch": 0.7455665867853475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1969,
      "step": 21778
    },
    {
      "epoch": 0.7456008216364259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1656,
      "step": 21779
    },
    {
      "epoch": 0.7456350564875043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.191,
      "step": 21780
    },
    {
      "epoch": 0.7456692913385827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1716,
      "step": 21781
    },
    {
      "epoch": 0.7457035261896611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1946,
      "step": 21782
    },
    {
      "epoch": 0.7457377610407395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1824,
      "step": 21783
    },
    {
      "epoch": 0.7457719958918179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1914,
      "step": 21784
    },
    {
      "epoch": 0.7458062307428963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1818,
      "step": 21785
    },
    {
      "epoch": 0.7458404655939747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1729,
      "step": 21786
    },
    {
      "epoch": 0.7458747004450531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1822,
      "step": 21787
    },
    {
      "epoch": 0.7459089352961314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2049,
      "step": 21788
    },
    {
      "epoch": 0.7459431701472099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1801,
      "step": 21789
    },
    {
      "epoch": 0.7459774049982882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.185,
      "step": 21790
    },
    {
      "epoch": 0.7460116398493667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1876,
      "step": 21791
    },
    {
      "epoch": 0.746045874700445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1927,
      "step": 21792
    },
    {
      "epoch": 0.7460801095515235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1794,
      "step": 21793
    },
    {
      "epoch": 0.7461143444026018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1855,
      "step": 21794
    },
    {
      "epoch": 0.7461485792536803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1977,
      "step": 21795
    },
    {
      "epoch": 0.7461828141047586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1954,
      "step": 21796
    },
    {
      "epoch": 0.7462170489558371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1625,
      "step": 21797
    },
    {
      "epoch": 0.7462512838069154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1591,
      "step": 21798
    },
    {
      "epoch": 0.7462855186579939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1996,
      "step": 21799
    },
    {
      "epoch": 0.7463197535090722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1903,
      "step": 21800
    },
    {
      "epoch": 0.7463539883601507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1739,
      "step": 21801
    },
    {
      "epoch": 0.746388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1867,
      "step": 21802
    },
    {
      "epoch": 0.7464224580623074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.201,
      "step": 21803
    },
    {
      "epoch": 0.7464566929133858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1747,
      "step": 21804
    },
    {
      "epoch": 0.7464909277644642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1845,
      "step": 21805
    },
    {
      "epoch": 0.7465251626155426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1753,
      "step": 21806
    },
    {
      "epoch": 0.746559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1808,
      "step": 21807
    },
    {
      "epoch": 0.7465936323176994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1866,
      "step": 21808
    },
    {
      "epoch": 0.7466278671687778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1891,
      "step": 21809
    },
    {
      "epoch": 0.7466621020198562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1783,
      "step": 21810
    },
    {
      "epoch": 0.7466963368709346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1966,
      "step": 21811
    },
    {
      "epoch": 0.746730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1809,
      "step": 21812
    },
    {
      "epoch": 0.7467648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1924,
      "step": 21813
    },
    {
      "epoch": 0.7467990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.187,
      "step": 21814
    },
    {
      "epoch": 0.7468332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1867,
      "step": 21815
    },
    {
      "epoch": 0.7468675111263267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1868,
      "step": 21816
    },
    {
      "epoch": 0.746901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1924,
      "step": 21817
    },
    {
      "epoch": 0.7469359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1626,
      "step": 21818
    },
    {
      "epoch": 0.7469702156795618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1657,
      "step": 21819
    },
    {
      "epoch": 0.7470044505306402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1873,
      "step": 21820
    },
    {
      "epoch": 0.7470386853817186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1825,
      "step": 21821
    },
    {
      "epoch": 0.747072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1803,
      "step": 21822
    },
    {
      "epoch": 0.7471071550838754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1749,
      "step": 21823
    },
    {
      "epoch": 0.7471413899349538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1696,
      "step": 21824
    },
    {
      "epoch": 0.7471756247860322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.179,
      "step": 21825
    },
    {
      "epoch": 0.7472098596371106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1798,
      "step": 21826
    },
    {
      "epoch": 0.747244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2003,
      "step": 21827
    },
    {
      "epoch": 0.7472783293392674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1743,
      "step": 21828
    },
    {
      "epoch": 0.7473125641903458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1918,
      "step": 21829
    },
    {
      "epoch": 0.7473467990414242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.179,
      "step": 21830
    },
    {
      "epoch": 0.7473810338925025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1891,
      "step": 21831
    },
    {
      "epoch": 0.747415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1857,
      "step": 21832
    },
    {
      "epoch": 0.7474495035946593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1982,
      "step": 21833
    },
    {
      "epoch": 0.7474837384457378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1677,
      "step": 21834
    },
    {
      "epoch": 0.7475179732968161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1938,
      "step": 21835
    },
    {
      "epoch": 0.7475522081478946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1765,
      "step": 21836
    },
    {
      "epoch": 0.7475864429989729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1763,
      "step": 21837
    },
    {
      "epoch": 0.7476206778500514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1908,
      "step": 21838
    },
    {
      "epoch": 0.7476549127011297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1923,
      "step": 21839
    },
    {
      "epoch": 0.7476891475522082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1751,
      "step": 21840
    },
    {
      "epoch": 0.7477233824032865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1694,
      "step": 21841
    },
    {
      "epoch": 0.747757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1641,
      "step": 21842
    },
    {
      "epoch": 0.7477918521054433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1735,
      "step": 21843
    },
    {
      "epoch": 0.7478260869565218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1807,
      "step": 21844
    },
    {
      "epoch": 0.7478603218076001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1852,
      "step": 21845
    },
    {
      "epoch": 0.7478945566586785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1797,
      "step": 21846
    },
    {
      "epoch": 0.7479287915097569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1741,
      "step": 21847
    },
    {
      "epoch": 0.7479630263608353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1766,
      "step": 21848
    },
    {
      "epoch": 0.7479972612119137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1795,
      "step": 21849
    },
    {
      "epoch": 0.7480314960629921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1748,
      "step": 21850
    },
    {
      "epoch": 0.7480657309140705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1808,
      "step": 21851
    },
    {
      "epoch": 0.7480999657651489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1824,
      "step": 21852
    },
    {
      "epoch": 0.7481342006162274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1556,
      "step": 21853
    },
    {
      "epoch": 0.7481684354673057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1823,
      "step": 21854
    },
    {
      "epoch": 0.7482026703183842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1828,
      "step": 21855
    },
    {
      "epoch": 0.7482369051694625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1688,
      "step": 21856
    },
    {
      "epoch": 0.748271140020541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1843,
      "step": 21857
    },
    {
      "epoch": 0.7483053748716193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1576,
      "step": 21858
    },
    {
      "epoch": 0.7483396097226978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1907,
      "step": 21859
    },
    {
      "epoch": 0.7483738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1804,
      "step": 21860
    },
    {
      "epoch": 0.7484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1837,
      "step": 21861
    },
    {
      "epoch": 0.7484423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1688,
      "step": 21862
    },
    {
      "epoch": 0.7484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1928,
      "step": 21863
    },
    {
      "epoch": 0.7485107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1949,
      "step": 21864
    },
    {
      "epoch": 0.7485450188291681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1837,
      "step": 21865
    },
    {
      "epoch": 0.7485792536802465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.181,
      "step": 21866
    },
    {
      "epoch": 0.7486134885313249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1692,
      "step": 21867
    },
    {
      "epoch": 0.7486477233824033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1922,
      "step": 21868
    },
    {
      "epoch": 0.7486819582334817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1935,
      "step": 21869
    },
    {
      "epoch": 0.74871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.17,
      "step": 21870
    },
    {
      "epoch": 0.7487504279356385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1818,
      "step": 21871
    },
    {
      "epoch": 0.7487846627867168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1873,
      "step": 21872
    },
    {
      "epoch": 0.7488188976377953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1906,
      "step": 21873
    },
    {
      "epoch": 0.7488531324888736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1608,
      "step": 21874
    },
    {
      "epoch": 0.7488873673399521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1981,
      "step": 21875
    },
    {
      "epoch": 0.7489216021910304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1798,
      "step": 21876
    },
    {
      "epoch": 0.7489558370421089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1791,
      "step": 21877
    },
    {
      "epoch": 0.7489900718931872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1884,
      "step": 21878
    },
    {
      "epoch": 0.7490243067442657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1749,
      "step": 21879
    },
    {
      "epoch": 0.749058541595344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1855,
      "step": 21880
    },
    {
      "epoch": 0.7490927764464225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1941,
      "step": 21881
    },
    {
      "epoch": 0.7491270112975008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1912,
      "step": 21882
    },
    {
      "epoch": 0.7491612461485793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1818,
      "step": 21883
    },
    {
      "epoch": 0.7491954809996576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1682,
      "step": 21884
    },
    {
      "epoch": 0.749229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1736,
      "step": 21885
    },
    {
      "epoch": 0.7492639507018144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1758,
      "step": 21886
    },
    {
      "epoch": 0.7492981855528928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1813,
      "step": 21887
    },
    {
      "epoch": 0.7493324204039712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1717,
      "step": 21888
    },
    {
      "epoch": 0.7493666552550496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1846,
      "step": 21889
    },
    {
      "epoch": 0.7494008901061281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1924,
      "step": 21890
    },
    {
      "epoch": 0.7494351249572064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1813,
      "step": 21891
    },
    {
      "epoch": 0.7494693598082849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1862,
      "step": 21892
    },
    {
      "epoch": 0.7495035946593632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1883,
      "step": 21893
    },
    {
      "epoch": 0.7495378295104417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1827,
      "step": 21894
    },
    {
      "epoch": 0.74957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1638,
      "step": 21895
    },
    {
      "epoch": 0.7496062992125985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1858,
      "step": 21896
    },
    {
      "epoch": 0.7496405340636768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1746,
      "step": 21897
    },
    {
      "epoch": 0.7496747689147553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1834,
      "step": 21898
    },
    {
      "epoch": 0.7497090037658336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1806,
      "step": 21899
    },
    {
      "epoch": 0.749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1652,
      "step": 21900
    },
    {
      "epoch": 0.7497774734679904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1981,
      "step": 21901
    },
    {
      "epoch": 0.7498117083190688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.181,
      "step": 21902
    },
    {
      "epoch": 0.7498459431701472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1589,
      "step": 21903
    },
    {
      "epoch": 0.7498801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1768,
      "step": 21904
    },
    {
      "epoch": 0.749914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1665,
      "step": 21905
    },
    {
      "epoch": 0.7499486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1866,
      "step": 21906
    },
    {
      "epoch": 0.7499828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1944,
      "step": 21907
    },
    {
      "epoch": 0.7500171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1838,
      "step": 21908
    },
    {
      "epoch": 0.7500513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.178,
      "step": 21909
    },
    {
      "epoch": 0.750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1848,
      "step": 21910
    },
    {
      "epoch": 0.7501198219787744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1957,
      "step": 21911
    },
    {
      "epoch": 0.7501540568298528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1989,
      "step": 21912
    },
    {
      "epoch": 0.7501882916809312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2002,
      "step": 21913
    },
    {
      "epoch": 0.7502225265320096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1842,
      "step": 21914
    },
    {
      "epoch": 0.750256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1715,
      "step": 21915
    },
    {
      "epoch": 0.7502909962341664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1704,
      "step": 21916
    },
    {
      "epoch": 0.7503252310852447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1905,
      "step": 21917
    },
    {
      "epoch": 0.7503594659363232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1854,
      "step": 21918
    },
    {
      "epoch": 0.7503937007874015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1741,
      "step": 21919
    },
    {
      "epoch": 0.75042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1784,
      "step": 21920
    },
    {
      "epoch": 0.7504621704895583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1943,
      "step": 21921
    },
    {
      "epoch": 0.7504964053406368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.198,
      "step": 21922
    },
    {
      "epoch": 0.7505306401917151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1795,
      "step": 21923
    },
    {
      "epoch": 0.7505648750427936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1765,
      "step": 21924
    },
    {
      "epoch": 0.7505991098938719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1745,
      "step": 21925
    },
    {
      "epoch": 0.7506333447449504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1745,
      "step": 21926
    },
    {
      "epoch": 0.7506675795960288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1834,
      "step": 21927
    },
    {
      "epoch": 0.7507018144471072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.181,
      "step": 21928
    },
    {
      "epoch": 0.7507360492981856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1852,
      "step": 21929
    },
    {
      "epoch": 0.750770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1788,
      "step": 21930
    },
    {
      "epoch": 0.7508045190003424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1519,
      "step": 21931
    },
    {
      "epoch": 0.7508387538514207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1889,
      "step": 21932
    },
    {
      "epoch": 0.7508729887024992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1874,
      "step": 21933
    },
    {
      "epoch": 0.7509072235535775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.166,
      "step": 21934
    },
    {
      "epoch": 0.750941458404656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.167,
      "step": 21935
    },
    {
      "epoch": 0.7509756932557343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1795,
      "step": 21936
    },
    {
      "epoch": 0.7510099281068128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1899,
      "step": 21937
    },
    {
      "epoch": 0.7510441629578911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1906,
      "step": 21938
    },
    {
      "epoch": 0.7510783978089696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.166,
      "step": 21939
    },
    {
      "epoch": 0.7511126326600479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1635,
      "step": 21940
    },
    {
      "epoch": 0.7511468675111264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.173,
      "step": 21941
    },
    {
      "epoch": 0.7511811023622047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1826,
      "step": 21942
    },
    {
      "epoch": 0.7512153372132832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1896,
      "step": 21943
    },
    {
      "epoch": 0.7512495720643615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1752,
      "step": 21944
    },
    {
      "epoch": 0.75128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1847,
      "step": 21945
    },
    {
      "epoch": 0.7513180417665183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1882,
      "step": 21946
    },
    {
      "epoch": 0.7513522766175967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1798,
      "step": 21947
    },
    {
      "epoch": 0.7513865114686751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1871,
      "step": 21948
    },
    {
      "epoch": 0.7514207463197535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1723,
      "step": 21949
    },
    {
      "epoch": 0.7514549811708319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1955,
      "step": 21950
    },
    {
      "epoch": 0.7514892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1726,
      "step": 21951
    },
    {
      "epoch": 0.7515234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1781,
      "step": 21952
    },
    {
      "epoch": 0.7515576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1833,
      "step": 21953
    },
    {
      "epoch": 0.7515919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1733,
      "step": 21954
    },
    {
      "epoch": 0.7516261554262239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1707,
      "step": 21955
    },
    {
      "epoch": 0.7516603902773022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.177,
      "step": 21956
    },
    {
      "epoch": 0.7516946251283807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.169,
      "step": 21957
    },
    {
      "epoch": 0.751728859979459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1901,
      "step": 21958
    },
    {
      "epoch": 0.7517630948305375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1703,
      "step": 21959
    },
    {
      "epoch": 0.7517973296816158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1888,
      "step": 21960
    },
    {
      "epoch": 0.7518315645326943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1827,
      "step": 21961
    },
    {
      "epoch": 0.7518657993837726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1626,
      "step": 21962
    },
    {
      "epoch": 0.7519000342348511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1679,
      "step": 21963
    },
    {
      "epoch": 0.7519342690859295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1654,
      "step": 21964
    },
    {
      "epoch": 0.7519685039370079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.174,
      "step": 21965
    },
    {
      "epoch": 0.7520027387880863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1733,
      "step": 21966
    },
    {
      "epoch": 0.7520369736391647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1692,
      "step": 21967
    },
    {
      "epoch": 0.7520712084902431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1746,
      "step": 21968
    },
    {
      "epoch": 0.7521054433413215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1706,
      "step": 21969
    },
    {
      "epoch": 0.7521396781923999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1945,
      "step": 21970
    },
    {
      "epoch": 0.7521739130434782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.188,
      "step": 21971
    },
    {
      "epoch": 0.7522081478945567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1657,
      "step": 21972
    },
    {
      "epoch": 0.752242382745635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1775,
      "step": 21973
    },
    {
      "epoch": 0.7522766175967135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1896,
      "step": 21974
    },
    {
      "epoch": 0.7523108524477918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1892,
      "step": 21975
    },
    {
      "epoch": 0.7523450872988703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1753,
      "step": 21976
    },
    {
      "epoch": 0.7523793221499486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1754,
      "step": 21977
    },
    {
      "epoch": 0.7524135570010271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.178,
      "step": 21978
    },
    {
      "epoch": 0.7524477918521054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1827,
      "step": 21979
    },
    {
      "epoch": 0.7524820267031839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1879,
      "step": 21980
    },
    {
      "epoch": 0.7525162615542622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.168,
      "step": 21981
    },
    {
      "epoch": 0.7525504964053407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1954,
      "step": 21982
    },
    {
      "epoch": 0.752584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1819,
      "step": 21983
    },
    {
      "epoch": 0.7526189661074975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1907,
      "step": 21984
    },
    {
      "epoch": 0.7526532009585758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1746,
      "step": 21985
    },
    {
      "epoch": 0.7526874358096542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1561,
      "step": 21986
    },
    {
      "epoch": 0.7527216706607326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1775,
      "step": 21987
    },
    {
      "epoch": 0.752755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1914,
      "step": 21988
    },
    {
      "epoch": 0.7527901403628894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1719,
      "step": 21989
    },
    {
      "epoch": 0.7528243752139678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1888,
      "step": 21990
    },
    {
      "epoch": 0.7528586100650462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1893,
      "step": 21991
    },
    {
      "epoch": 0.7528928449161246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1663,
      "step": 21992
    },
    {
      "epoch": 0.752927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1888,
      "step": 21993
    },
    {
      "epoch": 0.7529613146182814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1951,
      "step": 21994
    },
    {
      "epoch": 0.7529955494693598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1761,
      "step": 21995
    },
    {
      "epoch": 0.7530297843204382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1614,
      "step": 21996
    },
    {
      "epoch": 0.7530640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1811,
      "step": 21997
    },
    {
      "epoch": 0.753098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1637,
      "step": 21998
    },
    {
      "epoch": 0.7531324888736733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1874,
      "step": 21999
    },
    {
      "epoch": 0.7531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1877,
      "step": 22000
    },
    {
      "epoch": 0.7532009585758302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1664,
      "step": 22001
    },
    {
      "epoch": 0.7532351934269086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1857,
      "step": 22002
    },
    {
      "epoch": 0.753269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1856,
      "step": 22003
    },
    {
      "epoch": 0.7533036631290654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1787,
      "step": 22004
    },
    {
      "epoch": 0.7533378979801438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1787,
      "step": 22005
    },
    {
      "epoch": 0.7533721328312222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1816,
      "step": 22006
    },
    {
      "epoch": 0.7534063676823006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.187,
      "step": 22007
    },
    {
      "epoch": 0.753440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1987,
      "step": 22008
    },
    {
      "epoch": 0.7534748373844574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1873,
      "step": 22009
    },
    {
      "epoch": 0.7535090722355358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1845,
      "step": 22010
    },
    {
      "epoch": 0.7535433070866142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1704,
      "step": 22011
    },
    {
      "epoch": 0.7535775419376926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1766,
      "step": 22012
    },
    {
      "epoch": 0.753611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1812,
      "step": 22013
    },
    {
      "epoch": 0.7536460116398493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1781,
      "step": 22014
    },
    {
      "epoch": 0.7536802464909278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1754,
      "step": 22015
    },
    {
      "epoch": 0.7537144813420061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.181,
      "step": 22016
    },
    {
      "epoch": 0.7537487161930846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1691,
      "step": 22017
    },
    {
      "epoch": 0.7537829510441629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1791,
      "step": 22018
    },
    {
      "epoch": 0.7538171858952414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1705,
      "step": 22019
    },
    {
      "epoch": 0.7538514207463197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1699,
      "step": 22020
    },
    {
      "epoch": 0.7538856555973982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.175,
      "step": 22021
    },
    {
      "epoch": 0.7539198904484765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.187,
      "step": 22022
    },
    {
      "epoch": 0.753954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1779,
      "step": 22023
    },
    {
      "epoch": 0.7539883601506333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1821,
      "step": 22024
    },
    {
      "epoch": 0.7540225950017118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.173,
      "step": 22025
    },
    {
      "epoch": 0.7540568298527901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1791,
      "step": 22026
    },
    {
      "epoch": 0.7540910647038686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1894,
      "step": 22027
    },
    {
      "epoch": 0.7541252995549469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1689,
      "step": 22028
    },
    {
      "epoch": 0.7541595344060253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1693,
      "step": 22029
    },
    {
      "epoch": 0.7541937692571037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1716,
      "step": 22030
    },
    {
      "epoch": 0.7542280041081821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.168,
      "step": 22031
    },
    {
      "epoch": 0.7542622389592605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1855,
      "step": 22032
    },
    {
      "epoch": 0.7542964738103389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1667,
      "step": 22033
    },
    {
      "epoch": 0.7543307086614173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1754,
      "step": 22034
    },
    {
      "epoch": 0.7543649435124957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1833,
      "step": 22035
    },
    {
      "epoch": 0.7543991783635741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.182,
      "step": 22036
    },
    {
      "epoch": 0.7544334132146525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1876,
      "step": 22037
    },
    {
      "epoch": 0.754467648065731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1721,
      "step": 22038
    },
    {
      "epoch": 0.7545018829168093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1911,
      "step": 22039
    },
    {
      "epoch": 0.7545361177678878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1841,
      "step": 22040
    },
    {
      "epoch": 0.7545703526189661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.18,
      "step": 22041
    },
    {
      "epoch": 0.7546045874700446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1745,
      "step": 22042
    },
    {
      "epoch": 0.7546388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1683,
      "step": 22043
    },
    {
      "epoch": 0.7546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.176,
      "step": 22044
    },
    {
      "epoch": 0.7547072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1804,
      "step": 22045
    },
    {
      "epoch": 0.7547415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1876,
      "step": 22046
    },
    {
      "epoch": 0.7547757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1835,
      "step": 22047
    },
    {
      "epoch": 0.7548099965765149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1837,
      "step": 22048
    },
    {
      "epoch": 0.7548442314275933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1835,
      "step": 22049
    },
    {
      "epoch": 0.7548784662786717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1796,
      "step": 22050
    },
    {
      "epoch": 0.7549127011297501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1988,
      "step": 22051
    },
    {
      "epoch": 0.7549469359808285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1797,
      "step": 22052
    },
    {
      "epoch": 0.7549811708319069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1798,
      "step": 22053
    },
    {
      "epoch": 0.7550154056829853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1819,
      "step": 22054
    },
    {
      "epoch": 0.7550496405340636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1887,
      "step": 22055
    },
    {
      "epoch": 0.7550838753851421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.183,
      "step": 22056
    },
    {
      "epoch": 0.7551181102362204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1722,
      "step": 22057
    },
    {
      "epoch": 0.7551523450872989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1717,
      "step": 22058
    },
    {
      "epoch": 0.7551865799383772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1822,
      "step": 22059
    },
    {
      "epoch": 0.7552208147894557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1728,
      "step": 22060
    },
    {
      "epoch": 0.755255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1726,
      "step": 22061
    },
    {
      "epoch": 0.7552892844916125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1714,
      "step": 22062
    },
    {
      "epoch": 0.7553235193426908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1721,
      "step": 22063
    },
    {
      "epoch": 0.7553577541937693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1532,
      "step": 22064
    },
    {
      "epoch": 0.7553919890448476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1885,
      "step": 22065
    },
    {
      "epoch": 0.7554262238959261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1817,
      "step": 22066
    },
    {
      "epoch": 0.7554604587470044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1786,
      "step": 22067
    },
    {
      "epoch": 0.7554946935980829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1565,
      "step": 22068
    },
    {
      "epoch": 0.7555289284491612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1901,
      "step": 22069
    },
    {
      "epoch": 0.7555631633002396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.182,
      "step": 22070
    },
    {
      "epoch": 0.755597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1838,
      "step": 22071
    },
    {
      "epoch": 0.7556316330023964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1877,
      "step": 22072
    },
    {
      "epoch": 0.7556658678534748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1739,
      "step": 22073
    },
    {
      "epoch": 0.7557001027045532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1848,
      "step": 22074
    },
    {
      "epoch": 0.7557343375556317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1677,
      "step": 22075
    },
    {
      "epoch": 0.75576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.178,
      "step": 22076
    },
    {
      "epoch": 0.7558028072577885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1711,
      "step": 22077
    },
    {
      "epoch": 0.7558370421088668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1681,
      "step": 22078
    },
    {
      "epoch": 0.7558712769599453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1786,
      "step": 22079
    },
    {
      "epoch": 0.7559055118110236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1717,
      "step": 22080
    },
    {
      "epoch": 0.7559397466621021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1847,
      "step": 22081
    },
    {
      "epoch": 0.7559739815131804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1815,
      "step": 22082
    },
    {
      "epoch": 0.7560082163642589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.177,
      "step": 22083
    },
    {
      "epoch": 0.7560424512153372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1796,
      "step": 22084
    },
    {
      "epoch": 0.7560766860664156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1599,
      "step": 22085
    },
    {
      "epoch": 0.756110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1835,
      "step": 22086
    },
    {
      "epoch": 0.7561451557685724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1596,
      "step": 22087
    },
    {
      "epoch": 0.7561793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2067,
      "step": 22088
    },
    {
      "epoch": 0.7562136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1705,
      "step": 22089
    },
    {
      "epoch": 0.7562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1732,
      "step": 22090
    },
    {
      "epoch": 0.756282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1807,
      "step": 22091
    },
    {
      "epoch": 0.7563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1682,
      "step": 22092
    },
    {
      "epoch": 0.7563505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1762,
      "step": 22093
    },
    {
      "epoch": 0.7563847997261212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1807,
      "step": 22094
    },
    {
      "epoch": 0.7564190345771996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1773,
      "step": 22095
    },
    {
      "epoch": 0.756453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2009,
      "step": 22096
    },
    {
      "epoch": 0.7564875042793564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1625,
      "step": 22097
    },
    {
      "epoch": 0.7565217391304347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.188,
      "step": 22098
    },
    {
      "epoch": 0.7565559739815132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1825,
      "step": 22099
    },
    {
      "epoch": 0.7565902088325915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1718,
      "step": 22100
    },
    {
      "epoch": 0.75662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1589,
      "step": 22101
    },
    {
      "epoch": 0.7566586785347483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1746,
      "step": 22102
    },
    {
      "epoch": 0.7566929133858268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1852,
      "step": 22103
    },
    {
      "epoch": 0.7567271482369051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1768,
      "step": 22104
    },
    {
      "epoch": 0.7567613830879836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1738,
      "step": 22105
    },
    {
      "epoch": 0.7567956179390619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1617,
      "step": 22106
    },
    {
      "epoch": 0.7568298527901404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1697,
      "step": 22107
    },
    {
      "epoch": 0.7568640876412187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1814,
      "step": 22108
    },
    {
      "epoch": 0.7568983224922972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1724,
      "step": 22109
    },
    {
      "epoch": 0.7569325573433756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.186,
      "step": 22110
    },
    {
      "epoch": 0.756966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1745,
      "step": 22111
    },
    {
      "epoch": 0.7570010270455324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1702,
      "step": 22112
    },
    {
      "epoch": 0.7570352618966107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1615,
      "step": 22113
    },
    {
      "epoch": 0.7570694967476892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1744,
      "step": 22114
    },
    {
      "epoch": 0.7571037315987675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1735,
      "step": 22115
    },
    {
      "epoch": 0.757137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1809,
      "step": 22116
    },
    {
      "epoch": 0.7571722013009243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1813,
      "step": 22117
    },
    {
      "epoch": 0.7572064361520028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1758,
      "step": 22118
    },
    {
      "epoch": 0.7572406710030811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1675,
      "step": 22119
    },
    {
      "epoch": 0.7572749058541596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1967,
      "step": 22120
    },
    {
      "epoch": 0.7573091407052379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1843,
      "step": 22121
    },
    {
      "epoch": 0.7573433755563164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1618,
      "step": 22122
    },
    {
      "epoch": 0.7573776104073947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1738,
      "step": 22123
    },
    {
      "epoch": 0.7574118452584732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1751,
      "step": 22124
    },
    {
      "epoch": 0.7574460801095515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1647,
      "step": 22125
    },
    {
      "epoch": 0.75748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1757,
      "step": 22126
    },
    {
      "epoch": 0.7575145498117083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1905,
      "step": 22127
    },
    {
      "epoch": 0.7575487846627867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.152,
      "step": 22128
    },
    {
      "epoch": 0.7575830195138651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1908,
      "step": 22129
    },
    {
      "epoch": 0.7576172543649435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1639,
      "step": 22130
    },
    {
      "epoch": 0.7576514892160219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1676,
      "step": 22131
    },
    {
      "epoch": 0.7576857240671003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1857,
      "step": 22132
    },
    {
      "epoch": 0.7577199589181787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.178,
      "step": 22133
    },
    {
      "epoch": 0.7577541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1791,
      "step": 22134
    },
    {
      "epoch": 0.7577884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.168,
      "step": 22135
    },
    {
      "epoch": 0.7578226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1733,
      "step": 22136
    },
    {
      "epoch": 0.7578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1878,
      "step": 22137
    },
    {
      "epoch": 0.7578911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.176,
      "step": 22138
    },
    {
      "epoch": 0.757925368024649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1623,
      "step": 22139
    },
    {
      "epoch": 0.7579596028757275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1707,
      "step": 22140
    },
    {
      "epoch": 0.7579938377268058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1815,
      "step": 22141
    },
    {
      "epoch": 0.7580280725778843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1783,
      "step": 22142
    },
    {
      "epoch": 0.7580623074289626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1574,
      "step": 22143
    },
    {
      "epoch": 0.7580965422800411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1831,
      "step": 22144
    },
    {
      "epoch": 0.7581307771311194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1552,
      "step": 22145
    },
    {
      "epoch": 0.7581650119821979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1769,
      "step": 22146
    },
    {
      "epoch": 0.7581992468332763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1758,
      "step": 22147
    },
    {
      "epoch": 0.7582334816843547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1695,
      "step": 22148
    },
    {
      "epoch": 0.7582677165354331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1703,
      "step": 22149
    },
    {
      "epoch": 0.7583019513865115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1741,
      "step": 22150
    },
    {
      "epoch": 0.7583361862375899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1533,
      "step": 22151
    },
    {
      "epoch": 0.7583704210886683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1727,
      "step": 22152
    },
    {
      "epoch": 0.7584046559397467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1668,
      "step": 22153
    },
    {
      "epoch": 0.758438890790825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1894,
      "step": 22154
    },
    {
      "epoch": 0.7584731256419035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1684,
      "step": 22155
    },
    {
      "epoch": 0.7585073604929818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1788,
      "step": 22156
    },
    {
      "epoch": 0.7585415953440603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1994,
      "step": 22157
    },
    {
      "epoch": 0.7585758301951386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.183,
      "step": 22158
    },
    {
      "epoch": 0.7586100650462171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1811,
      "step": 22159
    },
    {
      "epoch": 0.7586442998972954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1721,
      "step": 22160
    },
    {
      "epoch": 0.7586785347483739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1857,
      "step": 22161
    },
    {
      "epoch": 0.7587127695994522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1762,
      "step": 22162
    },
    {
      "epoch": 0.7587470044505307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1705,
      "step": 22163
    },
    {
      "epoch": 0.758781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1848,
      "step": 22164
    },
    {
      "epoch": 0.7588154741526875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1801,
      "step": 22165
    },
    {
      "epoch": 0.7588497090037658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1716,
      "step": 22166
    },
    {
      "epoch": 0.7588839438548443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1646,
      "step": 22167
    },
    {
      "epoch": 0.7589181787059226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1726,
      "step": 22168
    },
    {
      "epoch": 0.758952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.191,
      "step": 22169
    },
    {
      "epoch": 0.7589866484080794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1746,
      "step": 22170
    },
    {
      "epoch": 0.7590208832591578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1781,
      "step": 22171
    },
    {
      "epoch": 0.7590551181102362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1643,
      "step": 22172
    },
    {
      "epoch": 0.7590893529613146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1836,
      "step": 22173
    },
    {
      "epoch": 0.759123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1729,
      "step": 22174
    },
    {
      "epoch": 0.7591578226634714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1803,
      "step": 22175
    },
    {
      "epoch": 0.7591920575145498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1605,
      "step": 22176
    },
    {
      "epoch": 0.7592262923656282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1764,
      "step": 22177
    },
    {
      "epoch": 0.7592605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1703,
      "step": 22178
    },
    {
      "epoch": 0.759294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1755,
      "step": 22179
    },
    {
      "epoch": 0.7593289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1812,
      "step": 22180
    },
    {
      "epoch": 0.7593632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1828,
      "step": 22181
    },
    {
      "epoch": 0.7593974666210201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1933,
      "step": 22182
    },
    {
      "epoch": 0.7594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1815,
      "step": 22183
    },
    {
      "epoch": 0.759465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1871,
      "step": 22184
    },
    {
      "epoch": 0.7595001711742554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1743,
      "step": 22185
    },
    {
      "epoch": 0.7595344060253338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.174,
      "step": 22186
    },
    {
      "epoch": 0.7595686408764122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.204,
      "step": 22187
    },
    {
      "epoch": 0.7596028757274906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2083,
      "step": 22188
    },
    {
      "epoch": 0.759637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1612,
      "step": 22189
    },
    {
      "epoch": 0.7596713454296474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1899,
      "step": 22190
    },
    {
      "epoch": 0.7597055802807258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1635,
      "step": 22191
    },
    {
      "epoch": 0.7597398151318042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1642,
      "step": 22192
    },
    {
      "epoch": 0.7597740499828826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1663,
      "step": 22193
    },
    {
      "epoch": 0.759808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1826,
      "step": 22194
    },
    {
      "epoch": 0.7598425196850394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1798,
      "step": 22195
    },
    {
      "epoch": 0.7598767545361178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1862,
      "step": 22196
    },
    {
      "epoch": 0.7599109893871961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1745,
      "step": 22197
    },
    {
      "epoch": 0.7599452242382746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1784,
      "step": 22198
    },
    {
      "epoch": 0.7599794590893529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1703,
      "step": 22199
    },
    {
      "epoch": 0.7600136939404314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1736,
      "step": 22200
    },
    {
      "epoch": 0.7600479287915097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.199,
      "step": 22201
    },
    {
      "epoch": 0.7600821636425882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1919,
      "step": 22202
    },
    {
      "epoch": 0.7601163984936665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1923,
      "step": 22203
    },
    {
      "epoch": 0.760150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1754,
      "step": 22204
    },
    {
      "epoch": 0.7601848681958233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1671,
      "step": 22205
    },
    {
      "epoch": 0.7602191030469018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1715,
      "step": 22206
    },
    {
      "epoch": 0.7602533378979801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1724,
      "step": 22207
    },
    {
      "epoch": 0.7602875727490586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1701,
      "step": 22208
    },
    {
      "epoch": 0.7603218076001369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1865,
      "step": 22209
    },
    {
      "epoch": 0.7603560424512154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1762,
      "step": 22210
    },
    {
      "epoch": 0.7603902773022937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1697,
      "step": 22211
    },
    {
      "epoch": 0.7604245121533721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1794,
      "step": 22212
    },
    {
      "epoch": 0.7604587470044505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1599,
      "step": 22213
    },
    {
      "epoch": 0.7604929818555289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1686,
      "step": 22214
    },
    {
      "epoch": 0.7605272167066073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1829,
      "step": 22215
    },
    {
      "epoch": 0.7605614515576857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1695,
      "step": 22216
    },
    {
      "epoch": 0.7605956864087641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1904,
      "step": 22217
    },
    {
      "epoch": 0.7606299212598425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.188,
      "step": 22218
    },
    {
      "epoch": 0.7606641561109209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1775,
      "step": 22219
    },
    {
      "epoch": 0.7606983909619993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1654,
      "step": 22220
    },
    {
      "epoch": 0.7607326258130778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.173,
      "step": 22221
    },
    {
      "epoch": 0.7607668606641561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1855,
      "step": 22222
    },
    {
      "epoch": 0.7608010955152346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1731,
      "step": 22223
    },
    {
      "epoch": 0.7608353303663129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1791,
      "step": 22224
    },
    {
      "epoch": 0.7608695652173914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1741,
      "step": 22225
    },
    {
      "epoch": 0.7609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1733,
      "step": 22226
    },
    {
      "epoch": 0.7609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.174,
      "step": 22227
    },
    {
      "epoch": 0.7609722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1822,
      "step": 22228
    },
    {
      "epoch": 0.7610065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1702,
      "step": 22229
    },
    {
      "epoch": 0.7610407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1849,
      "step": 22230
    },
    {
      "epoch": 0.7610749743238617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1718,
      "step": 22231
    },
    {
      "epoch": 0.7611092091749401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1831,
      "step": 22232
    },
    {
      "epoch": 0.7611434440260185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.182,
      "step": 22233
    },
    {
      "epoch": 0.7611776788770969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1784,
      "step": 22234
    },
    {
      "epoch": 0.7612119137281753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1746,
      "step": 22235
    },
    {
      "epoch": 0.7612461485792537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1779,
      "step": 22236
    },
    {
      "epoch": 0.7612803834303321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1827,
      "step": 22237
    },
    {
      "epoch": 0.7613146182814104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1821,
      "step": 22238
    },
    {
      "epoch": 0.7613488531324889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1758,
      "step": 22239
    },
    {
      "epoch": 0.7613830879835672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1728,
      "step": 22240
    },
    {
      "epoch": 0.7614173228346457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1679,
      "step": 22241
    },
    {
      "epoch": 0.761451557685724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1815,
      "step": 22242
    },
    {
      "epoch": 0.7614857925368025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1718,
      "step": 22243
    },
    {
      "epoch": 0.7615200273878808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.176,
      "step": 22244
    },
    {
      "epoch": 0.7615542622389593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.15,
      "step": 22245
    },
    {
      "epoch": 0.7615884970900376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1847,
      "step": 22246
    },
    {
      "epoch": 0.7616227319411161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1774,
      "step": 22247
    },
    {
      "epoch": 0.7616569667921944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1705,
      "step": 22248
    },
    {
      "epoch": 0.7616912016432729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1721,
      "step": 22249
    },
    {
      "epoch": 0.7617254364943512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.18,
      "step": 22250
    },
    {
      "epoch": 0.7617596713454297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1682,
      "step": 22251
    },
    {
      "epoch": 0.761793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.165,
      "step": 22252
    },
    {
      "epoch": 0.7618281410475864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1696,
      "step": 22253
    },
    {
      "epoch": 0.7618623758986648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1773,
      "step": 22254
    },
    {
      "epoch": 0.7618966107497432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1748,
      "step": 22255
    },
    {
      "epoch": 0.7619308456008216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1697,
      "step": 22256
    },
    {
      "epoch": 0.7619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1772,
      "step": 22257
    },
    {
      "epoch": 0.7619993153029785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1738,
      "step": 22258
    },
    {
      "epoch": 0.7620335501540568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1806,
      "step": 22259
    },
    {
      "epoch": 0.7620677850051353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1697,
      "step": 22260
    },
    {
      "epoch": 0.7621020198562136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1572,
      "step": 22261
    },
    {
      "epoch": 0.7621362547072921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1701,
      "step": 22262
    },
    {
      "epoch": 0.7621704895583704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1789,
      "step": 22263
    },
    {
      "epoch": 0.7622047244094489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1787,
      "step": 22264
    },
    {
      "epoch": 0.7622389592605272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1802,
      "step": 22265
    },
    {
      "epoch": 0.7622731941116057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1664,
      "step": 22266
    },
    {
      "epoch": 0.762307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1646,
      "step": 22267
    },
    {
      "epoch": 0.7623416638137624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1717,
      "step": 22268
    },
    {
      "epoch": 0.7623758986648408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1848,
      "step": 22269
    },
    {
      "epoch": 0.7624101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1583,
      "step": 22270
    },
    {
      "epoch": 0.7624443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1631,
      "step": 22271
    },
    {
      "epoch": 0.762478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1863,
      "step": 22272
    },
    {
      "epoch": 0.7625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1733,
      "step": 22273
    },
    {
      "epoch": 0.7625470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1878,
      "step": 22274
    },
    {
      "epoch": 0.7625813077713112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1674,
      "step": 22275
    },
    {
      "epoch": 0.7626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1762,
      "step": 22276
    },
    {
      "epoch": 0.762649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1661,
      "step": 22277
    },
    {
      "epoch": 0.7626840123245464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1781,
      "step": 22278
    },
    {
      "epoch": 0.7627182471756248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1794,
      "step": 22279
    },
    {
      "epoch": 0.7627524820267032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.179,
      "step": 22280
    },
    {
      "epoch": 0.7627867168777815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1776,
      "step": 22281
    },
    {
      "epoch": 0.76282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1599,
      "step": 22282
    },
    {
      "epoch": 0.7628551865799383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1584,
      "step": 22283
    },
    {
      "epoch": 0.7628894214310168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1862,
      "step": 22284
    },
    {
      "epoch": 0.7629236562820951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1822,
      "step": 22285
    },
    {
      "epoch": 0.7629578911331736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1672,
      "step": 22286
    },
    {
      "epoch": 0.7629921259842519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1529,
      "step": 22287
    },
    {
      "epoch": 0.7630263608353304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1649,
      "step": 22288
    },
    {
      "epoch": 0.7630605956864087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.174,
      "step": 22289
    },
    {
      "epoch": 0.7630948305374872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1617,
      "step": 22290
    },
    {
      "epoch": 0.7631290653885655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1631,
      "step": 22291
    },
    {
      "epoch": 0.763163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.196,
      "step": 22292
    },
    {
      "epoch": 0.7631975350907223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1692,
      "step": 22293
    },
    {
      "epoch": 0.7632317699418008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1739,
      "step": 22294
    },
    {
      "epoch": 0.7632660047928792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1847,
      "step": 22295
    },
    {
      "epoch": 0.7633002396439575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1774,
      "step": 22296
    },
    {
      "epoch": 0.763334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1756,
      "step": 22297
    },
    {
      "epoch": 0.7633687093461143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1698,
      "step": 22298
    },
    {
      "epoch": 0.7634029441971928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1596,
      "step": 22299
    },
    {
      "epoch": 0.7634371790482711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1721,
      "step": 22300
    },
    {
      "epoch": 0.7634714138993496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1769,
      "step": 22301
    },
    {
      "epoch": 0.7635056487504279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1731,
      "step": 22302
    },
    {
      "epoch": 0.7635398836015064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1892,
      "step": 22303
    },
    {
      "epoch": 0.7635741184525847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1884,
      "step": 22304
    },
    {
      "epoch": 0.7636083533036632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1705,
      "step": 22305
    },
    {
      "epoch": 0.7636425881547415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1724,
      "step": 22306
    },
    {
      "epoch": 0.76367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1697,
      "step": 22307
    },
    {
      "epoch": 0.7637110578568983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1635,
      "step": 22308
    },
    {
      "epoch": 0.7637452927079768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1576,
      "step": 22309
    },
    {
      "epoch": 0.7637795275590551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1745,
      "step": 22310
    },
    {
      "epoch": 0.7638137624101335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1698,
      "step": 22311
    },
    {
      "epoch": 0.7638479972612119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1693,
      "step": 22312
    },
    {
      "epoch": 0.7638822321122903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1736,
      "step": 22313
    },
    {
      "epoch": 0.7639164669633687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.154,
      "step": 22314
    },
    {
      "epoch": 0.7639507018144471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1553,
      "step": 22315
    },
    {
      "epoch": 0.7639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1772,
      "step": 22316
    },
    {
      "epoch": 0.7640191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1582,
      "step": 22317
    },
    {
      "epoch": 0.7640534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1852,
      "step": 22318
    },
    {
      "epoch": 0.7640876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1795,
      "step": 22319
    },
    {
      "epoch": 0.7641218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1802,
      "step": 22320
    },
    {
      "epoch": 0.7641561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1768,
      "step": 22321
    },
    {
      "epoch": 0.7641903457719958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1522,
      "step": 22322
    },
    {
      "epoch": 0.7642245806230743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1859,
      "step": 22323
    },
    {
      "epoch": 0.7642588154741526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1826,
      "step": 22324
    },
    {
      "epoch": 0.7642930503252311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1589,
      "step": 22325
    },
    {
      "epoch": 0.7643272851763094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1525,
      "step": 22326
    },
    {
      "epoch": 0.7643615200273879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1813,
      "step": 22327
    },
    {
      "epoch": 0.7643957548784662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1668,
      "step": 22328
    },
    {
      "epoch": 0.7644299897295447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1878,
      "step": 22329
    },
    {
      "epoch": 0.764464224580623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1664,
      "step": 22330
    },
    {
      "epoch": 0.7644984594317015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1725,
      "step": 22331
    },
    {
      "epoch": 0.7645326942827799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1831,
      "step": 22332
    },
    {
      "epoch": 0.7645669291338583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.181,
      "step": 22333
    },
    {
      "epoch": 0.7646011639849367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1559,
      "step": 22334
    },
    {
      "epoch": 0.7646353988360151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1712,
      "step": 22335
    },
    {
      "epoch": 0.7646696336870935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.195,
      "step": 22336
    },
    {
      "epoch": 0.7647038685381718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1785,
      "step": 22337
    },
    {
      "epoch": 0.7647381033892503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1924,
      "step": 22338
    },
    {
      "epoch": 0.7647723382403286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.18,
      "step": 22339
    },
    {
      "epoch": 0.7648065730914071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1817,
      "step": 22340
    },
    {
      "epoch": 0.7648408079424854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1616,
      "step": 22341
    },
    {
      "epoch": 0.7648750427935639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.167,
      "step": 22342
    },
    {
      "epoch": 0.7649092776446422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.182,
      "step": 22343
    },
    {
      "epoch": 0.7649435124957207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1685,
      "step": 22344
    },
    {
      "epoch": 0.764977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1881,
      "step": 22345
    },
    {
      "epoch": 0.7650119821978775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.181,
      "step": 22346
    },
    {
      "epoch": 0.7650462170489558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1722,
      "step": 22347
    },
    {
      "epoch": 0.7650804519000343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.2021,
      "step": 22348
    },
    {
      "epoch": 0.7651146867511126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.17,
      "step": 22349
    },
    {
      "epoch": 0.7651489216021911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.176,
      "step": 22350
    },
    {
      "epoch": 0.7651831564532694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1584,
      "step": 22351
    },
    {
      "epoch": 0.7652173913043478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1879,
      "step": 22352
    },
    {
      "epoch": 0.7652516261554262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1724,
      "step": 22353
    },
    {
      "epoch": 0.7652858610065046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1567,
      "step": 22354
    },
    {
      "epoch": 0.765320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1797,
      "step": 22355
    },
    {
      "epoch": 0.7653543307086614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.176,
      "step": 22356
    },
    {
      "epoch": 0.7653885655597398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1741,
      "step": 22357
    },
    {
      "epoch": 0.7654228004108182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1935,
      "step": 22358
    },
    {
      "epoch": 0.7654570352618966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1782,
      "step": 22359
    },
    {
      "epoch": 0.765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.172,
      "step": 22360
    },
    {
      "epoch": 0.7655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1625,
      "step": 22361
    },
    {
      "epoch": 0.7655597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1694,
      "step": 22362
    },
    {
      "epoch": 0.7655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1608,
      "step": 22363
    },
    {
      "epoch": 0.7656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1857,
      "step": 22364
    },
    {
      "epoch": 0.765662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1611,
      "step": 22365
    },
    {
      "epoch": 0.7656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1877,
      "step": 22366
    },
    {
      "epoch": 0.7657309140705237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1614,
      "step": 22367
    },
    {
      "epoch": 0.7657651489216022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1761,
      "step": 22368
    },
    {
      "epoch": 0.7657993837726806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1727,
      "step": 22369
    },
    {
      "epoch": 0.765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1694,
      "step": 22370
    },
    {
      "epoch": 0.7658678534748374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1699,
      "step": 22371
    },
    {
      "epoch": 0.7659020883259158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1504,
      "step": 22372
    },
    {
      "epoch": 0.7659363231769942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1629,
      "step": 22373
    },
    {
      "epoch": 0.7659705580280726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1645,
      "step": 22374
    },
    {
      "epoch": 0.766004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1845,
      "step": 22375
    },
    {
      "epoch": 0.7660390277302294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1796,
      "step": 22376
    },
    {
      "epoch": 0.7660732625813078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1823,
      "step": 22377
    },
    {
      "epoch": 0.7661074974323862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.176,
      "step": 22378
    },
    {
      "epoch": 0.7661417322834646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1586,
      "step": 22379
    },
    {
      "epoch": 0.766175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1653,
      "step": 22380
    },
    {
      "epoch": 0.7662102019856214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1696,
      "step": 22381
    },
    {
      "epoch": 0.7662444368366997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1702,
      "step": 22382
    },
    {
      "epoch": 0.7662786716877782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.165,
      "step": 22383
    },
    {
      "epoch": 0.7663129065388565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1702,
      "step": 22384
    },
    {
      "epoch": 0.766347141389935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1604,
      "step": 22385
    },
    {
      "epoch": 0.7663813762410133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.174,
      "step": 22386
    },
    {
      "epoch": 0.7664156110920918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1708,
      "step": 22387
    },
    {
      "epoch": 0.7664498459431701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.171,
      "step": 22388
    },
    {
      "epoch": 0.7664840807942486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1795,
      "step": 22389
    },
    {
      "epoch": 0.7665183156453269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1825,
      "step": 22390
    },
    {
      "epoch": 0.7665525504964054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1617,
      "step": 22391
    },
    {
      "epoch": 0.7665867853474837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1787,
      "step": 22392
    },
    {
      "epoch": 0.7666210201985622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.148,
      "step": 22393
    },
    {
      "epoch": 0.7666552550496405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1803,
      "step": 22394
    },
    {
      "epoch": 0.766689489900719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1641,
      "step": 22395
    },
    {
      "epoch": 0.7667237247517973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1529,
      "step": 22396
    },
    {
      "epoch": 0.7667579596028757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1684,
      "step": 22397
    },
    {
      "epoch": 0.7667921944539541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1732,
      "step": 22398
    },
    {
      "epoch": 0.7668264293050325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1805,
      "step": 22399
    },
    {
      "epoch": 0.7668606641561109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1696,
      "step": 22400
    },
    {
      "epoch": 0.7668948990071893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1739,
      "step": 22401
    },
    {
      "epoch": 0.7669291338582677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1776,
      "step": 22402
    },
    {
      "epoch": 0.7669633687093461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1877,
      "step": 22403
    },
    {
      "epoch": 0.7669976035604245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1591,
      "step": 22404
    },
    {
      "epoch": 0.7670318384115029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1745,
      "step": 22405
    },
    {
      "epoch": 0.7670660732625814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1661,
      "step": 22406
    },
    {
      "epoch": 0.7671003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1674,
      "step": 22407
    },
    {
      "epoch": 0.7671345429647382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1638,
      "step": 22408
    },
    {
      "epoch": 0.7671687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1816,
      "step": 22409
    },
    {
      "epoch": 0.767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1708,
      "step": 22410
    },
    {
      "epoch": 0.7672372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1715,
      "step": 22411
    },
    {
      "epoch": 0.7672714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.195,
      "step": 22412
    },
    {
      "epoch": 0.7673057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1731,
      "step": 22413
    },
    {
      "epoch": 0.7673399520712085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1643,
      "step": 22414
    },
    {
      "epoch": 0.7673741869222869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1649,
      "step": 22415
    },
    {
      "epoch": 0.7674084217733653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1772,
      "step": 22416
    },
    {
      "epoch": 0.7674426566244437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1705,
      "step": 22417
    },
    {
      "epoch": 0.7674768914755221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1846,
      "step": 22418
    },
    {
      "epoch": 0.7675111263266005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1768,
      "step": 22419
    },
    {
      "epoch": 0.7675453611776789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1722,
      "step": 22420
    },
    {
      "epoch": 0.7675795960287572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1753,
      "step": 22421
    },
    {
      "epoch": 0.7676138308798357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1623,
      "step": 22422
    },
    {
      "epoch": 0.767648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1743,
      "step": 22423
    },
    {
      "epoch": 0.7676823005819925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.159,
      "step": 22424
    },
    {
      "epoch": 0.7677165354330708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1561,
      "step": 22425
    },
    {
      "epoch": 0.7677507702841493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1802,
      "step": 22426
    },
    {
      "epoch": 0.7677850051352276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1671,
      "step": 22427
    },
    {
      "epoch": 0.7678192399863061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1408,
      "step": 22428
    },
    {
      "epoch": 0.7678534748373844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1744,
      "step": 22429
    },
    {
      "epoch": 0.7678877096884629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1671,
      "step": 22430
    },
    {
      "epoch": 0.7679219445395412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1697,
      "step": 22431
    },
    {
      "epoch": 0.7679561793906197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1599,
      "step": 22432
    },
    {
      "epoch": 0.767990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1587,
      "step": 22433
    },
    {
      "epoch": 0.7680246490927765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1626,
      "step": 22434
    },
    {
      "epoch": 0.7680588839438548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1736,
      "step": 22435
    },
    {
      "epoch": 0.7680931187949332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1893,
      "step": 22436
    },
    {
      "epoch": 0.7681273536460116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1733,
      "step": 22437
    },
    {
      "epoch": 0.76816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1776,
      "step": 22438
    },
    {
      "epoch": 0.7681958233481684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1934,
      "step": 22439
    },
    {
      "epoch": 0.7682300581992468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1815,
      "step": 22440
    },
    {
      "epoch": 0.7682642930503253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.182,
      "step": 22441
    },
    {
      "epoch": 0.7682985279014036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1569,
      "step": 22442
    },
    {
      "epoch": 0.7683327627524821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1758,
      "step": 22443
    },
    {
      "epoch": 0.7683669976035604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1792,
      "step": 22444
    },
    {
      "epoch": 0.7684012324546389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.172,
      "step": 22445
    },
    {
      "epoch": 0.7684354673057172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.148,
      "step": 22446
    },
    {
      "epoch": 0.7684697021567957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1624,
      "step": 22447
    },
    {
      "epoch": 0.768503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.181,
      "step": 22448
    },
    {
      "epoch": 0.7685381718589525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.179,
      "step": 22449
    },
    {
      "epoch": 0.7685724067100308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1614,
      "step": 22450
    },
    {
      "epoch": 0.7686066415611092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1607,
      "step": 22451
    },
    {
      "epoch": 0.7686408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1651,
      "step": 22452
    },
    {
      "epoch": 0.768675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1611,
      "step": 22453
    },
    {
      "epoch": 0.7687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1599,
      "step": 22454
    },
    {
      "epoch": 0.7687435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.165,
      "step": 22455
    },
    {
      "epoch": 0.7687778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.162,
      "step": 22456
    },
    {
      "epoch": 0.7688120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1692,
      "step": 22457
    },
    {
      "epoch": 0.768846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1483,
      "step": 22458
    },
    {
      "epoch": 0.7688805203697364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1586,
      "step": 22459
    },
    {
      "epoch": 0.7689147552208148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1571,
      "step": 22460
    },
    {
      "epoch": 0.7689489900718932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1569,
      "step": 22461
    },
    {
      "epoch": 0.7689832249229716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1595,
      "step": 22462
    },
    {
      "epoch": 0.76901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1699,
      "step": 22463
    },
    {
      "epoch": 0.7690516946251283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1645,
      "step": 22464
    },
    {
      "epoch": 0.7690859294762068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1563,
      "step": 22465
    },
    {
      "epoch": 0.7691201643272851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1575,
      "step": 22466
    },
    {
      "epoch": 0.7691543991783636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1632,
      "step": 22467
    },
    {
      "epoch": 0.7691886340294419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1701,
      "step": 22468
    },
    {
      "epoch": 0.7692228688805204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1733,
      "step": 22469
    },
    {
      "epoch": 0.7692571037315987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1763,
      "step": 22470
    },
    {
      "epoch": 0.7692913385826772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1699,
      "step": 22471
    },
    {
      "epoch": 0.7693255734337555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1669,
      "step": 22472
    },
    {
      "epoch": 0.769359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1468,
      "step": 22473
    },
    {
      "epoch": 0.7693940431359123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1749,
      "step": 22474
    },
    {
      "epoch": 0.7694282779869908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1793,
      "step": 22475
    },
    {
      "epoch": 0.7694625128380691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1723,
      "step": 22476
    },
    {
      "epoch": 0.7694967476891476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1797,
      "step": 22477
    },
    {
      "epoch": 0.769530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.192,
      "step": 22478
    },
    {
      "epoch": 0.7695652173913043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1698,
      "step": 22479
    },
    {
      "epoch": 0.7695994522423828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.171,
      "step": 22480
    },
    {
      "epoch": 0.7696336870934611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1633,
      "step": 22481
    },
    {
      "epoch": 0.7696679219445396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.185,
      "step": 22482
    },
    {
      "epoch": 0.7697021567956179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1647,
      "step": 22483
    },
    {
      "epoch": 0.7697363916466964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.182,
      "step": 22484
    },
    {
      "epoch": 0.7697706264977747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1751,
      "step": 22485
    },
    {
      "epoch": 0.7698048613488532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1623,
      "step": 22486
    },
    {
      "epoch": 0.7698390961999315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1766,
      "step": 22487
    },
    {
      "epoch": 0.76987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1603,
      "step": 22488
    },
    {
      "epoch": 0.7699075659020883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1704,
      "step": 22489
    },
    {
      "epoch": 0.7699418007531668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1741,
      "step": 22490
    },
    {
      "epoch": 0.7699760356042451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1807,
      "step": 22491
    },
    {
      "epoch": 0.7700102704553236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1818,
      "step": 22492
    },
    {
      "epoch": 0.7700445053064019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1859,
      "step": 22493
    },
    {
      "epoch": 0.7700787401574803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1751,
      "step": 22494
    },
    {
      "epoch": 0.7701129750085587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1643,
      "step": 22495
    },
    {
      "epoch": 0.7701472098596371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1832,
      "step": 22496
    },
    {
      "epoch": 0.7701814447107155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1577,
      "step": 22497
    },
    {
      "epoch": 0.7702156795617939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1857,
      "step": 22498
    },
    {
      "epoch": 0.7702499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1782,
      "step": 22499
    },
    {
      "epoch": 0.7702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1643,
      "step": 22500
    },
    {
      "epoch": 0.7703183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1742,
      "step": 22501
    },
    {
      "epoch": 0.7703526189661075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1795,
      "step": 22502
    },
    {
      "epoch": 0.7703868538171859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.194,
      "step": 22503
    },
    {
      "epoch": 0.7704210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1677,
      "step": 22504
    },
    {
      "epoch": 0.7704553235193426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1707,
      "step": 22505
    },
    {
      "epoch": 0.7704895583704211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.174,
      "step": 22506
    },
    {
      "epoch": 0.7705237932214994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1678,
      "step": 22507
    },
    {
      "epoch": 0.7705580280725779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1716,
      "step": 22508
    },
    {
      "epoch": 0.7705922629236562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1527,
      "step": 22509
    },
    {
      "epoch": 0.7706264977747347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1798,
      "step": 22510
    },
    {
      "epoch": 0.770660732625813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1711,
      "step": 22511
    },
    {
      "epoch": 0.7706949674768915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1452,
      "step": 22512
    },
    {
      "epoch": 0.7707292023279698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.16,
      "step": 22513
    },
    {
      "epoch": 0.7707634371790483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1786,
      "step": 22514
    },
    {
      "epoch": 0.7707976720301267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.171,
      "step": 22515
    },
    {
      "epoch": 0.7708319068812051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.157,
      "step": 22516
    },
    {
      "epoch": 0.7708661417322835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1577,
      "step": 22517
    },
    {
      "epoch": 0.7709003765833619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1814,
      "step": 22518
    },
    {
      "epoch": 0.7709346114344403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1712,
      "step": 22519
    },
    {
      "epoch": 0.7709688462855186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1522,
      "step": 22520
    },
    {
      "epoch": 0.7710030811365971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1679,
      "step": 22521
    },
    {
      "epoch": 0.7710373159876754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1596,
      "step": 22522
    },
    {
      "epoch": 0.7710715508387539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1581,
      "step": 22523
    },
    {
      "epoch": 0.7711057856898322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1685,
      "step": 22524
    },
    {
      "epoch": 0.7711400205409107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1602,
      "step": 22525
    },
    {
      "epoch": 0.771174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1838,
      "step": 22526
    },
    {
      "epoch": 0.7712084902430675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1828,
      "step": 22527
    },
    {
      "epoch": 0.7712427250941458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1649,
      "step": 22528
    },
    {
      "epoch": 0.7712769599452243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1673,
      "step": 22529
    },
    {
      "epoch": 0.7713111947963026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1769,
      "step": 22530
    },
    {
      "epoch": 0.7713454296473811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.179,
      "step": 22531
    },
    {
      "epoch": 0.7713796644984594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1686,
      "step": 22532
    },
    {
      "epoch": 0.7714138993495379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1791,
      "step": 22533
    },
    {
      "epoch": 0.7714481342006162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1607,
      "step": 22534
    },
    {
      "epoch": 0.7714823690516946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1784,
      "step": 22535
    },
    {
      "epoch": 0.771516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.159,
      "step": 22536
    },
    {
      "epoch": 0.7715508387538514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1722,
      "step": 22537
    },
    {
      "epoch": 0.7715850736049298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1586,
      "step": 22538
    },
    {
      "epoch": 0.7716193084560082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1823,
      "step": 22539
    },
    {
      "epoch": 0.7716535433070866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1754,
      "step": 22540
    },
    {
      "epoch": 0.771687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1668,
      "step": 22541
    },
    {
      "epoch": 0.7717220130092434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1606,
      "step": 22542
    },
    {
      "epoch": 0.7717562478603218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1686,
      "step": 22543
    },
    {
      "epoch": 0.7717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1743,
      "step": 22544
    },
    {
      "epoch": 0.7718247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1753,
      "step": 22545
    },
    {
      "epoch": 0.771858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1716,
      "step": 22546
    },
    {
      "epoch": 0.7718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1621,
      "step": 22547
    },
    {
      "epoch": 0.7719274221157137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.166,
      "step": 22548
    },
    {
      "epoch": 0.7719616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1697,
      "step": 22549
    },
    {
      "epoch": 0.7719958918178705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1689,
      "step": 22550
    },
    {
      "epoch": 0.772030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1848,
      "step": 22551
    },
    {
      "epoch": 0.7720643615200274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1481,
      "step": 22552
    },
    {
      "epoch": 0.7720985963711058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1795,
      "step": 22553
    },
    {
      "epoch": 0.7721328312221842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1534,
      "step": 22554
    },
    {
      "epoch": 0.7721670660732626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1632,
      "step": 22555
    },
    {
      "epoch": 0.772201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1707,
      "step": 22556
    },
    {
      "epoch": 0.7722355357754194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1883,
      "step": 22557
    },
    {
      "epoch": 0.7722697706264978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1722,
      "step": 22558
    },
    {
      "epoch": 0.7723040054775762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1629,
      "step": 22559
    },
    {
      "epoch": 0.7723382403286546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.16,
      "step": 22560
    },
    {
      "epoch": 0.772372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1771,
      "step": 22561
    },
    {
      "epoch": 0.7724067100308114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1705,
      "step": 22562
    },
    {
      "epoch": 0.7724409448818897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1734,
      "step": 22563
    },
    {
      "epoch": 0.7724751797329682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1732,
      "step": 22564
    },
    {
      "epoch": 0.7725094145840465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1781,
      "step": 22565
    },
    {
      "epoch": 0.772543649435125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1508,
      "step": 22566
    },
    {
      "epoch": 0.7725778842862033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1635,
      "step": 22567
    },
    {
      "epoch": 0.7726121191372818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1829,
      "step": 22568
    },
    {
      "epoch": 0.7726463539883601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1623,
      "step": 22569
    },
    {
      "epoch": 0.7726805888394386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1559,
      "step": 22570
    },
    {
      "epoch": 0.7727148236905169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.171,
      "step": 22571
    },
    {
      "epoch": 0.7727490585415954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1807,
      "step": 22572
    },
    {
      "epoch": 0.7727832933926737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1666,
      "step": 22573
    },
    {
      "epoch": 0.7728175282437522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1972,
      "step": 22574
    },
    {
      "epoch": 0.7728517630948305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.169,
      "step": 22575
    },
    {
      "epoch": 0.772885997945909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1595,
      "step": 22576
    },
    {
      "epoch": 0.7729202327969873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.155,
      "step": 22577
    },
    {
      "epoch": 0.7729544676480657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1555,
      "step": 22578
    },
    {
      "epoch": 0.7729887024991441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1581,
      "step": 22579
    },
    {
      "epoch": 0.7730229373502225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1593,
      "step": 22580
    },
    {
      "epoch": 0.7730571722013009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1723,
      "step": 22581
    },
    {
      "epoch": 0.7730914070523793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1445,
      "step": 22582
    },
    {
      "epoch": 0.7731256419034577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1656,
      "step": 22583
    },
    {
      "epoch": 0.7731598767545361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1682,
      "step": 22584
    },
    {
      "epoch": 0.7731941116056145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1779,
      "step": 22585
    },
    {
      "epoch": 0.7732283464566929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1655,
      "step": 22586
    },
    {
      "epoch": 0.7732625813077713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1772,
      "step": 22587
    },
    {
      "epoch": 0.7732968161588497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.163,
      "step": 22588
    },
    {
      "epoch": 0.7733310510099282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1674,
      "step": 22589
    },
    {
      "epoch": 0.7733652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1625,
      "step": 22590
    },
    {
      "epoch": 0.773399520712085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1596,
      "step": 22591
    },
    {
      "epoch": 0.7734337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1542,
      "step": 22592
    },
    {
      "epoch": 0.7734679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1644,
      "step": 22593
    },
    {
      "epoch": 0.7735022252653201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1685,
      "step": 22594
    },
    {
      "epoch": 0.7735364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1516,
      "step": 22595
    },
    {
      "epoch": 0.7735706949674769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 22596
    },
    {
      "epoch": 0.7736049298185553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1727,
      "step": 22597
    },
    {
      "epoch": 0.7736391646696337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1652,
      "step": 22598
    },
    {
      "epoch": 0.7736733995207121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1845,
      "step": 22599
    },
    {
      "epoch": 0.7737076343717905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1654,
      "step": 22600
    },
    {
      "epoch": 0.7737418692228689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1757,
      "step": 22601
    },
    {
      "epoch": 0.7737761040739473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1854,
      "step": 22602
    },
    {
      "epoch": 0.7738103389250257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1768,
      "step": 22603
    },
    {
      "epoch": 0.773844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1548,
      "step": 22604
    },
    {
      "epoch": 0.7738788086271825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1517,
      "step": 22605
    },
    {
      "epoch": 0.7739130434782608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1566,
      "step": 22606
    },
    {
      "epoch": 0.7739472783293393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1628,
      "step": 22607
    },
    {
      "epoch": 0.7739815131804176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1674,
      "step": 22608
    },
    {
      "epoch": 0.7740157480314961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1694,
      "step": 22609
    },
    {
      "epoch": 0.7740499828825744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1597,
      "step": 22610
    },
    {
      "epoch": 0.7740842177336529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1704,
      "step": 22611
    },
    {
      "epoch": 0.7741184525847312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.174,
      "step": 22612
    },
    {
      "epoch": 0.7741526874358097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.171,
      "step": 22613
    },
    {
      "epoch": 0.774186922286888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1732,
      "step": 22614
    },
    {
      "epoch": 0.7742211571379665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1641,
      "step": 22615
    },
    {
      "epoch": 0.7742553919890448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1725,
      "step": 22616
    },
    {
      "epoch": 0.7742896268401233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1676,
      "step": 22617
    },
    {
      "epoch": 0.7743238616912016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1649,
      "step": 22618
    },
    {
      "epoch": 0.77435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1751,
      "step": 22619
    },
    {
      "epoch": 0.7743923313933584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.153,
      "step": 22620
    },
    {
      "epoch": 0.7744265662444368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.145,
      "step": 22621
    },
    {
      "epoch": 0.7744608010955152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1618,
      "step": 22622
    },
    {
      "epoch": 0.7744950359465936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1531,
      "step": 22623
    },
    {
      "epoch": 0.774529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1699,
      "step": 22624
    },
    {
      "epoch": 0.7745635056487504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1505,
      "step": 22625
    },
    {
      "epoch": 0.7745977404998289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.152,
      "step": 22626
    },
    {
      "epoch": 0.7746319753509072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1776,
      "step": 22627
    },
    {
      "epoch": 0.7746662102019857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1769,
      "step": 22628
    },
    {
      "epoch": 0.774700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1563,
      "step": 22629
    },
    {
      "epoch": 0.7747346799041425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1683,
      "step": 22630
    },
    {
      "epoch": 0.7747689147552208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1716,
      "step": 22631
    },
    {
      "epoch": 0.7748031496062993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1729,
      "step": 22632
    },
    {
      "epoch": 0.7748373844573776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1562,
      "step": 22633
    },
    {
      "epoch": 0.774871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1719,
      "step": 22634
    },
    {
      "epoch": 0.7749058541595344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1666,
      "step": 22635
    },
    {
      "epoch": 0.7749400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1688,
      "step": 22636
    },
    {
      "epoch": 0.7749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1863,
      "step": 22637
    },
    {
      "epoch": 0.7750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1707,
      "step": 22638
    },
    {
      "epoch": 0.775042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1831,
      "step": 22639
    },
    {
      "epoch": 0.7750770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1754,
      "step": 22640
    },
    {
      "epoch": 0.7751112632660048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1648,
      "step": 22641
    },
    {
      "epoch": 0.7751454981170832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1593,
      "step": 22642
    },
    {
      "epoch": 0.7751797329681616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1736,
      "step": 22643
    },
    {
      "epoch": 0.77521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1659,
      "step": 22644
    },
    {
      "epoch": 0.7752482026703184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1537,
      "step": 22645
    },
    {
      "epoch": 0.7752824375213968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1714,
      "step": 22646
    },
    {
      "epoch": 0.7753166723724751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1723,
      "step": 22647
    },
    {
      "epoch": 0.7753509072235536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1437,
      "step": 22648
    },
    {
      "epoch": 0.7753851420746319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1782,
      "step": 22649
    },
    {
      "epoch": 0.7754193769257104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1466,
      "step": 22650
    },
    {
      "epoch": 0.7754536117767887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1631,
      "step": 22651
    },
    {
      "epoch": 0.7754878466278672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1643,
      "step": 22652
    },
    {
      "epoch": 0.7755220814789455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1613,
      "step": 22653
    },
    {
      "epoch": 0.775556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.163,
      "step": 22654
    },
    {
      "epoch": 0.7755905511811023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1698,
      "step": 22655
    },
    {
      "epoch": 0.7756247860321808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.172,
      "step": 22656
    },
    {
      "epoch": 0.7756590208832591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1657,
      "step": 22657
    },
    {
      "epoch": 0.7756932557343376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1628,
      "step": 22658
    },
    {
      "epoch": 0.7757274905854159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1527,
      "step": 22659
    },
    {
      "epoch": 0.7757617254364944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.169,
      "step": 22660
    },
    {
      "epoch": 0.7757959602875727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.164,
      "step": 22661
    },
    {
      "epoch": 0.7758301951386511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.174,
      "step": 22662
    },
    {
      "epoch": 0.7758644299897296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1589,
      "step": 22663
    },
    {
      "epoch": 0.7758986648408079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1461,
      "step": 22664
    },
    {
      "epoch": 0.7759328996918864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1713,
      "step": 22665
    },
    {
      "epoch": 0.7759671345429647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1559,
      "step": 22666
    },
    {
      "epoch": 0.7760013693940432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1694,
      "step": 22667
    },
    {
      "epoch": 0.7760356042451215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1593,
      "step": 22668
    },
    {
      "epoch": 0.7760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1556,
      "step": 22669
    },
    {
      "epoch": 0.7761040739472783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1685,
      "step": 22670
    },
    {
      "epoch": 0.7761383087983568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1553,
      "step": 22671
    },
    {
      "epoch": 0.7761725436494351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1688,
      "step": 22672
    },
    {
      "epoch": 0.7762067785005136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1639,
      "step": 22673
    },
    {
      "epoch": 0.7762410133515919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1649,
      "step": 22674
    },
    {
      "epoch": 0.7762752482026704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1652,
      "step": 22675
    },
    {
      "epoch": 0.7763094830537487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1747,
      "step": 22676
    },
    {
      "epoch": 0.7763437179048271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1562,
      "step": 22677
    },
    {
      "epoch": 0.7763779527559055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1819,
      "step": 22678
    },
    {
      "epoch": 0.7764121876069839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1692,
      "step": 22679
    },
    {
      "epoch": 0.7764464224580623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1618,
      "step": 22680
    },
    {
      "epoch": 0.7764806573091407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1783,
      "step": 22681
    },
    {
      "epoch": 0.7765148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.189,
      "step": 22682
    },
    {
      "epoch": 0.7765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1754,
      "step": 22683
    },
    {
      "epoch": 0.7765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 22684
    },
    {
      "epoch": 0.7766175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1599,
      "step": 22685
    },
    {
      "epoch": 0.7766518315645327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1768,
      "step": 22686
    },
    {
      "epoch": 0.7766860664156111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1787,
      "step": 22687
    },
    {
      "epoch": 0.7767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1663,
      "step": 22688
    },
    {
      "epoch": 0.7767545361177679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1788,
      "step": 22689
    },
    {
      "epoch": 0.7767887709688462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1677,
      "step": 22690
    },
    {
      "epoch": 0.7768230058199247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1547,
      "step": 22691
    },
    {
      "epoch": 0.776857240671003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1638,
      "step": 22692
    },
    {
      "epoch": 0.7768914755220815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1478,
      "step": 22693
    },
    {
      "epoch": 0.7769257103731598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1649,
      "step": 22694
    },
    {
      "epoch": 0.7769599452242383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1617,
      "step": 22695
    },
    {
      "epoch": 0.7769941800753166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1729,
      "step": 22696
    },
    {
      "epoch": 0.7770284149263951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1567,
      "step": 22697
    },
    {
      "epoch": 0.7770626497774734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1572,
      "step": 22698
    },
    {
      "epoch": 0.7770968846285519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1653,
      "step": 22699
    },
    {
      "epoch": 0.7771311194796303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1722,
      "step": 22700
    },
    {
      "epoch": 0.7771653543307087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1727,
      "step": 22701
    },
    {
      "epoch": 0.7771995891817871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1542,
      "step": 22702
    },
    {
      "epoch": 0.7772338240328655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1675,
      "step": 22703
    },
    {
      "epoch": 0.7772680588839439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1709,
      "step": 22704
    },
    {
      "epoch": 0.7773022937350222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.185,
      "step": 22705
    },
    {
      "epoch": 0.7773365285861007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1686,
      "step": 22706
    },
    {
      "epoch": 0.777370763437179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1513,
      "step": 22707
    },
    {
      "epoch": 0.7774049982882575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1628,
      "step": 22708
    },
    {
      "epoch": 0.7774392331393358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1534,
      "step": 22709
    },
    {
      "epoch": 0.7774734679904143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1627,
      "step": 22710
    },
    {
      "epoch": 0.7775077028414926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1471,
      "step": 22711
    },
    {
      "epoch": 0.7775419376925711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.177,
      "step": 22712
    },
    {
      "epoch": 0.7775761725436494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1565,
      "step": 22713
    },
    {
      "epoch": 0.7776104073947279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1679,
      "step": 22714
    },
    {
      "epoch": 0.7776446422458062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1499,
      "step": 22715
    },
    {
      "epoch": 0.7776788770968847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1645,
      "step": 22716
    },
    {
      "epoch": 0.777713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1612,
      "step": 22717
    },
    {
      "epoch": 0.7777473467990415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1705,
      "step": 22718
    },
    {
      "epoch": 0.7777815816501198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1732,
      "step": 22719
    },
    {
      "epoch": 0.7778158165011982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 22720
    },
    {
      "epoch": 0.7778500513522766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.17,
      "step": 22721
    },
    {
      "epoch": 0.777884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1623,
      "step": 22722
    },
    {
      "epoch": 0.7779185210544334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.168,
      "step": 22723
    },
    {
      "epoch": 0.7779527559055118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1576,
      "step": 22724
    },
    {
      "epoch": 0.7779869907565902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1719,
      "step": 22725
    },
    {
      "epoch": 0.7780212256076686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1615,
      "step": 22726
    },
    {
      "epoch": 0.778055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1626,
      "step": 22727
    },
    {
      "epoch": 0.7780896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1808,
      "step": 22728
    },
    {
      "epoch": 0.7781239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1661,
      "step": 22729
    },
    {
      "epoch": 0.7781581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1654,
      "step": 22730
    },
    {
      "epoch": 0.7781923998630605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1624,
      "step": 22731
    },
    {
      "epoch": 0.778226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.148,
      "step": 22732
    },
    {
      "epoch": 0.7782608695652173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1469,
      "step": 22733
    },
    {
      "epoch": 0.7782951044162958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1739,
      "step": 22734
    },
    {
      "epoch": 0.7783293392673741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1707,
      "step": 22735
    },
    {
      "epoch": 0.7783635741184526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1737,
      "step": 22736
    },
    {
      "epoch": 0.778397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.149,
      "step": 22737
    },
    {
      "epoch": 0.7784320438206094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1618,
      "step": 22738
    },
    {
      "epoch": 0.7784662786716878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1509,
      "step": 22739
    },
    {
      "epoch": 0.7785005135227662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1724,
      "step": 22740
    },
    {
      "epoch": 0.7785347483738446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1721,
      "step": 22741
    },
    {
      "epoch": 0.778568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1616,
      "step": 22742
    },
    {
      "epoch": 0.7786032180760014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1599,
      "step": 22743
    },
    {
      "epoch": 0.7786374529270798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1632,
      "step": 22744
    },
    {
      "epoch": 0.7786716877781582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1584,
      "step": 22745
    },
    {
      "epoch": 0.7787059226292365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1716,
      "step": 22746
    },
    {
      "epoch": 0.778740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1682,
      "step": 22747
    },
    {
      "epoch": 0.7787743923313933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1615,
      "step": 22748
    },
    {
      "epoch": 0.7788086271824718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1723,
      "step": 22749
    },
    {
      "epoch": 0.7788428620335501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1705,
      "step": 22750
    },
    {
      "epoch": 0.7788770968846286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1724,
      "step": 22751
    },
    {
      "epoch": 0.7789113317357069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1566,
      "step": 22752
    },
    {
      "epoch": 0.7789455665867854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1679,
      "step": 22753
    },
    {
      "epoch": 0.7789798014378637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1686,
      "step": 22754
    },
    {
      "epoch": 0.7790140362889422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1616,
      "step": 22755
    },
    {
      "epoch": 0.7790482711400205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1539,
      "step": 22756
    },
    {
      "epoch": 0.779082505991099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.167,
      "step": 22757
    },
    {
      "epoch": 0.7791167408421773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1708,
      "step": 22758
    },
    {
      "epoch": 0.7791509756932558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1704,
      "step": 22759
    },
    {
      "epoch": 0.7791852105443341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1454,
      "step": 22760
    },
    {
      "epoch": 0.7792194453954125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1635,
      "step": 22761
    },
    {
      "epoch": 0.7792536802464909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1487,
      "step": 22762
    },
    {
      "epoch": 0.7792879150975693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1675,
      "step": 22763
    },
    {
      "epoch": 0.7793221499486477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1563,
      "step": 22764
    },
    {
      "epoch": 0.7793563847997261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1606,
      "step": 22765
    },
    {
      "epoch": 0.7793906196508045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1408,
      "step": 22766
    },
    {
      "epoch": 0.7794248545018829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1713,
      "step": 22767
    },
    {
      "epoch": 0.7794590893529613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1507,
      "step": 22768
    },
    {
      "epoch": 0.7794933242040397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1466,
      "step": 22769
    },
    {
      "epoch": 0.7795275590551181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1542,
      "step": 22770
    },
    {
      "epoch": 0.7795617939061965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1688,
      "step": 22771
    },
    {
      "epoch": 0.7795960287572749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1585,
      "step": 22772
    },
    {
      "epoch": 0.7796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.144,
      "step": 22773
    },
    {
      "epoch": 0.7796644984594318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.163,
      "step": 22774
    },
    {
      "epoch": 0.7796987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1538,
      "step": 22775
    },
    {
      "epoch": 0.7797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1627,
      "step": 22776
    },
    {
      "epoch": 0.7797672030126669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1736,
      "step": 22777
    },
    {
      "epoch": 0.7798014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.179,
      "step": 22778
    },
    {
      "epoch": 0.7798356727148237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1509,
      "step": 22779
    },
    {
      "epoch": 0.7798699075659021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1605,
      "step": 22780
    },
    {
      "epoch": 0.7799041424169805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1656,
      "step": 22781
    },
    {
      "epoch": 0.7799383772680589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1754,
      "step": 22782
    },
    {
      "epoch": 0.7799726121191373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1627,
      "step": 22783
    },
    {
      "epoch": 0.7800068469702157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1753,
      "step": 22784
    },
    {
      "epoch": 0.7800410818212941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1735,
      "step": 22785
    },
    {
      "epoch": 0.7800753166723725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1686,
      "step": 22786
    },
    {
      "epoch": 0.7801095515234509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1551,
      "step": 22787
    },
    {
      "epoch": 0.7801437863745293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1566,
      "step": 22788
    },
    {
      "epoch": 0.7801780212256076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1605,
      "step": 22789
    },
    {
      "epoch": 0.7802122560766861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1538,
      "step": 22790
    },
    {
      "epoch": 0.7802464909277644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1712,
      "step": 22791
    },
    {
      "epoch": 0.7802807257788429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1403,
      "step": 22792
    },
    {
      "epoch": 0.7803149606299212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1656,
      "step": 22793
    },
    {
      "epoch": 0.7803491954809997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1656,
      "step": 22794
    },
    {
      "epoch": 0.780383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1765,
      "step": 22795
    },
    {
      "epoch": 0.7804176651831565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1578,
      "step": 22796
    },
    {
      "epoch": 0.7804519000342348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1562,
      "step": 22797
    },
    {
      "epoch": 0.7804861348853133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1748,
      "step": 22798
    },
    {
      "epoch": 0.7805203697363916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1596,
      "step": 22799
    },
    {
      "epoch": 0.7805546045874701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1749,
      "step": 22800
    },
    {
      "epoch": 0.7805888394385484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1577,
      "step": 22801
    },
    {
      "epoch": 0.7806230742896269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1569,
      "step": 22802
    },
    {
      "epoch": 0.7806573091407052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1602,
      "step": 22803
    },
    {
      "epoch": 0.7806915439917836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1649,
      "step": 22804
    },
    {
      "epoch": 0.780725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1725,
      "step": 22805
    },
    {
      "epoch": 0.7807600136939404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1787,
      "step": 22806
    },
    {
      "epoch": 0.7807942485450188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1723,
      "step": 22807
    },
    {
      "epoch": 0.7808284833960972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1748,
      "step": 22808
    },
    {
      "epoch": 0.7808627182471757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.168,
      "step": 22809
    },
    {
      "epoch": 0.780896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1729,
      "step": 22810
    },
    {
      "epoch": 0.7809311879493325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1671,
      "step": 22811
    },
    {
      "epoch": 0.7809654228004108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.161,
      "step": 22812
    },
    {
      "epoch": 0.7809996576514893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.149,
      "step": 22813
    },
    {
      "epoch": 0.7810338925025676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1686,
      "step": 22814
    },
    {
      "epoch": 0.7810681273536461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1658,
      "step": 22815
    },
    {
      "epoch": 0.7811023622047244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1607,
      "step": 22816
    },
    {
      "epoch": 0.7811365970558029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1545,
      "step": 22817
    },
    {
      "epoch": 0.7811708319068812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1682,
      "step": 22818
    },
    {
      "epoch": 0.7812050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1599,
      "step": 22819
    },
    {
      "epoch": 0.781239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1486,
      "step": 22820
    },
    {
      "epoch": 0.7812735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1577,
      "step": 22821
    },
    {
      "epoch": 0.7813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1625,
      "step": 22822
    },
    {
      "epoch": 0.7813420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1528,
      "step": 22823
    },
    {
      "epoch": 0.7813762410133516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1372,
      "step": 22824
    },
    {
      "epoch": 0.78141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1554,
      "step": 22825
    },
    {
      "epoch": 0.7814447107155084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.132,
      "step": 22826
    },
    {
      "epoch": 0.7814789455665868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1707,
      "step": 22827
    },
    {
      "epoch": 0.7815131804176652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1625,
      "step": 22828
    },
    {
      "epoch": 0.7815474152687436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1434,
      "step": 22829
    },
    {
      "epoch": 0.781581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1677,
      "step": 22830
    },
    {
      "epoch": 0.7816158849709004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1794,
      "step": 22831
    },
    {
      "epoch": 0.7816501198219787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.157,
      "step": 22832
    },
    {
      "epoch": 0.7816843546730572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1681,
      "step": 22833
    },
    {
      "epoch": 0.7817185895241355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1747,
      "step": 22834
    },
    {
      "epoch": 0.781752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1628,
      "step": 22835
    },
    {
      "epoch": 0.7817870592262923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1605,
      "step": 22836
    },
    {
      "epoch": 0.7818212940773708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1778,
      "step": 22837
    },
    {
      "epoch": 0.7818555289284491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1837,
      "step": 22838
    },
    {
      "epoch": 0.7818897637795276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.139,
      "step": 22839
    },
    {
      "epoch": 0.7819239986306059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1645,
      "step": 22840
    },
    {
      "epoch": 0.7819582334816844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1468,
      "step": 22841
    },
    {
      "epoch": 0.7819924683327627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1651,
      "step": 22842
    },
    {
      "epoch": 0.7820267031838412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1599,
      "step": 22843
    },
    {
      "epoch": 0.7820609380349195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.162,
      "step": 22844
    },
    {
      "epoch": 0.782095172885998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1584,
      "step": 22845
    },
    {
      "epoch": 0.7821294077370764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1646,
      "step": 22846
    },
    {
      "epoch": 0.7821636425881547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1612,
      "step": 22847
    },
    {
      "epoch": 0.7821978774392332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1491,
      "step": 22848
    },
    {
      "epoch": 0.7822321122903115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1636,
      "step": 22849
    },
    {
      "epoch": 0.78226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1577,
      "step": 22850
    },
    {
      "epoch": 0.7823005819924683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1747,
      "step": 22851
    },
    {
      "epoch": 0.7823348168435468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1565,
      "step": 22852
    },
    {
      "epoch": 0.7823690516946251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1588,
      "step": 22853
    },
    {
      "epoch": 0.7824032865457036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1772,
      "step": 22854
    },
    {
      "epoch": 0.7824375213967819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.166,
      "step": 22855
    },
    {
      "epoch": 0.7824717562478604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1658,
      "step": 22856
    },
    {
      "epoch": 0.7825059910989387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1559,
      "step": 22857
    },
    {
      "epoch": 0.7825402259500172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1707,
      "step": 22858
    },
    {
      "epoch": 0.7825744608010955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1717,
      "step": 22859
    },
    {
      "epoch": 0.782608695652174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1717,
      "step": 22860
    },
    {
      "epoch": 0.7826429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.159,
      "step": 22861
    },
    {
      "epoch": 0.7826771653543307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1606,
      "step": 22862
    },
    {
      "epoch": 0.7827114002054091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1672,
      "step": 22863
    },
    {
      "epoch": 0.7827456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1691,
      "step": 22864
    },
    {
      "epoch": 0.7827798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1662,
      "step": 22865
    },
    {
      "epoch": 0.7828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1588,
      "step": 22866
    },
    {
      "epoch": 0.7828483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1501,
      "step": 22867
    },
    {
      "epoch": 0.7828825744608011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1646,
      "step": 22868
    },
    {
      "epoch": 0.7829168093118795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1549,
      "step": 22869
    },
    {
      "epoch": 0.7829510441629579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1524,
      "step": 22870
    },
    {
      "epoch": 0.7829852790140363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1664,
      "step": 22871
    },
    {
      "epoch": 0.7830195138651147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1531,
      "step": 22872
    },
    {
      "epoch": 0.783053748716193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1642,
      "step": 22873
    },
    {
      "epoch": 0.7830879835672715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1625,
      "step": 22874
    },
    {
      "epoch": 0.7831222184183498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1603,
      "step": 22875
    },
    {
      "epoch": 0.7831564532694283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1683,
      "step": 22876
    },
    {
      "epoch": 0.7831906881205066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1657,
      "step": 22877
    },
    {
      "epoch": 0.7832249229715851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1577,
      "step": 22878
    },
    {
      "epoch": 0.7832591578226634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.17,
      "step": 22879
    },
    {
      "epoch": 0.7832933926737419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1802,
      "step": 22880
    },
    {
      "epoch": 0.7833276275248202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1469,
      "step": 22881
    },
    {
      "epoch": 0.7833618623758987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.17,
      "step": 22882
    },
    {
      "epoch": 0.7833960972269771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1401,
      "step": 22883
    },
    {
      "epoch": 0.7834303320780555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1525,
      "step": 22884
    },
    {
      "epoch": 0.7834645669291339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1637,
      "step": 22885
    },
    {
      "epoch": 0.7834988017802123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1545,
      "step": 22886
    },
    {
      "epoch": 0.7835330366312907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1647,
      "step": 22887
    },
    {
      "epoch": 0.783567271482369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1602,
      "step": 22888
    },
    {
      "epoch": 0.7836015063334475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1542,
      "step": 22889
    },
    {
      "epoch": 0.7836357411845258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1699,
      "step": 22890
    },
    {
      "epoch": 0.7836699760356043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 22891
    },
    {
      "epoch": 0.7837042108866826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1626,
      "step": 22892
    },
    {
      "epoch": 0.7837384457377611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1552,
      "step": 22893
    },
    {
      "epoch": 0.7837726805888394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1658,
      "step": 22894
    },
    {
      "epoch": 0.7838069154399179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1639,
      "step": 22895
    },
    {
      "epoch": 0.7838411502909962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1619,
      "step": 22896
    },
    {
      "epoch": 0.7838753851420747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1737,
      "step": 22897
    },
    {
      "epoch": 0.783909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1635,
      "step": 22898
    },
    {
      "epoch": 0.7839438548442315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1631,
      "step": 22899
    },
    {
      "epoch": 0.7839780896953098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1692,
      "step": 22900
    },
    {
      "epoch": 0.7840123245463883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1635,
      "step": 22901
    },
    {
      "epoch": 0.7840465593974666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1736,
      "step": 22902
    },
    {
      "epoch": 0.784080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1691,
      "step": 22903
    },
    {
      "epoch": 0.7841150290996234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1667,
      "step": 22904
    },
    {
      "epoch": 0.7841492639507018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1526,
      "step": 22905
    },
    {
      "epoch": 0.7841834988017802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1604,
      "step": 22906
    },
    {
      "epoch": 0.7842177336528586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1533,
      "step": 22907
    },
    {
      "epoch": 0.784251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1689,
      "step": 22908
    },
    {
      "epoch": 0.7842862033550154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1651,
      "step": 22909
    },
    {
      "epoch": 0.7843204382060938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1633,
      "step": 22910
    },
    {
      "epoch": 0.7843546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1522,
      "step": 22911
    },
    {
      "epoch": 0.7843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1555,
      "step": 22912
    },
    {
      "epoch": 0.784423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1581,
      "step": 22913
    },
    {
      "epoch": 0.7844573776104073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1503,
      "step": 22914
    },
    {
      "epoch": 0.7844916124614858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1604,
      "step": 22915
    },
    {
      "epoch": 0.7845258473125641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1507,
      "step": 22916
    },
    {
      "epoch": 0.7845600821636426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1604,
      "step": 22917
    },
    {
      "epoch": 0.7845943170147209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1699,
      "step": 22918
    },
    {
      "epoch": 0.7846285518657994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1491,
      "step": 22919
    },
    {
      "epoch": 0.7846627867168778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1482,
      "step": 22920
    },
    {
      "epoch": 0.7846970215679562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.157,
      "step": 22921
    },
    {
      "epoch": 0.7847312564190346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1566,
      "step": 22922
    },
    {
      "epoch": 0.784765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1581,
      "step": 22923
    },
    {
      "epoch": 0.7847997261211914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1634,
      "step": 22924
    },
    {
      "epoch": 0.7848339609722698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1701,
      "step": 22925
    },
    {
      "epoch": 0.7848681958233482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1349,
      "step": 22926
    },
    {
      "epoch": 0.7849024306744266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1689,
      "step": 22927
    },
    {
      "epoch": 0.784936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1494,
      "step": 22928
    },
    {
      "epoch": 0.7849709003765833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1679,
      "step": 22929
    },
    {
      "epoch": 0.7850051352276618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1622,
      "step": 22930
    },
    {
      "epoch": 0.7850393700787401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1637,
      "step": 22931
    },
    {
      "epoch": 0.7850736049298186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1492,
      "step": 22932
    },
    {
      "epoch": 0.7851078397808969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1633,
      "step": 22933
    },
    {
      "epoch": 0.7851420746319754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1466,
      "step": 22934
    },
    {
      "epoch": 0.7851763094830537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1637,
      "step": 22935
    },
    {
      "epoch": 0.7852105443341322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1608,
      "step": 22936
    },
    {
      "epoch": 0.7852447791852105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1535,
      "step": 22937
    },
    {
      "epoch": 0.785279014036289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1661,
      "step": 22938
    },
    {
      "epoch": 0.7853132488873673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1673,
      "step": 22939
    },
    {
      "epoch": 0.7853474837384458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.135,
      "step": 22940
    },
    {
      "epoch": 0.7853817185895241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1733,
      "step": 22941
    },
    {
      "epoch": 0.7854159534406026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1552,
      "step": 22942
    },
    {
      "epoch": 0.7854501882916809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1634,
      "step": 22943
    },
    {
      "epoch": 0.7854844231427593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1457,
      "step": 22944
    },
    {
      "epoch": 0.7855186579938377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.145,
      "step": 22945
    },
    {
      "epoch": 0.7855528928449161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.165,
      "step": 22946
    },
    {
      "epoch": 0.7855871276959945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1641,
      "step": 22947
    },
    {
      "epoch": 0.7856213625470729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1535,
      "step": 22948
    },
    {
      "epoch": 0.7856555973981513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1728,
      "step": 22949
    },
    {
      "epoch": 0.7856898322492297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1639,
      "step": 22950
    },
    {
      "epoch": 0.7857240671003081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1538,
      "step": 22951
    },
    {
      "epoch": 0.7857583019513865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1666,
      "step": 22952
    },
    {
      "epoch": 0.7857925368024649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1638,
      "step": 22953
    },
    {
      "epoch": 0.7858267716535433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1689,
      "step": 22954
    },
    {
      "epoch": 0.7858610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1891,
      "step": 22955
    },
    {
      "epoch": 0.7858952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1651,
      "step": 22956
    },
    {
      "epoch": 0.7859294762067786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1619,
      "step": 22957
    },
    {
      "epoch": 0.7859637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1752,
      "step": 22958
    },
    {
      "epoch": 0.7859979459089353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1576,
      "step": 22959
    },
    {
      "epoch": 0.7860321807600137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1463,
      "step": 22960
    },
    {
      "epoch": 0.7860664156110921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1549,
      "step": 22961
    },
    {
      "epoch": 0.7861006504621705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1551,
      "step": 22962
    },
    {
      "epoch": 0.7861348853132489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1616,
      "step": 22963
    },
    {
      "epoch": 0.7861691201643273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.148,
      "step": 22964
    },
    {
      "epoch": 0.7862033550154057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.168,
      "step": 22965
    },
    {
      "epoch": 0.7862375898664841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1686,
      "step": 22966
    },
    {
      "epoch": 0.7862718247175625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1413,
      "step": 22967
    },
    {
      "epoch": 0.7863060595686409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1624,
      "step": 22968
    },
    {
      "epoch": 0.7863402944197193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1574,
      "step": 22969
    },
    {
      "epoch": 0.7863745292707977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1606,
      "step": 22970
    },
    {
      "epoch": 0.7864087641218761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1599,
      "step": 22971
    },
    {
      "epoch": 0.7864429989729544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1687,
      "step": 22972
    },
    {
      "epoch": 0.7864772338240329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1608,
      "step": 22973
    },
    {
      "epoch": 0.7865114686751112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1624,
      "step": 22974
    },
    {
      "epoch": 0.7865457035261897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1552,
      "step": 22975
    },
    {
      "epoch": 0.786579938377268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1593,
      "step": 22976
    },
    {
      "epoch": 0.7866141732283465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1754,
      "step": 22977
    },
    {
      "epoch": 0.7866484080794248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1633,
      "step": 22978
    },
    {
      "epoch": 0.7866826429305033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1579,
      "step": 22979
    },
    {
      "epoch": 0.7867168777815816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1542,
      "step": 22980
    },
    {
      "epoch": 0.7867511126326601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.167,
      "step": 22981
    },
    {
      "epoch": 0.7867853474837384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1776,
      "step": 22982
    },
    {
      "epoch": 0.7868195823348169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1576,
      "step": 22983
    },
    {
      "epoch": 0.7868538171858952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1631,
      "step": 22984
    },
    {
      "epoch": 0.7868880520369737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1746,
      "step": 22985
    },
    {
      "epoch": 0.786922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1716,
      "step": 22986
    },
    {
      "epoch": 0.7869565217391304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1601,
      "step": 22987
    },
    {
      "epoch": 0.7869907565902088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1514,
      "step": 22988
    },
    {
      "epoch": 0.7870249914412872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1646,
      "step": 22989
    },
    {
      "epoch": 0.7870592262923656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1595,
      "step": 22990
    },
    {
      "epoch": 0.787093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1707,
      "step": 22991
    },
    {
      "epoch": 0.7871276959945224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1652,
      "step": 22992
    },
    {
      "epoch": 0.7871619308456008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1737,
      "step": 22993
    },
    {
      "epoch": 0.7871961656966793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1492,
      "step": 22994
    },
    {
      "epoch": 0.7872304005477576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1565,
      "step": 22995
    },
    {
      "epoch": 0.7872646353988361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1828,
      "step": 22996
    },
    {
      "epoch": 0.7872988702499144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.173,
      "step": 22997
    },
    {
      "epoch": 0.7873331051009929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1597,
      "step": 22998
    },
    {
      "epoch": 0.7873673399520712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1606,
      "step": 22999
    },
    {
      "epoch": 0.7874015748031497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1663,
      "step": 23000
    },
    {
      "epoch": 0.787435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1525,
      "step": 23001
    },
    {
      "epoch": 0.7874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 23002
    },
    {
      "epoch": 0.7875042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1745,
      "step": 23003
    },
    {
      "epoch": 0.7875385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1664,
      "step": 23004
    },
    {
      "epoch": 0.7875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1578,
      "step": 23005
    },
    {
      "epoch": 0.78760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1643,
      "step": 23006
    },
    {
      "epoch": 0.7876412187606984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1765,
      "step": 23007
    },
    {
      "epoch": 0.7876754536117768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1564,
      "step": 23008
    },
    {
      "epoch": 0.7877096884628552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1615,
      "step": 23009
    },
    {
      "epoch": 0.7877439233139336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.161,
      "step": 23010
    },
    {
      "epoch": 0.787778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1514,
      "step": 23011
    },
    {
      "epoch": 0.7878123930160904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1505,
      "step": 23012
    },
    {
      "epoch": 0.7878466278671687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1616,
      "step": 23013
    },
    {
      "epoch": 0.7878808627182472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1676,
      "step": 23014
    },
    {
      "epoch": 0.7879150975693255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1668,
      "step": 23015
    },
    {
      "epoch": 0.787949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1666,
      "step": 23016
    },
    {
      "epoch": 0.7879835672714823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1623,
      "step": 23017
    },
    {
      "epoch": 0.7880178021225608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1686,
      "step": 23018
    },
    {
      "epoch": 0.7880520369736391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1564,
      "step": 23019
    },
    {
      "epoch": 0.7880862718247176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1373,
      "step": 23020
    },
    {
      "epoch": 0.7881205066757959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1472,
      "step": 23021
    },
    {
      "epoch": 0.7881547415268744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1551,
      "step": 23022
    },
    {
      "epoch": 0.7881889763779527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1538,
      "step": 23023
    },
    {
      "epoch": 0.7882232112290312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1717,
      "step": 23024
    },
    {
      "epoch": 0.7882574460801095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1473,
      "step": 23025
    },
    {
      "epoch": 0.788291680931188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1467,
      "step": 23026
    },
    {
      "epoch": 0.7883259157822663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1591,
      "step": 23027
    },
    {
      "epoch": 0.7883601506333447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1464,
      "step": 23028
    },
    {
      "epoch": 0.7883943854844231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.153,
      "step": 23029
    },
    {
      "epoch": 0.7884286203355015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1812,
      "step": 23030
    },
    {
      "epoch": 0.78846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1627,
      "step": 23031
    },
    {
      "epoch": 0.7884970900376583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1396,
      "step": 23032
    },
    {
      "epoch": 0.7885313248887368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.157,
      "step": 23033
    },
    {
      "epoch": 0.7885655597398151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1607,
      "step": 23034
    },
    {
      "epoch": 0.7885997945908936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1619,
      "step": 23035
    },
    {
      "epoch": 0.7886340294419719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1506,
      "step": 23036
    },
    {
      "epoch": 0.7886682642930504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.156,
      "step": 23037
    },
    {
      "epoch": 0.7887024991441287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1689,
      "step": 23038
    },
    {
      "epoch": 0.7887367339952072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1632,
      "step": 23039
    },
    {
      "epoch": 0.7887709688462855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1489,
      "step": 23040
    },
    {
      "epoch": 0.788805203697364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1573,
      "step": 23041
    },
    {
      "epoch": 0.7888394385484423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1811,
      "step": 23042
    },
    {
      "epoch": 0.7888736733995207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1687,
      "step": 23043
    },
    {
      "epoch": 0.7889079082505991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1572,
      "step": 23044
    },
    {
      "epoch": 0.7889421431016775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1521,
      "step": 23045
    },
    {
      "epoch": 0.7889763779527559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1551,
      "step": 23046
    },
    {
      "epoch": 0.7890106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1744,
      "step": 23047
    },
    {
      "epoch": 0.7890448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1597,
      "step": 23048
    },
    {
      "epoch": 0.7890790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1607,
      "step": 23049
    },
    {
      "epoch": 0.7891133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1676,
      "step": 23050
    },
    {
      "epoch": 0.7891475522081479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1716,
      "step": 23051
    },
    {
      "epoch": 0.7891817870592263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1499,
      "step": 23052
    },
    {
      "epoch": 0.7892160219103047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1454,
      "step": 23053
    },
    {
      "epoch": 0.789250256761383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1673,
      "step": 23054
    },
    {
      "epoch": 0.7892844916124615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1585,
      "step": 23055
    },
    {
      "epoch": 0.7893187264635398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1678,
      "step": 23056
    },
    {
      "epoch": 0.7893529613146183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.146,
      "step": 23057
    },
    {
      "epoch": 0.7893871961656966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1713,
      "step": 23058
    },
    {
      "epoch": 0.7894214310167751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1607,
      "step": 23059
    },
    {
      "epoch": 0.7894556658678534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1462,
      "step": 23060
    },
    {
      "epoch": 0.7894899007189319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1708,
      "step": 23061
    },
    {
      "epoch": 0.7895241355700102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1697,
      "step": 23062
    },
    {
      "epoch": 0.7895583704210887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1579,
      "step": 23063
    },
    {
      "epoch": 0.789592605272167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1667,
      "step": 23064
    },
    {
      "epoch": 0.7896268401232455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1672,
      "step": 23065
    },
    {
      "epoch": 0.7896610749743238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1635,
      "step": 23066
    },
    {
      "epoch": 0.7896953098254023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1629,
      "step": 23067
    },
    {
      "epoch": 0.7897295446764807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1565,
      "step": 23068
    },
    {
      "epoch": 0.789763779527559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1807,
      "step": 23069
    },
    {
      "epoch": 0.7897980143786375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1466,
      "step": 23070
    },
    {
      "epoch": 0.7898322492297158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1557,
      "step": 23071
    },
    {
      "epoch": 0.7898664840807943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1646,
      "step": 23072
    },
    {
      "epoch": 0.7899007189318726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 23073
    },
    {
      "epoch": 0.7899349537829511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1454,
      "step": 23074
    },
    {
      "epoch": 0.7899691886340294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1503,
      "step": 23075
    },
    {
      "epoch": 0.7900034234851079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1707,
      "step": 23076
    },
    {
      "epoch": 0.7900376583361862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1523,
      "step": 23077
    },
    {
      "epoch": 0.7900718931872647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1493,
      "step": 23078
    },
    {
      "epoch": 0.790106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 23079
    },
    {
      "epoch": 0.7901403628894215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1717,
      "step": 23080
    },
    {
      "epoch": 0.7901745977404998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1439,
      "step": 23081
    },
    {
      "epoch": 0.7902088325915783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1517,
      "step": 23082
    },
    {
      "epoch": 0.7902430674426566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1549,
      "step": 23083
    },
    {
      "epoch": 0.790277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1441,
      "step": 23084
    },
    {
      "epoch": 0.7903115371448134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1607,
      "step": 23085
    },
    {
      "epoch": 0.7903457719958918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1496,
      "step": 23086
    },
    {
      "epoch": 0.7903800068469702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1726,
      "step": 23087
    },
    {
      "epoch": 0.7904142416980486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1568,
      "step": 23088
    },
    {
      "epoch": 0.790448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1492,
      "step": 23089
    },
    {
      "epoch": 0.7904827114002054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1591,
      "step": 23090
    },
    {
      "epoch": 0.7905169462512838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.165,
      "step": 23091
    },
    {
      "epoch": 0.7905511811023622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1573,
      "step": 23092
    },
    {
      "epoch": 0.7905854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1505,
      "step": 23093
    },
    {
      "epoch": 0.790619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1416,
      "step": 23094
    },
    {
      "epoch": 0.7906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1478,
      "step": 23095
    },
    {
      "epoch": 0.7906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1624,
      "step": 23096
    },
    {
      "epoch": 0.7907223553577541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1603,
      "step": 23097
    },
    {
      "epoch": 0.7907565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1526,
      "step": 23098
    },
    {
      "epoch": 0.7907908250599109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1612,
      "step": 23099
    },
    {
      "epoch": 0.7908250599109894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1545,
      "step": 23100
    },
    {
      "epoch": 0.7908592947620677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1695,
      "step": 23101
    },
    {
      "epoch": 0.7908935296131462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1693,
      "step": 23102
    },
    {
      "epoch": 0.7909277644642245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1632,
      "step": 23103
    },
    {
      "epoch": 0.790961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1531,
      "step": 23104
    },
    {
      "epoch": 0.7909962341663814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1711,
      "step": 23105
    },
    {
      "epoch": 0.7910304690174598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1537,
      "step": 23106
    },
    {
      "epoch": 0.7910647038685382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1468,
      "step": 23107
    },
    {
      "epoch": 0.7910989387196166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1522,
      "step": 23108
    },
    {
      "epoch": 0.791133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1535,
      "step": 23109
    },
    {
      "epoch": 0.7911674084217734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1491,
      "step": 23110
    },
    {
      "epoch": 0.7912016432728518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1563,
      "step": 23111
    },
    {
      "epoch": 0.7912358781239301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.156,
      "step": 23112
    },
    {
      "epoch": 0.7912701129750086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1657,
      "step": 23113
    },
    {
      "epoch": 0.7913043478260869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.159,
      "step": 23114
    },
    {
      "epoch": 0.7913385826771654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1614,
      "step": 23115
    },
    {
      "epoch": 0.7913728175282437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.163,
      "step": 23116
    },
    {
      "epoch": 0.7914070523793222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1401,
      "step": 23117
    },
    {
      "epoch": 0.7914412872304005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1719,
      "step": 23118
    },
    {
      "epoch": 0.791475522081479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1575,
      "step": 23119
    },
    {
      "epoch": 0.7915097569325573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1597,
      "step": 23120
    },
    {
      "epoch": 0.7915439917836358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.139,
      "step": 23121
    },
    {
      "epoch": 0.7915782266347141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1614,
      "step": 23122
    },
    {
      "epoch": 0.7916124614857926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1578,
      "step": 23123
    },
    {
      "epoch": 0.7916466963368709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1592,
      "step": 23124
    },
    {
      "epoch": 0.7916809311879494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1457,
      "step": 23125
    },
    {
      "epoch": 0.7917151660390277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1553,
      "step": 23126
    },
    {
      "epoch": 0.7917494008901061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.174,
      "step": 23127
    },
    {
      "epoch": 0.7917836357411845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1607,
      "step": 23128
    },
    {
      "epoch": 0.7918178705922629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1585,
      "step": 23129
    },
    {
      "epoch": 0.7918521054433413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1604,
      "step": 23130
    },
    {
      "epoch": 0.7918863402944197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1599,
      "step": 23131
    },
    {
      "epoch": 0.7919205751454981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1722,
      "step": 23132
    },
    {
      "epoch": 0.7919548099965765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.157,
      "step": 23133
    },
    {
      "epoch": 0.7919890448476549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1627,
      "step": 23134
    },
    {
      "epoch": 0.7920232796987333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.141,
      "step": 23135
    },
    {
      "epoch": 0.7920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.158,
      "step": 23136
    },
    {
      "epoch": 0.7920917494008901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1626,
      "step": 23137
    },
    {
      "epoch": 0.7921259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1494,
      "step": 23138
    },
    {
      "epoch": 0.7921602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.168,
      "step": 23139
    },
    {
      "epoch": 0.7921944539541254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1626,
      "step": 23140
    },
    {
      "epoch": 0.7922286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1591,
      "step": 23141
    },
    {
      "epoch": 0.7922629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1584,
      "step": 23142
    },
    {
      "epoch": 0.7922971585073605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.155,
      "step": 23143
    },
    {
      "epoch": 0.7923313933584389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1584,
      "step": 23144
    },
    {
      "epoch": 0.7923656282095173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.169,
      "step": 23145
    },
    {
      "epoch": 0.7923998630605957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1635,
      "step": 23146
    },
    {
      "epoch": 0.7924340979116741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1718,
      "step": 23147
    },
    {
      "epoch": 0.7924683327627525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1536,
      "step": 23148
    },
    {
      "epoch": 0.7925025676138309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1463,
      "step": 23149
    },
    {
      "epoch": 0.7925368024649093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1509,
      "step": 23150
    },
    {
      "epoch": 0.7925710373159877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1555,
      "step": 23151
    },
    {
      "epoch": 0.7926052721670661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1625,
      "step": 23152
    },
    {
      "epoch": 0.7926395070181445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1632,
      "step": 23153
    },
    {
      "epoch": 0.7926737418692229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1575,
      "step": 23154
    },
    {
      "epoch": 0.7927079767203012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1678,
      "step": 23155
    },
    {
      "epoch": 0.7927422115713797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.153,
      "step": 23156
    },
    {
      "epoch": 0.792776446422458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1626,
      "step": 23157
    },
    {
      "epoch": 0.7928106812735365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1494,
      "step": 23158
    },
    {
      "epoch": 0.7928449161246148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.141,
      "step": 23159
    },
    {
      "epoch": 0.7928791509756933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1727,
      "step": 23160
    },
    {
      "epoch": 0.7929133858267716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1717,
      "step": 23161
    },
    {
      "epoch": 0.7929476206778501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1557,
      "step": 23162
    },
    {
      "epoch": 0.7929818555289284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1713,
      "step": 23163
    },
    {
      "epoch": 0.7930160903800069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1467,
      "step": 23164
    },
    {
      "epoch": 0.7930503252310852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1529,
      "step": 23165
    },
    {
      "epoch": 0.7930845600821637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1622,
      "step": 23166
    },
    {
      "epoch": 0.793118794933242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1777,
      "step": 23167
    },
    {
      "epoch": 0.7931530297843205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1699,
      "step": 23168
    },
    {
      "epoch": 0.7931872646353988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1617,
      "step": 23169
    },
    {
      "epoch": 0.7932214994864772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1501,
      "step": 23170
    },
    {
      "epoch": 0.7932557343375556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1505,
      "step": 23171
    },
    {
      "epoch": 0.793289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1544,
      "step": 23172
    },
    {
      "epoch": 0.7933242040397124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1436,
      "step": 23173
    },
    {
      "epoch": 0.7933584388907908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1647,
      "step": 23174
    },
    {
      "epoch": 0.7933926737418692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1472,
      "step": 23175
    },
    {
      "epoch": 0.7934269085929476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1469,
      "step": 23176
    },
    {
      "epoch": 0.7934611434440261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.167,
      "step": 23177
    },
    {
      "epoch": 0.7934953782951044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1636,
      "step": 23178
    },
    {
      "epoch": 0.7935296131461829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1647,
      "step": 23179
    },
    {
      "epoch": 0.7935638479972612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1828,
      "step": 23180
    },
    {
      "epoch": 0.7935980828483397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1405,
      "step": 23181
    },
    {
      "epoch": 0.793632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1551,
      "step": 23182
    },
    {
      "epoch": 0.7936665525504965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1628,
      "step": 23183
    },
    {
      "epoch": 0.7937007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1723,
      "step": 23184
    },
    {
      "epoch": 0.7937350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1683,
      "step": 23185
    },
    {
      "epoch": 0.7937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1663,
      "step": 23186
    },
    {
      "epoch": 0.79380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.168,
      "step": 23187
    },
    {
      "epoch": 0.7938377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1531,
      "step": 23188
    },
    {
      "epoch": 0.7938719616569668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1548,
      "step": 23189
    },
    {
      "epoch": 0.7939061965080452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1662,
      "step": 23190
    },
    {
      "epoch": 0.7939404313591236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.159,
      "step": 23191
    },
    {
      "epoch": 0.793974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1631,
      "step": 23192
    },
    {
      "epoch": 0.7940089010612804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1565,
      "step": 23193
    },
    {
      "epoch": 0.7940431359123588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1497,
      "step": 23194
    },
    {
      "epoch": 0.7940773707634372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.16,
      "step": 23195
    },
    {
      "epoch": 0.7941116056145155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1664,
      "step": 23196
    },
    {
      "epoch": 0.794145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1384,
      "step": 23197
    },
    {
      "epoch": 0.7941800753166723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1704,
      "step": 23198
    },
    {
      "epoch": 0.7942143101677508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.151,
      "step": 23199
    },
    {
      "epoch": 0.7942485450188291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1521,
      "step": 23200
    },
    {
      "epoch": 0.7942827798699076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1612,
      "step": 23201
    },
    {
      "epoch": 0.7943170147209859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1508,
      "step": 23202
    },
    {
      "epoch": 0.7943512495720644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1498,
      "step": 23203
    },
    {
      "epoch": 0.7943854844231427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1586,
      "step": 23204
    },
    {
      "epoch": 0.7944197192742212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.155,
      "step": 23205
    },
    {
      "epoch": 0.7944539541252995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1576,
      "step": 23206
    },
    {
      "epoch": 0.794488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1547,
      "step": 23207
    },
    {
      "epoch": 0.7945224238274563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1747,
      "step": 23208
    },
    {
      "epoch": 0.7945566586785348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1646,
      "step": 23209
    },
    {
      "epoch": 0.7945908935296131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1657,
      "step": 23210
    },
    {
      "epoch": 0.7946251283806915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1413,
      "step": 23211
    },
    {
      "epoch": 0.7946593632317699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1551,
      "step": 23212
    },
    {
      "epoch": 0.7946935980828483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1698,
      "step": 23213
    },
    {
      "epoch": 0.7947278329339268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1516,
      "step": 23214
    },
    {
      "epoch": 0.7947620677850051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1578,
      "step": 23215
    },
    {
      "epoch": 0.7947963026360836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1611,
      "step": 23216
    },
    {
      "epoch": 0.7948305374871619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1499,
      "step": 23217
    },
    {
      "epoch": 0.7948647723382404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1544,
      "step": 23218
    },
    {
      "epoch": 0.7948990071893187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1608,
      "step": 23219
    },
    {
      "epoch": 0.7949332420403972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1591,
      "step": 23220
    },
    {
      "epoch": 0.7949674768914755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.161,
      "step": 23221
    },
    {
      "epoch": 0.795001711742554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1616,
      "step": 23222
    },
    {
      "epoch": 0.7950359465936323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1534,
      "step": 23223
    },
    {
      "epoch": 0.7950701814447108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1509,
      "step": 23224
    },
    {
      "epoch": 0.7951044162957891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1498,
      "step": 23225
    },
    {
      "epoch": 0.7951386511468675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1676,
      "step": 23226
    },
    {
      "epoch": 0.7951728859979459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 23227
    },
    {
      "epoch": 0.7952071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1506,
      "step": 23228
    },
    {
      "epoch": 0.7952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1708,
      "step": 23229
    },
    {
      "epoch": 0.7952755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1625,
      "step": 23230
    },
    {
      "epoch": 0.7953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1417,
      "step": 23231
    },
    {
      "epoch": 0.7953440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1659,
      "step": 23232
    },
    {
      "epoch": 0.7953782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1554,
      "step": 23233
    },
    {
      "epoch": 0.7954125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1487,
      "step": 23234
    },
    {
      "epoch": 0.7954467648065731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1656,
      "step": 23235
    },
    {
      "epoch": 0.7954809996576515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1516,
      "step": 23236
    },
    {
      "epoch": 0.7955152345087299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1582,
      "step": 23237
    },
    {
      "epoch": 0.7955494693598083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1438,
      "step": 23238
    },
    {
      "epoch": 0.7955837042108866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1471,
      "step": 23239
    },
    {
      "epoch": 0.7956179390619651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1574,
      "step": 23240
    },
    {
      "epoch": 0.7956521739130434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1537,
      "step": 23241
    },
    {
      "epoch": 0.7956864087641219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1549,
      "step": 23242
    },
    {
      "epoch": 0.7957206436152002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1511,
      "step": 23243
    },
    {
      "epoch": 0.7957548784662787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1376,
      "step": 23244
    },
    {
      "epoch": 0.795789113317357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1477,
      "step": 23245
    },
    {
      "epoch": 0.7958233481684355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1547,
      "step": 23246
    },
    {
      "epoch": 0.7958575830195138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1545,
      "step": 23247
    },
    {
      "epoch": 0.7958918178705923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1561,
      "step": 23248
    },
    {
      "epoch": 0.7959260527216706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1411,
      "step": 23249
    },
    {
      "epoch": 0.7959602875727491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1517,
      "step": 23250
    },
    {
      "epoch": 0.7959945224238275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1589,
      "step": 23251
    },
    {
      "epoch": 0.7960287572749059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1551,
      "step": 23252
    },
    {
      "epoch": 0.7960629921259843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1747,
      "step": 23253
    },
    {
      "epoch": 0.7960972269770626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1596,
      "step": 23254
    },
    {
      "epoch": 0.7961314618281411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1443,
      "step": 23255
    },
    {
      "epoch": 0.7961656966792194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1503,
      "step": 23256
    },
    {
      "epoch": 0.7961999315302979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1494,
      "step": 23257
    },
    {
      "epoch": 0.7962341663813762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.152,
      "step": 23258
    },
    {
      "epoch": 0.7962684012324547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1461,
      "step": 23259
    },
    {
      "epoch": 0.796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1516,
      "step": 23260
    },
    {
      "epoch": 0.7963368709346115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1714,
      "step": 23261
    },
    {
      "epoch": 0.7963711057856898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1476,
      "step": 23262
    },
    {
      "epoch": 0.7964053406367683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1661,
      "step": 23263
    },
    {
      "epoch": 0.7964395754878466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1495,
      "step": 23264
    },
    {
      "epoch": 0.7964738103389251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1306,
      "step": 23265
    },
    {
      "epoch": 0.7965080451900034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.149,
      "step": 23266
    },
    {
      "epoch": 0.7965422800410819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1466,
      "step": 23267
    },
    {
      "epoch": 0.7965765148921602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.153,
      "step": 23268
    },
    {
      "epoch": 0.7966107497432386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1467,
      "step": 23269
    },
    {
      "epoch": 0.796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1469,
      "step": 23270
    },
    {
      "epoch": 0.7966792194453954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1487,
      "step": 23271
    },
    {
      "epoch": 0.7967134542964738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1555,
      "step": 23272
    },
    {
      "epoch": 0.7967476891475522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1405,
      "step": 23273
    },
    {
      "epoch": 0.7967819239986306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1282,
      "step": 23274
    },
    {
      "epoch": 0.796816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1451,
      "step": 23275
    },
    {
      "epoch": 0.7968503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 23276
    },
    {
      "epoch": 0.7968846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1665,
      "step": 23277
    },
    {
      "epoch": 0.7969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.154,
      "step": 23278
    },
    {
      "epoch": 0.7969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1606,
      "step": 23279
    },
    {
      "epoch": 0.796987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.154,
      "step": 23280
    },
    {
      "epoch": 0.7970215679561794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.155,
      "step": 23281
    },
    {
      "epoch": 0.7970558028072577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1393,
      "step": 23282
    },
    {
      "epoch": 0.7970900376583362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1446,
      "step": 23283
    },
    {
      "epoch": 0.7971242725094145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1723,
      "step": 23284
    },
    {
      "epoch": 0.797158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1494,
      "step": 23285
    },
    {
      "epoch": 0.7971927422115713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1544,
      "step": 23286
    },
    {
      "epoch": 0.7972269770626498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1467,
      "step": 23287
    },
    {
      "epoch": 0.7972612119137282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1558,
      "step": 23288
    },
    {
      "epoch": 0.7972954467648066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1496,
      "step": 23289
    },
    {
      "epoch": 0.797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1723,
      "step": 23290
    },
    {
      "epoch": 0.7973639164669634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1696,
      "step": 23291
    },
    {
      "epoch": 0.7973981513180418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1712,
      "step": 23292
    },
    {
      "epoch": 0.7974323861691202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1516,
      "step": 23293
    },
    {
      "epoch": 0.7974666210201986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1381,
      "step": 23294
    },
    {
      "epoch": 0.797500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1385,
      "step": 23295
    },
    {
      "epoch": 0.7975350907223554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1545,
      "step": 23296
    },
    {
      "epoch": 0.7975693255734337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1543,
      "step": 23297
    },
    {
      "epoch": 0.7976035604245122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1376,
      "step": 23298
    },
    {
      "epoch": 0.7976377952755905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1798,
      "step": 23299
    },
    {
      "epoch": 0.797672030126669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1384,
      "step": 23300
    },
    {
      "epoch": 0.7977062649777473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1616,
      "step": 23301
    },
    {
      "epoch": 0.7977404998288258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1685,
      "step": 23302
    },
    {
      "epoch": 0.7977747346799041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1464,
      "step": 23303
    },
    {
      "epoch": 0.7978089695309826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1413,
      "step": 23304
    },
    {
      "epoch": 0.7978432043820609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1453,
      "step": 23305
    },
    {
      "epoch": 0.7978774392331394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1423,
      "step": 23306
    },
    {
      "epoch": 0.7979116740842177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.153,
      "step": 23307
    },
    {
      "epoch": 0.7979459089352962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1369,
      "step": 23308
    },
    {
      "epoch": 0.7979801437863745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1634,
      "step": 23309
    },
    {
      "epoch": 0.798014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1439,
      "step": 23310
    },
    {
      "epoch": 0.7980486134885313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1568,
      "step": 23311
    },
    {
      "epoch": 0.7980828483396097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1448,
      "step": 23312
    },
    {
      "epoch": 0.7981170831906881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.153,
      "step": 23313
    },
    {
      "epoch": 0.7981513180417665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1713,
      "step": 23314
    },
    {
      "epoch": 0.7981855528928449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1362,
      "step": 23315
    },
    {
      "epoch": 0.7982197877439233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1727,
      "step": 23316
    },
    {
      "epoch": 0.7982540225950017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1592,
      "step": 23317
    },
    {
      "epoch": 0.7982882574460801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1471,
      "step": 23318
    },
    {
      "epoch": 0.7983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1364,
      "step": 23319
    },
    {
      "epoch": 0.7983567271482369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1487,
      "step": 23320
    },
    {
      "epoch": 0.7983909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1487,
      "step": 23321
    },
    {
      "epoch": 0.7984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1722,
      "step": 23322
    },
    {
      "epoch": 0.798459431701472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1516,
      "step": 23323
    },
    {
      "epoch": 0.7984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1407,
      "step": 23324
    },
    {
      "epoch": 0.798527901403629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1456,
      "step": 23325
    },
    {
      "epoch": 0.7985621362547073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1491,
      "step": 23326
    },
    {
      "epoch": 0.7985963711057857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1583,
      "step": 23327
    },
    {
      "epoch": 0.7986306059568641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1466,
      "step": 23328
    },
    {
      "epoch": 0.7986648408079425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1498,
      "step": 23329
    },
    {
      "epoch": 0.7986990756590209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1642,
      "step": 23330
    },
    {
      "epoch": 0.7987333105100993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1495,
      "step": 23331
    },
    {
      "epoch": 0.7987675453611777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1615,
      "step": 23332
    },
    {
      "epoch": 0.7988017802122561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1492,
      "step": 23333
    },
    {
      "epoch": 0.7988360150633345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1722,
      "step": 23334
    },
    {
      "epoch": 0.7988702499144129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1473,
      "step": 23335
    },
    {
      "epoch": 0.7989044847654913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 23336
    },
    {
      "epoch": 0.7989387196165697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1613,
      "step": 23337
    },
    {
      "epoch": 0.798972954467648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1714,
      "step": 23338
    },
    {
      "epoch": 0.7990071893187265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1615,
      "step": 23339
    },
    {
      "epoch": 0.7990414241698048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1496,
      "step": 23340
    },
    {
      "epoch": 0.7990756590208833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1488,
      "step": 23341
    },
    {
      "epoch": 0.7991098938719616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1514,
      "step": 23342
    },
    {
      "epoch": 0.7991441287230401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.16,
      "step": 23343
    },
    {
      "epoch": 0.7991783635741184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1502,
      "step": 23344
    },
    {
      "epoch": 0.7992125984251969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1478,
      "step": 23345
    },
    {
      "epoch": 0.7992468332762752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1523,
      "step": 23346
    },
    {
      "epoch": 0.7992810681273537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.133,
      "step": 23347
    },
    {
      "epoch": 0.799315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1405,
      "step": 23348
    },
    {
      "epoch": 0.7993495378295105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1592,
      "step": 23349
    },
    {
      "epoch": 0.7993837726805888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1314,
      "step": 23350
    },
    {
      "epoch": 0.7994180075316673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1564,
      "step": 23351
    },
    {
      "epoch": 0.7994522423827456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1655,
      "step": 23352
    },
    {
      "epoch": 0.799486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1498,
      "step": 23353
    },
    {
      "epoch": 0.7995207120849024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1413,
      "step": 23354
    },
    {
      "epoch": 0.7995549469359808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1456,
      "step": 23355
    },
    {
      "epoch": 0.7995891817870592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1487,
      "step": 23356
    },
    {
      "epoch": 0.7996234166381376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1572,
      "step": 23357
    },
    {
      "epoch": 0.799657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1644,
      "step": 23358
    },
    {
      "epoch": 0.7996918863402944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1711,
      "step": 23359
    },
    {
      "epoch": 0.7997261211913728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1659,
      "step": 23360
    },
    {
      "epoch": 0.7997603560424512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.135,
      "step": 23361
    },
    {
      "epoch": 0.7997945908935297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1584,
      "step": 23362
    },
    {
      "epoch": 0.799828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1497,
      "step": 23363
    },
    {
      "epoch": 0.7998630605956865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1619,
      "step": 23364
    },
    {
      "epoch": 0.7998972954467648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1529,
      "step": 23365
    },
    {
      "epoch": 0.7999315302978433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1491,
      "step": 23366
    },
    {
      "epoch": 0.7999657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1334,
      "step": 23367
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1575,
      "step": 23368
    },
    {
      "epoch": 0.8000342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1539,
      "step": 23369
    },
    {
      "epoch": 0.8000684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.144,
      "step": 23370
    },
    {
      "epoch": 0.8001027045532352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1525,
      "step": 23371
    },
    {
      "epoch": 0.8001369394043136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1535,
      "step": 23372
    },
    {
      "epoch": 0.800171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1522,
      "step": 23373
    },
    {
      "epoch": 0.8002054091064704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1395,
      "step": 23374
    },
    {
      "epoch": 0.8002396439575488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1542,
      "step": 23375
    },
    {
      "epoch": 0.8002738788086272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1638,
      "step": 23376
    },
    {
      "epoch": 0.8003081136597056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1489,
      "step": 23377
    },
    {
      "epoch": 0.800342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1467,
      "step": 23378
    },
    {
      "epoch": 0.8003765833618623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1479,
      "step": 23379
    },
    {
      "epoch": 0.8004108182129408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1515,
      "step": 23380
    },
    {
      "epoch": 0.8004450530640191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.165,
      "step": 23381
    },
    {
      "epoch": 0.8004792879150976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.167,
      "step": 23382
    },
    {
      "epoch": 0.8005135227661759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1517,
      "step": 23383
    },
    {
      "epoch": 0.8005477576172544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1567,
      "step": 23384
    },
    {
      "epoch": 0.8005819924683327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1361,
      "step": 23385
    },
    {
      "epoch": 0.8006162273194112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1465,
      "step": 23386
    },
    {
      "epoch": 0.8006504621704895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1377,
      "step": 23387
    },
    {
      "epoch": 0.800684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1416,
      "step": 23388
    },
    {
      "epoch": 0.8007189318726463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1596,
      "step": 23389
    },
    {
      "epoch": 0.8007531667237248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1424,
      "step": 23390
    },
    {
      "epoch": 0.8007874015748031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1637,
      "step": 23391
    },
    {
      "epoch": 0.8008216364258816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1684,
      "step": 23392
    },
    {
      "epoch": 0.8008558712769599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1401,
      "step": 23393
    },
    {
      "epoch": 0.8008901061280383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.157,
      "step": 23394
    },
    {
      "epoch": 0.8009243409791167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.151,
      "step": 23395
    },
    {
      "epoch": 0.8009585758301951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1613,
      "step": 23396
    },
    {
      "epoch": 0.8009928106812735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1398,
      "step": 23397
    },
    {
      "epoch": 0.8010270455323519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1632,
      "step": 23398
    },
    {
      "epoch": 0.8010612803834304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1519,
      "step": 23399
    },
    {
      "epoch": 0.8010955152345087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1656,
      "step": 23400
    },
    {
      "epoch": 0.8011297500855872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1525,
      "step": 23401
    },
    {
      "epoch": 0.8011639849366655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1597,
      "step": 23402
    },
    {
      "epoch": 0.801198219787744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1322,
      "step": 23403
    },
    {
      "epoch": 0.8012324546388223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1368,
      "step": 23404
    },
    {
      "epoch": 0.8012666894899008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1515,
      "step": 23405
    },
    {
      "epoch": 0.8013009243409791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1474,
      "step": 23406
    },
    {
      "epoch": 0.8013351591920576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1516,
      "step": 23407
    },
    {
      "epoch": 0.8013693940431359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1361,
      "step": 23408
    },
    {
      "epoch": 0.8014036288942143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1457,
      "step": 23409
    },
    {
      "epoch": 0.8014378637452927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1742,
      "step": 23410
    },
    {
      "epoch": 0.8014720985963711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1588,
      "step": 23411
    },
    {
      "epoch": 0.8015063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1397,
      "step": 23412
    },
    {
      "epoch": 0.8015405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1591,
      "step": 23413
    },
    {
      "epoch": 0.8015748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.151,
      "step": 23414
    },
    {
      "epoch": 0.8016090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1638,
      "step": 23415
    },
    {
      "epoch": 0.8016432728517631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1589,
      "step": 23416
    },
    {
      "epoch": 0.8016775077028415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1529,
      "step": 23417
    },
    {
      "epoch": 0.8017117425539199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1499,
      "step": 23418
    },
    {
      "epoch": 0.8017459774049983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1543,
      "step": 23419
    },
    {
      "epoch": 0.8017802122560767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1499,
      "step": 23420
    },
    {
      "epoch": 0.8018144471071551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1622,
      "step": 23421
    },
    {
      "epoch": 0.8018486819582334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.144,
      "step": 23422
    },
    {
      "epoch": 0.8018829168093119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1511,
      "step": 23423
    },
    {
      "epoch": 0.8019171516603902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1595,
      "step": 23424
    },
    {
      "epoch": 0.8019513865114687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1448,
      "step": 23425
    },
    {
      "epoch": 0.801985621362547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.179,
      "step": 23426
    },
    {
      "epoch": 0.8020198562136255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1452,
      "step": 23427
    },
    {
      "epoch": 0.8020540910647038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1424,
      "step": 23428
    },
    {
      "epoch": 0.8020883259157823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1391,
      "step": 23429
    },
    {
      "epoch": 0.8021225607668606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1402,
      "step": 23430
    },
    {
      "epoch": 0.8021567956179391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1488,
      "step": 23431
    },
    {
      "epoch": 0.8021910304690174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1487,
      "step": 23432
    },
    {
      "epoch": 0.8022252653200959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1529,
      "step": 23433
    },
    {
      "epoch": 0.8022595001711742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.146,
      "step": 23434
    },
    {
      "epoch": 0.8022937350222527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 23435
    },
    {
      "epoch": 0.8023279698733311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1271,
      "step": 23436
    },
    {
      "epoch": 0.8023622047244094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 23437
    },
    {
      "epoch": 0.8023964395754879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1455,
      "step": 23438
    },
    {
      "epoch": 0.8024306744265662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 23439
    },
    {
      "epoch": 0.8024649092776447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1523,
      "step": 23440
    },
    {
      "epoch": 0.802499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1547,
      "step": 23441
    },
    {
      "epoch": 0.8025333789798015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 23442
    },
    {
      "epoch": 0.8025676138308798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.146,
      "step": 23443
    },
    {
      "epoch": 0.8026018486819583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1556,
      "step": 23444
    },
    {
      "epoch": 0.8026360835330366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1506,
      "step": 23445
    },
    {
      "epoch": 0.8026703183841151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1598,
      "step": 23446
    },
    {
      "epoch": 0.8027045532351934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1386,
      "step": 23447
    },
    {
      "epoch": 0.8027387880862719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1548,
      "step": 23448
    },
    {
      "epoch": 0.8027730229373502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1404,
      "step": 23449
    },
    {
      "epoch": 0.8028072577884287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 23450
    },
    {
      "epoch": 0.802841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1677,
      "step": 23451
    },
    {
      "epoch": 0.8028757274905854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1301,
      "step": 23452
    },
    {
      "epoch": 0.8029099623416638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1463,
      "step": 23453
    },
    {
      "epoch": 0.8029441971927422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1463,
      "step": 23454
    },
    {
      "epoch": 0.8029784320438206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1541,
      "step": 23455
    },
    {
      "epoch": 0.803012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1534,
      "step": 23456
    },
    {
      "epoch": 0.8030469017459774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1609,
      "step": 23457
    },
    {
      "epoch": 0.8030811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1456,
      "step": 23458
    },
    {
      "epoch": 0.8031153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1463,
      "step": 23459
    },
    {
      "epoch": 0.8031496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1407,
      "step": 23460
    },
    {
      "epoch": 0.803183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1602,
      "step": 23461
    },
    {
      "epoch": 0.8032180760013694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1448,
      "step": 23462
    },
    {
      "epoch": 0.8032523108524477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.173,
      "step": 23463
    },
    {
      "epoch": 0.8032865457035262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.144,
      "step": 23464
    },
    {
      "epoch": 0.8033207805546045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1476,
      "step": 23465
    },
    {
      "epoch": 0.803355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1367,
      "step": 23466
    },
    {
      "epoch": 0.8033892502567613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1322,
      "step": 23467
    },
    {
      "epoch": 0.8034234851078398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1448,
      "step": 23468
    },
    {
      "epoch": 0.8034577199589181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1689,
      "step": 23469
    },
    {
      "epoch": 0.8034919548099966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1536,
      "step": 23470
    },
    {
      "epoch": 0.8035261896610749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1525,
      "step": 23471
    },
    {
      "epoch": 0.8035604245121534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1641,
      "step": 23472
    },
    {
      "epoch": 0.8035946593632318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1521,
      "step": 23473
    },
    {
      "epoch": 0.8036288942143102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1412,
      "step": 23474
    },
    {
      "epoch": 0.8036631290653886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1759,
      "step": 23475
    },
    {
      "epoch": 0.803697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1412,
      "step": 23476
    },
    {
      "epoch": 0.8037315987675454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1465,
      "step": 23477
    },
    {
      "epoch": 0.8037658336186237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1504,
      "step": 23478
    },
    {
      "epoch": 0.8038000684697022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1497,
      "step": 23479
    },
    {
      "epoch": 0.8038343033207805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.138,
      "step": 23480
    },
    {
      "epoch": 0.803868538171859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1577,
      "step": 23481
    },
    {
      "epoch": 0.8039027730229373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1557,
      "step": 23482
    },
    {
      "epoch": 0.8039370078740158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 23483
    },
    {
      "epoch": 0.8039712427250941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1477,
      "step": 23484
    },
    {
      "epoch": 0.8040054775761726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1638,
      "step": 23485
    },
    {
      "epoch": 0.8040397124272509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.156,
      "step": 23486
    },
    {
      "epoch": 0.8040739472783294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1601,
      "step": 23487
    },
    {
      "epoch": 0.8041081821294077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1416,
      "step": 23488
    },
    {
      "epoch": 0.8041424169804862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1453,
      "step": 23489
    },
    {
      "epoch": 0.8041766518315645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1635,
      "step": 23490
    },
    {
      "epoch": 0.804210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1652,
      "step": 23491
    },
    {
      "epoch": 0.8042451215337213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 23492
    },
    {
      "epoch": 0.8042793563847997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1573,
      "step": 23493
    },
    {
      "epoch": 0.8043135912358781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1484,
      "step": 23494
    },
    {
      "epoch": 0.8043478260869565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1552,
      "step": 23495
    },
    {
      "epoch": 0.8043820609380349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1388,
      "step": 23496
    },
    {
      "epoch": 0.8044162957891133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1435,
      "step": 23497
    },
    {
      "epoch": 0.8044505306401917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1618,
      "step": 23498
    },
    {
      "epoch": 0.8044847654912701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.143,
      "step": 23499
    },
    {
      "epoch": 0.8045190003423485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1534,
      "step": 23500
    },
    {
      "epoch": 0.8045532351934269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1551,
      "step": 23501
    },
    {
      "epoch": 0.8045874700445053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1417,
      "step": 23502
    },
    {
      "epoch": 0.8046217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1428,
      "step": 23503
    },
    {
      "epoch": 0.804655939746662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1377,
      "step": 23504
    },
    {
      "epoch": 0.8046901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.15,
      "step": 23505
    },
    {
      "epoch": 0.8047244094488188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1581,
      "step": 23506
    },
    {
      "epoch": 0.8047586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1692,
      "step": 23507
    },
    {
      "epoch": 0.8047928791509757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1357,
      "step": 23508
    },
    {
      "epoch": 0.8048271140020541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1568,
      "step": 23509
    },
    {
      "epoch": 0.8048613488531325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1589,
      "step": 23510
    },
    {
      "epoch": 0.8048955837042109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1574,
      "step": 23511
    },
    {
      "epoch": 0.8049298185552893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1317,
      "step": 23512
    },
    {
      "epoch": 0.8049640534063677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1368,
      "step": 23513
    },
    {
      "epoch": 0.8049982882574461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1457,
      "step": 23514
    },
    {
      "epoch": 0.8050325231085245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1591,
      "step": 23515
    },
    {
      "epoch": 0.8050667579596029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1461,
      "step": 23516
    },
    {
      "epoch": 0.8051009928106813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1547,
      "step": 23517
    },
    {
      "epoch": 0.8051352276617597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1488,
      "step": 23518
    },
    {
      "epoch": 0.805169462512838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1556,
      "step": 23519
    },
    {
      "epoch": 0.8052036973639165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1484,
      "step": 23520
    },
    {
      "epoch": 0.8052379322149948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1388,
      "step": 23521
    },
    {
      "epoch": 0.8052721670660733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.14,
      "step": 23522
    },
    {
      "epoch": 0.8053064019171516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1423,
      "step": 23523
    },
    {
      "epoch": 0.8053406367682301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1573,
      "step": 23524
    },
    {
      "epoch": 0.8053748716193084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1396,
      "step": 23525
    },
    {
      "epoch": 0.8054091064703869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1618,
      "step": 23526
    },
    {
      "epoch": 0.8054433413214652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.135,
      "step": 23527
    },
    {
      "epoch": 0.8054775761725437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1591,
      "step": 23528
    },
    {
      "epoch": 0.805511811023622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1803,
      "step": 23529
    },
    {
      "epoch": 0.8055460458747005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1497,
      "step": 23530
    },
    {
      "epoch": 0.8055802807257788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.145,
      "step": 23531
    },
    {
      "epoch": 0.8056145155768573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1469,
      "step": 23532
    },
    {
      "epoch": 0.8056487504279356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1404,
      "step": 23533
    },
    {
      "epoch": 0.805682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1587,
      "step": 23534
    },
    {
      "epoch": 0.8057172201300924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1398,
      "step": 23535
    },
    {
      "epoch": 0.8057514549811708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.144,
      "step": 23536
    },
    {
      "epoch": 0.8057856898322492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1566,
      "step": 23537
    },
    {
      "epoch": 0.8058199246833276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1502,
      "step": 23538
    },
    {
      "epoch": 0.805854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1476,
      "step": 23539
    },
    {
      "epoch": 0.8058883943854844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1472,
      "step": 23540
    },
    {
      "epoch": 0.8059226292365628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1606,
      "step": 23541
    },
    {
      "epoch": 0.8059568640876412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1448,
      "step": 23542
    },
    {
      "epoch": 0.8059910989387196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1548,
      "step": 23543
    },
    {
      "epoch": 0.806025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1551,
      "step": 23544
    },
    {
      "epoch": 0.8060595686408765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 23545
    },
    {
      "epoch": 0.8060938034919548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1421,
      "step": 23546
    },
    {
      "epoch": 0.8061280383430333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1458,
      "step": 23547
    },
    {
      "epoch": 0.8061622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1554,
      "step": 23548
    },
    {
      "epoch": 0.80619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1436,
      "step": 23549
    },
    {
      "epoch": 0.8062307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1418,
      "step": 23550
    },
    {
      "epoch": 0.8062649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1527,
      "step": 23551
    },
    {
      "epoch": 0.8062992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1565,
      "step": 23552
    },
    {
      "epoch": 0.8063334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1371,
      "step": 23553
    },
    {
      "epoch": 0.806367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1307,
      "step": 23554
    },
    {
      "epoch": 0.8064019171516604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1572,
      "step": 23555
    },
    {
      "epoch": 0.8064361520027388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.148,
      "step": 23556
    },
    {
      "epoch": 0.8064703868538172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1642,
      "step": 23557
    },
    {
      "epoch": 0.8065046217048956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.126,
      "step": 23558
    },
    {
      "epoch": 0.806538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1275,
      "step": 23559
    },
    {
      "epoch": 0.8065730914070524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1533,
      "step": 23560
    },
    {
      "epoch": 0.8066073262581308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1465,
      "step": 23561
    },
    {
      "epoch": 0.8066415611092091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1438,
      "step": 23562
    },
    {
      "epoch": 0.8066757959602876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1423,
      "step": 23563
    },
    {
      "epoch": 0.8067100308113659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1484,
      "step": 23564
    },
    {
      "epoch": 0.8067442656624444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1455,
      "step": 23565
    },
    {
      "epoch": 0.8067785005135227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1516,
      "step": 23566
    },
    {
      "epoch": 0.8068127353646012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1412,
      "step": 23567
    },
    {
      "epoch": 0.8068469702156795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1517,
      "step": 23568
    },
    {
      "epoch": 0.806881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1479,
      "step": 23569
    },
    {
      "epoch": 0.8069154399178363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1564,
      "step": 23570
    },
    {
      "epoch": 0.8069496747689148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.138,
      "step": 23571
    },
    {
      "epoch": 0.8069839096199931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1533,
      "step": 23572
    },
    {
      "epoch": 0.8070181444710716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1573,
      "step": 23573
    },
    {
      "epoch": 0.8070523793221499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1524,
      "step": 23574
    },
    {
      "epoch": 0.8070866141732284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1404,
      "step": 23575
    },
    {
      "epoch": 0.8071208490243067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1415,
      "step": 23576
    },
    {
      "epoch": 0.8071550838753851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1536,
      "step": 23577
    },
    {
      "epoch": 0.8071893187264635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1557,
      "step": 23578
    },
    {
      "epoch": 0.8072235535775419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1659,
      "step": 23579
    },
    {
      "epoch": 0.8072577884286203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1548,
      "step": 23580
    },
    {
      "epoch": 0.8072920232796987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1558,
      "step": 23581
    },
    {
      "epoch": 0.8073262581307772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1499,
      "step": 23582
    },
    {
      "epoch": 0.8073604929818555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1499,
      "step": 23583
    },
    {
      "epoch": 0.807394727832934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1513,
      "step": 23584
    },
    {
      "epoch": 0.8074289626840123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1628,
      "step": 23585
    },
    {
      "epoch": 0.8074631975350908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1461,
      "step": 23586
    },
    {
      "epoch": 0.8074974323861691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1321,
      "step": 23587
    },
    {
      "epoch": 0.8075316672372476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1523,
      "step": 23588
    },
    {
      "epoch": 0.8075659020883259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1366,
      "step": 23589
    },
    {
      "epoch": 0.8076001369394044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1519,
      "step": 23590
    },
    {
      "epoch": 0.8076343717904827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.157,
      "step": 23591
    },
    {
      "epoch": 0.8076686066415611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1595,
      "step": 23592
    },
    {
      "epoch": 0.8077028414926395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1489,
      "step": 23593
    },
    {
      "epoch": 0.8077370763437179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1555,
      "step": 23594
    },
    {
      "epoch": 0.8077713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1367,
      "step": 23595
    },
    {
      "epoch": 0.8078055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1508,
      "step": 23596
    },
    {
      "epoch": 0.8078397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1642,
      "step": 23597
    },
    {
      "epoch": 0.8078740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1464,
      "step": 23598
    },
    {
      "epoch": 0.8079082505991099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1488,
      "step": 23599
    },
    {
      "epoch": 0.8079424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1414,
      "step": 23600
    },
    {
      "epoch": 0.8079767203012667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1426,
      "step": 23601
    },
    {
      "epoch": 0.8080109551523451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1607,
      "step": 23602
    },
    {
      "epoch": 0.8080451900034235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1404,
      "step": 23603
    },
    {
      "epoch": 0.8080794248545019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1655,
      "step": 23604
    },
    {
      "epoch": 0.8081136597055802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.149,
      "step": 23605
    },
    {
      "epoch": 0.8081478945566587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1667,
      "step": 23606
    },
    {
      "epoch": 0.808182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1566,
      "step": 23607
    },
    {
      "epoch": 0.8082163642588155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1256,
      "step": 23608
    },
    {
      "epoch": 0.8082505991098938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1404,
      "step": 23609
    },
    {
      "epoch": 0.8082848339609723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1367,
      "step": 23610
    },
    {
      "epoch": 0.8083190688120506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 23611
    },
    {
      "epoch": 0.8083533036631291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1571,
      "step": 23612
    },
    {
      "epoch": 0.8083875385142074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 23613
    },
    {
      "epoch": 0.8084217733652859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1655,
      "step": 23614
    },
    {
      "epoch": 0.8084560082163642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.149,
      "step": 23615
    },
    {
      "epoch": 0.8084902430674427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1415,
      "step": 23616
    },
    {
      "epoch": 0.808524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1509,
      "step": 23617
    },
    {
      "epoch": 0.8085587127695995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1597,
      "step": 23618
    },
    {
      "epoch": 0.8085929476206779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1537,
      "step": 23619
    },
    {
      "epoch": 0.8086271824717562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1514,
      "step": 23620
    },
    {
      "epoch": 0.8086614173228347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1522,
      "step": 23621
    },
    {
      "epoch": 0.808695652173913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1411,
      "step": 23622
    },
    {
      "epoch": 0.8087298870249915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1424,
      "step": 23623
    },
    {
      "epoch": 0.8087641218760698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1581,
      "step": 23624
    },
    {
      "epoch": 0.8087983567271483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1456,
      "step": 23625
    },
    {
      "epoch": 0.8088325915782266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1526,
      "step": 23626
    },
    {
      "epoch": 0.8088668264293051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 23627
    },
    {
      "epoch": 0.8089010612803834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1625,
      "step": 23628
    },
    {
      "epoch": 0.8089352961314619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1531,
      "step": 23629
    },
    {
      "epoch": 0.8089695309825402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1384,
      "step": 23630
    },
    {
      "epoch": 0.8090037658336187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1498,
      "step": 23631
    },
    {
      "epoch": 0.809038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1576,
      "step": 23632
    },
    {
      "epoch": 0.8090722355357755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1419,
      "step": 23633
    },
    {
      "epoch": 0.8091064703868538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1336,
      "step": 23634
    },
    {
      "epoch": 0.8091407052379322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1463,
      "step": 23635
    },
    {
      "epoch": 0.8091749400890106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1467,
      "step": 23636
    },
    {
      "epoch": 0.809209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.154,
      "step": 23637
    },
    {
      "epoch": 0.8092434097911674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1544,
      "step": 23638
    },
    {
      "epoch": 0.8092776446422458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1414,
      "step": 23639
    },
    {
      "epoch": 0.8093118794933242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1563,
      "step": 23640
    },
    {
      "epoch": 0.8093461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1552,
      "step": 23641
    },
    {
      "epoch": 0.809380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1346,
      "step": 23642
    },
    {
      "epoch": 0.8094145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.163,
      "step": 23643
    },
    {
      "epoch": 0.8094488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1389,
      "step": 23644
    },
    {
      "epoch": 0.8094830537487162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1284,
      "step": 23645
    },
    {
      "epoch": 0.8095172885997945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1503,
      "step": 23646
    },
    {
      "epoch": 0.809551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1442,
      "step": 23647
    },
    {
      "epoch": 0.8095857583019513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1613,
      "step": 23648
    },
    {
      "epoch": 0.8096199931530298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1464,
      "step": 23649
    },
    {
      "epoch": 0.8096542280041081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.159,
      "step": 23650
    },
    {
      "epoch": 0.8096884628551866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 23651
    },
    {
      "epoch": 0.8097226977062649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1514,
      "step": 23652
    },
    {
      "epoch": 0.8097569325573434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1604,
      "step": 23653
    },
    {
      "epoch": 0.8097911674084217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1518,
      "step": 23654
    },
    {
      "epoch": 0.8098254022595002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.134,
      "step": 23655
    },
    {
      "epoch": 0.8098596371105786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1386,
      "step": 23656
    },
    {
      "epoch": 0.809893871961657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1483,
      "step": 23657
    },
    {
      "epoch": 0.8099281068127354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1343,
      "step": 23658
    },
    {
      "epoch": 0.8099623416638138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1529,
      "step": 23659
    },
    {
      "epoch": 0.8099965765148922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1398,
      "step": 23660
    },
    {
      "epoch": 0.8100308113659705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1395,
      "step": 23661
    },
    {
      "epoch": 0.810065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1364,
      "step": 23662
    },
    {
      "epoch": 0.8100992810681273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1477,
      "step": 23663
    },
    {
      "epoch": 0.8101335159192058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1552,
      "step": 23664
    },
    {
      "epoch": 0.8101677507702841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1609,
      "step": 23665
    },
    {
      "epoch": 0.8102019856213626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1457,
      "step": 23666
    },
    {
      "epoch": 0.8102362204724409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 23667
    },
    {
      "epoch": 0.8102704553235194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1426,
      "step": 23668
    },
    {
      "epoch": 0.8103046901745977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1404,
      "step": 23669
    },
    {
      "epoch": 0.8103389250256762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1557,
      "step": 23670
    },
    {
      "epoch": 0.8103731598767545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1416,
      "step": 23671
    },
    {
      "epoch": 0.810407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1391,
      "step": 23672
    },
    {
      "epoch": 0.8104416295789113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1402,
      "step": 23673
    },
    {
      "epoch": 0.8104758644299898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1516,
      "step": 23674
    },
    {
      "epoch": 0.8105100992810681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1279,
      "step": 23675
    },
    {
      "epoch": 0.8105443341321465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 23676
    },
    {
      "epoch": 0.8105785689832249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1608,
      "step": 23677
    },
    {
      "epoch": 0.8106128038343033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.139,
      "step": 23678
    },
    {
      "epoch": 0.8106470386853817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1382,
      "step": 23679
    },
    {
      "epoch": 0.8106812735364601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1411,
      "step": 23680
    },
    {
      "epoch": 0.8107155083875385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1553,
      "step": 23681
    },
    {
      "epoch": 0.8107497432386169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1418,
      "step": 23682
    },
    {
      "epoch": 0.8107839780896953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1463,
      "step": 23683
    },
    {
      "epoch": 0.8108182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1292,
      "step": 23684
    },
    {
      "epoch": 0.8108524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.141,
      "step": 23685
    },
    {
      "epoch": 0.8108866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1621,
      "step": 23686
    },
    {
      "epoch": 0.8109209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1587,
      "step": 23687
    },
    {
      "epoch": 0.8109551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1494,
      "step": 23688
    },
    {
      "epoch": 0.8109893871961656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1685,
      "step": 23689
    },
    {
      "epoch": 0.8110236220472441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1674,
      "step": 23690
    },
    {
      "epoch": 0.8110578568983224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.164,
      "step": 23691
    },
    {
      "epoch": 0.8110920917494009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1356,
      "step": 23692
    },
    {
      "epoch": 0.8111263266004793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1436,
      "step": 23693
    },
    {
      "epoch": 0.8111605614515577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.157,
      "step": 23694
    },
    {
      "epoch": 0.8111947963026361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1604,
      "step": 23695
    },
    {
      "epoch": 0.8112290311537145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1526,
      "step": 23696
    },
    {
      "epoch": 0.8112632660047929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1614,
      "step": 23697
    },
    {
      "epoch": 0.8112975008558713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1166,
      "step": 23698
    },
    {
      "epoch": 0.8113317357069497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1356,
      "step": 23699
    },
    {
      "epoch": 0.8113659705580281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1561,
      "step": 23700
    },
    {
      "epoch": 0.8114002054091065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1441,
      "step": 23701
    },
    {
      "epoch": 0.8114344402601849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1413,
      "step": 23702
    },
    {
      "epoch": 0.8114686751112633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 23703
    },
    {
      "epoch": 0.8115029099623416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1552,
      "step": 23704
    },
    {
      "epoch": 0.8115371448134201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1522,
      "step": 23705
    },
    {
      "epoch": 0.8115713796644984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1287,
      "step": 23706
    },
    {
      "epoch": 0.8116056145155769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1624,
      "step": 23707
    },
    {
      "epoch": 0.8116398493666552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.131,
      "step": 23708
    },
    {
      "epoch": 0.8116740842177337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1521,
      "step": 23709
    },
    {
      "epoch": 0.811708319068812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1527,
      "step": 23710
    },
    {
      "epoch": 0.8117425539198905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1489,
      "step": 23711
    },
    {
      "epoch": 0.8117767887709688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1583,
      "step": 23712
    },
    {
      "epoch": 0.8118110236220473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1516,
      "step": 23713
    },
    {
      "epoch": 0.8118452584731256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1467,
      "step": 23714
    },
    {
      "epoch": 0.8118794933242041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1492,
      "step": 23715
    },
    {
      "epoch": 0.8119137281752824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.152,
      "step": 23716
    },
    {
      "epoch": 0.8119479630263609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1334,
      "step": 23717
    },
    {
      "epoch": 0.8119821978774392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1437,
      "step": 23718
    },
    {
      "epoch": 0.8120164327285176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 23719
    },
    {
      "epoch": 0.812050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1584,
      "step": 23720
    },
    {
      "epoch": 0.8120849024306744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1547,
      "step": 23721
    },
    {
      "epoch": 0.8121191372817528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1717,
      "step": 23722
    },
    {
      "epoch": 0.8121533721328312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1649,
      "step": 23723
    },
    {
      "epoch": 0.8121876069839096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1354,
      "step": 23724
    },
    {
      "epoch": 0.812221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1348,
      "step": 23725
    },
    {
      "epoch": 0.8122560766860664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1529,
      "step": 23726
    },
    {
      "epoch": 0.8122903115371448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1522,
      "step": 23727
    },
    {
      "epoch": 0.8123245463882232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1572,
      "step": 23728
    },
    {
      "epoch": 0.8123587812393016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1417,
      "step": 23729
    },
    {
      "epoch": 0.8123930160903801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1542,
      "step": 23730
    },
    {
      "epoch": 0.8124272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1357,
      "step": 23731
    },
    {
      "epoch": 0.8124614857925369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1479,
      "step": 23732
    },
    {
      "epoch": 0.8124957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1511,
      "step": 23733
    },
    {
      "epoch": 0.8125299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.157,
      "step": 23734
    },
    {
      "epoch": 0.812564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1324,
      "step": 23735
    },
    {
      "epoch": 0.8125984251968504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1331,
      "step": 23736
    },
    {
      "epoch": 0.8126326600479288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1521,
      "step": 23737
    },
    {
      "epoch": 0.8126668948990072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 23738
    },
    {
      "epoch": 0.8127011297500856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1248,
      "step": 23739
    },
    {
      "epoch": 0.812735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1465,
      "step": 23740
    },
    {
      "epoch": 0.8127695994522424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1346,
      "step": 23741
    },
    {
      "epoch": 0.8128038343033208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1355,
      "step": 23742
    },
    {
      "epoch": 0.8128380691543992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1365,
      "step": 23743
    },
    {
      "epoch": 0.8128723040054776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1512,
      "step": 23744
    },
    {
      "epoch": 0.812906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1442,
      "step": 23745
    },
    {
      "epoch": 0.8129407737076344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1484,
      "step": 23746
    },
    {
      "epoch": 0.8129750085587127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1416,
      "step": 23747
    },
    {
      "epoch": 0.8130092434097912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1317,
      "step": 23748
    },
    {
      "epoch": 0.8130434782608695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.15,
      "step": 23749
    },
    {
      "epoch": 0.813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1268,
      "step": 23750
    },
    {
      "epoch": 0.8131119479630263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1399,
      "step": 23751
    },
    {
      "epoch": 0.8131461828141048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1571,
      "step": 23752
    },
    {
      "epoch": 0.8131804176651831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1514,
      "step": 23753
    },
    {
      "epoch": 0.8132146525162616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1561,
      "step": 23754
    },
    {
      "epoch": 0.8132488873673399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1419,
      "step": 23755
    },
    {
      "epoch": 0.8132831222184184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1603,
      "step": 23756
    },
    {
      "epoch": 0.8133173570694967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1338,
      "step": 23757
    },
    {
      "epoch": 0.8133515919205752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1527,
      "step": 23758
    },
    {
      "epoch": 0.8133858267716535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1334,
      "step": 23759
    },
    {
      "epoch": 0.813420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1338,
      "step": 23760
    },
    {
      "epoch": 0.8134542964738103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1519,
      "step": 23761
    },
    {
      "epoch": 0.8134885313248887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1452,
      "step": 23762
    },
    {
      "epoch": 0.8135227661759671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1345,
      "step": 23763
    },
    {
      "epoch": 0.8135570010270455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1436,
      "step": 23764
    },
    {
      "epoch": 0.8135912358781239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1517,
      "step": 23765
    },
    {
      "epoch": 0.8136254707292023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1468,
      "step": 23766
    },
    {
      "epoch": 0.8136597055802808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1461,
      "step": 23767
    },
    {
      "epoch": 0.8136939404313591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1373,
      "step": 23768
    },
    {
      "epoch": 0.8137281752824376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1548,
      "step": 23769
    },
    {
      "epoch": 0.8137624101335159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1468,
      "step": 23770
    },
    {
      "epoch": 0.8137966449845944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1472,
      "step": 23771
    },
    {
      "epoch": 0.8138308798356727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1437,
      "step": 23772
    },
    {
      "epoch": 0.8138651146867512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1352,
      "step": 23773
    },
    {
      "epoch": 0.8138993495378295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1443,
      "step": 23774
    },
    {
      "epoch": 0.813933584388908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1436,
      "step": 23775
    },
    {
      "epoch": 0.8139678192399863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1598,
      "step": 23776
    },
    {
      "epoch": 0.8140020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1509,
      "step": 23777
    },
    {
      "epoch": 0.8140362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1427,
      "step": 23778
    },
    {
      "epoch": 0.8140705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 23779
    },
    {
      "epoch": 0.8141047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 23780
    },
    {
      "epoch": 0.8141389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.147,
      "step": 23781
    },
    {
      "epoch": 0.8141732283464567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1368,
      "step": 23782
    },
    {
      "epoch": 0.8142074631975351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1452,
      "step": 23783
    },
    {
      "epoch": 0.8142416980486135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1562,
      "step": 23784
    },
    {
      "epoch": 0.8142759328996919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1277,
      "step": 23785
    },
    {
      "epoch": 0.8143101677507703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1526,
      "step": 23786
    },
    {
      "epoch": 0.8143444026018487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1515,
      "step": 23787
    },
    {
      "epoch": 0.814378637452927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 23788
    },
    {
      "epoch": 0.8144128723040055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1472,
      "step": 23789
    },
    {
      "epoch": 0.8144471071550838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.139,
      "step": 23790
    },
    {
      "epoch": 0.8144813420061623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1505,
      "step": 23791
    },
    {
      "epoch": 0.8145155768572406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1583,
      "step": 23792
    },
    {
      "epoch": 0.8145498117083191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.148,
      "step": 23793
    },
    {
      "epoch": 0.8145840465593974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1269,
      "step": 23794
    },
    {
      "epoch": 0.8146182814104759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.152,
      "step": 23795
    },
    {
      "epoch": 0.8146525162615542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1635,
      "step": 23796
    },
    {
      "epoch": 0.8146867511126327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1587,
      "step": 23797
    },
    {
      "epoch": 0.814720985963711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1415,
      "step": 23798
    },
    {
      "epoch": 0.8147552208147895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1398,
      "step": 23799
    },
    {
      "epoch": 0.8147894556658678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1383,
      "step": 23800
    },
    {
      "epoch": 0.8148236905169463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1478,
      "step": 23801
    },
    {
      "epoch": 0.8148579253680246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1436,
      "step": 23802
    },
    {
      "epoch": 0.814892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1569,
      "step": 23803
    },
    {
      "epoch": 0.8149263950701815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1655,
      "step": 23804
    },
    {
      "epoch": 0.8149606299212598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1409,
      "step": 23805
    },
    {
      "epoch": 0.8149948647723383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1594,
      "step": 23806
    },
    {
      "epoch": 0.8150290996234166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1509,
      "step": 23807
    },
    {
      "epoch": 0.8150633344744951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.143,
      "step": 23808
    },
    {
      "epoch": 0.8150975693255734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1618,
      "step": 23809
    },
    {
      "epoch": 0.8151318041766519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1743,
      "step": 23810
    },
    {
      "epoch": 0.8151660390277302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1514,
      "step": 23811
    },
    {
      "epoch": 0.8152002738788087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1632,
      "step": 23812
    },
    {
      "epoch": 0.815234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1443,
      "step": 23813
    },
    {
      "epoch": 0.8152687435809655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1503,
      "step": 23814
    },
    {
      "epoch": 0.8153029784320438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1712,
      "step": 23815
    },
    {
      "epoch": 0.8153372132831223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1522,
      "step": 23816
    },
    {
      "epoch": 0.8153714481342006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1444,
      "step": 23817
    },
    {
      "epoch": 0.815405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1617,
      "step": 23818
    },
    {
      "epoch": 0.8154399178363574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1628,
      "step": 23819
    },
    {
      "epoch": 0.8154741526874358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1596,
      "step": 23820
    },
    {
      "epoch": 0.8155083875385142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1651,
      "step": 23821
    },
    {
      "epoch": 0.8155426223895926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1527,
      "step": 23822
    },
    {
      "epoch": 0.815576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.166,
      "step": 23823
    },
    {
      "epoch": 0.8156110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1791,
      "step": 23824
    },
    {
      "epoch": 0.8156453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1512,
      "step": 23825
    },
    {
      "epoch": 0.8156795617939062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1615,
      "step": 23826
    },
    {
      "epoch": 0.8157137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1325,
      "step": 23827
    },
    {
      "epoch": 0.815748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1463,
      "step": 23828
    },
    {
      "epoch": 0.8157822663471413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 23829
    },
    {
      "epoch": 0.8158165011982198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1474,
      "step": 23830
    },
    {
      "epoch": 0.8158507360492981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1524,
      "step": 23831
    },
    {
      "epoch": 0.8158849709003766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1492,
      "step": 23832
    },
    {
      "epoch": 0.8159192057514549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1607,
      "step": 23833
    },
    {
      "epoch": 0.8159534406025334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1411,
      "step": 23834
    },
    {
      "epoch": 0.8159876754536117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.118,
      "step": 23835
    },
    {
      "epoch": 0.8160219103046902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1574,
      "step": 23836
    },
    {
      "epoch": 0.8160561451557685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1653,
      "step": 23837
    },
    {
      "epoch": 0.816090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1395,
      "step": 23838
    },
    {
      "epoch": 0.8161246148579254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1642,
      "step": 23839
    },
    {
      "epoch": 0.8161588497090038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1499,
      "step": 23840
    },
    {
      "epoch": 0.8161930845600822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1504,
      "step": 23841
    },
    {
      "epoch": 0.8162273194111606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1502,
      "step": 23842
    },
    {
      "epoch": 0.816261554262239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1677,
      "step": 23843
    },
    {
      "epoch": 0.8162957891133173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1564,
      "step": 23844
    },
    {
      "epoch": 0.8163300239643958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1606,
      "step": 23845
    },
    {
      "epoch": 0.8163642588154741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1423,
      "step": 23846
    },
    {
      "epoch": 0.8163984936665526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1519,
      "step": 23847
    },
    {
      "epoch": 0.8164327285176309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1441,
      "step": 23848
    },
    {
      "epoch": 0.8164669633687094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1538,
      "step": 23849
    },
    {
      "epoch": 0.8165011982197877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1333,
      "step": 23850
    },
    {
      "epoch": 0.8165354330708662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 23851
    },
    {
      "epoch": 0.8165696679219445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.143,
      "step": 23852
    },
    {
      "epoch": 0.816603902773023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1444,
      "step": 23853
    },
    {
      "epoch": 0.8166381376241013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1541,
      "step": 23854
    },
    {
      "epoch": 0.8166723724751798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1327,
      "step": 23855
    },
    {
      "epoch": 0.8167066073262581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1373,
      "step": 23856
    },
    {
      "epoch": 0.8167408421773366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1392,
      "step": 23857
    },
    {
      "epoch": 0.8167750770284149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1526,
      "step": 23858
    },
    {
      "epoch": 0.8168093118794933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1475,
      "step": 23859
    },
    {
      "epoch": 0.8168435467305717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1394,
      "step": 23860
    },
    {
      "epoch": 0.8168777815816501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1249,
      "step": 23861
    },
    {
      "epoch": 0.8169120164327285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1555,
      "step": 23862
    },
    {
      "epoch": 0.8169462512838069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1405,
      "step": 23863
    },
    {
      "epoch": 0.8169804861348853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1499,
      "step": 23864
    },
    {
      "epoch": 0.8170147209859637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1401,
      "step": 23865
    },
    {
      "epoch": 0.8170489558370421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1535,
      "step": 23866
    },
    {
      "epoch": 0.8170831906881205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.146,
      "step": 23867
    },
    {
      "epoch": 0.8171174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.157,
      "step": 23868
    },
    {
      "epoch": 0.8171516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1456,
      "step": 23869
    },
    {
      "epoch": 0.8171858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1185,
      "step": 23870
    },
    {
      "epoch": 0.8172201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1338,
      "step": 23871
    },
    {
      "epoch": 0.8172543649435124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1569,
      "step": 23872
    },
    {
      "epoch": 0.8172885997945909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1457,
      "step": 23873
    },
    {
      "epoch": 0.8173228346456692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1449,
      "step": 23874
    },
    {
      "epoch": 0.8173570694967477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1349,
      "step": 23875
    },
    {
      "epoch": 0.8173913043478261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.142,
      "step": 23876
    },
    {
      "epoch": 0.8174255391989045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 23877
    },
    {
      "epoch": 0.8174597740499829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1454,
      "step": 23878
    },
    {
      "epoch": 0.8174940089010613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1309,
      "step": 23879
    },
    {
      "epoch": 0.8175282437521397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1345,
      "step": 23880
    },
    {
      "epoch": 0.8175624786032181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1536,
      "step": 23881
    },
    {
      "epoch": 0.8175967134542965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1421,
      "step": 23882
    },
    {
      "epoch": 0.8176309483053749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1426,
      "step": 23883
    },
    {
      "epoch": 0.8176651831564533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.137,
      "step": 23884
    },
    {
      "epoch": 0.8176994180075317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1494,
      "step": 23885
    },
    {
      "epoch": 0.8177336528586101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1466,
      "step": 23886
    },
    {
      "epoch": 0.8177678877096884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1345,
      "step": 23887
    },
    {
      "epoch": 0.8178021225607669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1522,
      "step": 23888
    },
    {
      "epoch": 0.8178363574118452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1291,
      "step": 23889
    },
    {
      "epoch": 0.8178705922629237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1579,
      "step": 23890
    },
    {
      "epoch": 0.817904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1428,
      "step": 23891
    },
    {
      "epoch": 0.8179390619650805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.148,
      "step": 23892
    },
    {
      "epoch": 0.8179732968161588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1498,
      "step": 23893
    },
    {
      "epoch": 0.8180075316672373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1562,
      "step": 23894
    },
    {
      "epoch": 0.8180417665183156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1463,
      "step": 23895
    },
    {
      "epoch": 0.8180760013693941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1234,
      "step": 23896
    },
    {
      "epoch": 0.8181102362204724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1278,
      "step": 23897
    },
    {
      "epoch": 0.8181444710715509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1239,
      "step": 23898
    },
    {
      "epoch": 0.8181787059226292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1477,
      "step": 23899
    },
    {
      "epoch": 0.8182129407737077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1472,
      "step": 23900
    },
    {
      "epoch": 0.818247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1295,
      "step": 23901
    },
    {
      "epoch": 0.8182814104758644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1439,
      "step": 23902
    },
    {
      "epoch": 0.8183156453269428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1548,
      "step": 23903
    },
    {
      "epoch": 0.8183498801780212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 23904
    },
    {
      "epoch": 0.8183841150290996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1471,
      "step": 23905
    },
    {
      "epoch": 0.818418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1386,
      "step": 23906
    },
    {
      "epoch": 0.8184525847312564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1358,
      "step": 23907
    },
    {
      "epoch": 0.8184868195823348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1404,
      "step": 23908
    },
    {
      "epoch": 0.8185210544334132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1435,
      "step": 23909
    },
    {
      "epoch": 0.8185552892844916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 23910
    },
    {
      "epoch": 0.81858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1603,
      "step": 23911
    },
    {
      "epoch": 0.8186237589866484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1639,
      "step": 23912
    },
    {
      "epoch": 0.8186579938377269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1526,
      "step": 23913
    },
    {
      "epoch": 0.8186922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 23914
    },
    {
      "epoch": 0.8187264635398837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 23915
    },
    {
      "epoch": 0.818760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1554,
      "step": 23916
    },
    {
      "epoch": 0.8187949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1455,
      "step": 23917
    },
    {
      "epoch": 0.8188291680931188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1535,
      "step": 23918
    },
    {
      "epoch": 0.8188634029441972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1504,
      "step": 23919
    },
    {
      "epoch": 0.8188976377952756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1483,
      "step": 23920
    },
    {
      "epoch": 0.818931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1236,
      "step": 23921
    },
    {
      "epoch": 0.8189661074974324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1526,
      "step": 23922
    },
    {
      "epoch": 0.8190003423485108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1497,
      "step": 23923
    },
    {
      "epoch": 0.8190345771995892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1237,
      "step": 23924
    },
    {
      "epoch": 0.8190688120506676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1161,
      "step": 23925
    },
    {
      "epoch": 0.819103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1396,
      "step": 23926
    },
    {
      "epoch": 0.8191372817528244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1492,
      "step": 23927
    },
    {
      "epoch": 0.8191715166039027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1383,
      "step": 23928
    },
    {
      "epoch": 0.8192057514549812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1359,
      "step": 23929
    },
    {
      "epoch": 0.8192399863060595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1498,
      "step": 23930
    },
    {
      "epoch": 0.819274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1469,
      "step": 23931
    },
    {
      "epoch": 0.8193084560082163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1435,
      "step": 23932
    },
    {
      "epoch": 0.8193426908592948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1541,
      "step": 23933
    },
    {
      "epoch": 0.8193769257103731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1379,
      "step": 23934
    },
    {
      "epoch": 0.8194111605614516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1279,
      "step": 23935
    },
    {
      "epoch": 0.8194453954125299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1563,
      "step": 23936
    },
    {
      "epoch": 0.8194796302636084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1488,
      "step": 23937
    },
    {
      "epoch": 0.8195138651146867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1545,
      "step": 23938
    },
    {
      "epoch": 0.8195480999657652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1442,
      "step": 23939
    },
    {
      "epoch": 0.8195823348168435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 23940
    },
    {
      "epoch": 0.819616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1371,
      "step": 23941
    },
    {
      "epoch": 0.8196508045190003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1389,
      "step": 23942
    },
    {
      "epoch": 0.8196850393700787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1589,
      "step": 23943
    },
    {
      "epoch": 0.8197192742211571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1476,
      "step": 23944
    },
    {
      "epoch": 0.8197535090722355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1466,
      "step": 23945
    },
    {
      "epoch": 0.8197877439233139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1337,
      "step": 23946
    },
    {
      "epoch": 0.8198219787743923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1425,
      "step": 23947
    },
    {
      "epoch": 0.8198562136254707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1475,
      "step": 23948
    },
    {
      "epoch": 0.8198904484765491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 23949
    },
    {
      "epoch": 0.8199246833276276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1362,
      "step": 23950
    },
    {
      "epoch": 0.8199589181787059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1413,
      "step": 23951
    },
    {
      "epoch": 0.8199931530297844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1491,
      "step": 23952
    },
    {
      "epoch": 0.8200273878808627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1378,
      "step": 23953
    },
    {
      "epoch": 0.8200616227319412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1317,
      "step": 23954
    },
    {
      "epoch": 0.8200958575830195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1556,
      "step": 23955
    },
    {
      "epoch": 0.820130092434098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1437,
      "step": 23956
    },
    {
      "epoch": 0.8201643272851763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.151,
      "step": 23957
    },
    {
      "epoch": 0.8201985621362547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1403,
      "step": 23958
    },
    {
      "epoch": 0.8202327969873331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1501,
      "step": 23959
    },
    {
      "epoch": 0.8202670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1556,
      "step": 23960
    },
    {
      "epoch": 0.8203012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1351,
      "step": 23961
    },
    {
      "epoch": 0.8203355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1432,
      "step": 23962
    },
    {
      "epoch": 0.8203697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1555,
      "step": 23963
    },
    {
      "epoch": 0.8204039712427251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1416,
      "step": 23964
    },
    {
      "epoch": 0.8204382060938035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1296,
      "step": 23965
    },
    {
      "epoch": 0.8204724409448819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1512,
      "step": 23966
    },
    {
      "epoch": 0.8205066757959603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.139,
      "step": 23967
    },
    {
      "epoch": 0.8205409106470387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1446,
      "step": 23968
    },
    {
      "epoch": 0.820575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1458,
      "step": 23969
    },
    {
      "epoch": 0.8206093803491955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1481,
      "step": 23970
    },
    {
      "epoch": 0.8206436152002738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.141,
      "step": 23971
    },
    {
      "epoch": 0.8206778500513523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1556,
      "step": 23972
    },
    {
      "epoch": 0.8207120849024306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1403,
      "step": 23973
    },
    {
      "epoch": 0.8207463197535091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1387,
      "step": 23974
    },
    {
      "epoch": 0.8207805546045874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 23975
    },
    {
      "epoch": 0.8208147894556659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1465,
      "step": 23976
    },
    {
      "epoch": 0.8208490243067442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1357,
      "step": 23977
    },
    {
      "epoch": 0.8208832591578227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1501,
      "step": 23978
    },
    {
      "epoch": 0.820917494008901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1527,
      "step": 23979
    },
    {
      "epoch": 0.8209517288599795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1381,
      "step": 23980
    },
    {
      "epoch": 0.8209859637110578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1276,
      "step": 23981
    },
    {
      "epoch": 0.8210201985621363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1402,
      "step": 23982
    },
    {
      "epoch": 0.8210544334132146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.134,
      "step": 23983
    },
    {
      "epoch": 0.821088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.152,
      "step": 23984
    },
    {
      "epoch": 0.8211229031153714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1405,
      "step": 23985
    },
    {
      "epoch": 0.8211571379664498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1272,
      "step": 23986
    },
    {
      "epoch": 0.8211913728175283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1372,
      "step": 23987
    },
    {
      "epoch": 0.8212256076686066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.137,
      "step": 23988
    },
    {
      "epoch": 0.8212598425196851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1403,
      "step": 23989
    },
    {
      "epoch": 0.8212940773707634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1378,
      "step": 23990
    },
    {
      "epoch": 0.8213283122218419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1349,
      "step": 23991
    },
    {
      "epoch": 0.8213625470729202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1396,
      "step": 23992
    },
    {
      "epoch": 0.8213967819239987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1328,
      "step": 23993
    },
    {
      "epoch": 0.821431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1587,
      "step": 23994
    },
    {
      "epoch": 0.8214652516261555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1439,
      "step": 23995
    },
    {
      "epoch": 0.8214994864772338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1314,
      "step": 23996
    },
    {
      "epoch": 0.8215337213283123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1723,
      "step": 23997
    },
    {
      "epoch": 0.8215679561793906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1405,
      "step": 23998
    },
    {
      "epoch": 0.821602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1325,
      "step": 23999
    },
    {
      "epoch": 0.8216364258815474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1496,
      "step": 24000
    },
    {
      "epoch": 0.8216706607326258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1513,
      "step": 24001
    },
    {
      "epoch": 0.8217048955837042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1635,
      "step": 24002
    },
    {
      "epoch": 0.8217391304347826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1557,
      "step": 24003
    },
    {
      "epoch": 0.821773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1512,
      "step": 24004
    },
    {
      "epoch": 0.8218076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1602,
      "step": 24005
    },
    {
      "epoch": 0.8218418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.128,
      "step": 24006
    },
    {
      "epoch": 0.8218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1328,
      "step": 24007
    },
    {
      "epoch": 0.8219103046901746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1509,
      "step": 24008
    },
    {
      "epoch": 0.821944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1484,
      "step": 24009
    },
    {
      "epoch": 0.8219787743923314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1405,
      "step": 24010
    },
    {
      "epoch": 0.8220130092434098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1416,
      "step": 24011
    },
    {
      "epoch": 0.8220472440944881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1568,
      "step": 24012
    },
    {
      "epoch": 0.8220814789455666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.13,
      "step": 24013
    },
    {
      "epoch": 0.8221157137966449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.155,
      "step": 24014
    },
    {
      "epoch": 0.8221499486477234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1397,
      "step": 24015
    },
    {
      "epoch": 0.8221841834988017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1481,
      "step": 24016
    },
    {
      "epoch": 0.8222184183498802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1442,
      "step": 24017
    },
    {
      "epoch": 0.8222526532009585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1409,
      "step": 24018
    },
    {
      "epoch": 0.822286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1518,
      "step": 24019
    },
    {
      "epoch": 0.8223211229031153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1421,
      "step": 24020
    },
    {
      "epoch": 0.8223553577541938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1569,
      "step": 24021
    },
    {
      "epoch": 0.8223895926052721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1272,
      "step": 24022
    },
    {
      "epoch": 0.8224238274563506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.167,
      "step": 24023
    },
    {
      "epoch": 0.822458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1459,
      "step": 24024
    },
    {
      "epoch": 0.8224922971585074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1317,
      "step": 24025
    },
    {
      "epoch": 0.8225265320095858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.146,
      "step": 24026
    },
    {
      "epoch": 0.8225607668606641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1496,
      "step": 24027
    },
    {
      "epoch": 0.8225950017117426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1396,
      "step": 24028
    },
    {
      "epoch": 0.8226292365628209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1511,
      "step": 24029
    },
    {
      "epoch": 0.8226634714138994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.151,
      "step": 24030
    },
    {
      "epoch": 0.8226977062649777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1539,
      "step": 24031
    },
    {
      "epoch": 0.8227319411160562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1318,
      "step": 24032
    },
    {
      "epoch": 0.8227661759671345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1378,
      "step": 24033
    },
    {
      "epoch": 0.822800410818213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1403,
      "step": 24034
    },
    {
      "epoch": 0.8228346456692913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1433,
      "step": 24035
    },
    {
      "epoch": 0.8228688805203698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1469,
      "step": 24036
    },
    {
      "epoch": 0.8229031153714481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1381,
      "step": 24037
    },
    {
      "epoch": 0.8229373502225266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1486,
      "step": 24038
    },
    {
      "epoch": 0.8229715850736049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1322,
      "step": 24039
    },
    {
      "epoch": 0.8230058199246834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1385,
      "step": 24040
    },
    {
      "epoch": 0.8230400547757617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1414,
      "step": 24041
    },
    {
      "epoch": 0.8230742896268401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1446,
      "step": 24042
    },
    {
      "epoch": 0.8231085244779185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1333,
      "step": 24043
    },
    {
      "epoch": 0.8231427593289969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1475,
      "step": 24044
    },
    {
      "epoch": 0.8231769941800753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1504,
      "step": 24045
    },
    {
      "epoch": 0.8232112290311537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1459,
      "step": 24046
    },
    {
      "epoch": 0.8232454638822321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1366,
      "step": 24047
    },
    {
      "epoch": 0.8232796987333105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1465,
      "step": 24048
    },
    {
      "epoch": 0.8233139335843889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.143,
      "step": 24049
    },
    {
      "epoch": 0.8233481684354673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1436,
      "step": 24050
    },
    {
      "epoch": 0.8233824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.141,
      "step": 24051
    },
    {
      "epoch": 0.8234166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1431,
      "step": 24052
    },
    {
      "epoch": 0.8234508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1568,
      "step": 24053
    },
    {
      "epoch": 0.8234851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1494,
      "step": 24054
    },
    {
      "epoch": 0.8235193426908592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1337,
      "step": 24055
    },
    {
      "epoch": 0.8235535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1513,
      "step": 24056
    },
    {
      "epoch": 0.823587812393016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1366,
      "step": 24057
    },
    {
      "epoch": 0.8236220472440945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1386,
      "step": 24058
    },
    {
      "epoch": 0.8236562820951728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1471,
      "step": 24059
    },
    {
      "epoch": 0.8236905169462513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1422,
      "step": 24060
    },
    {
      "epoch": 0.8237247517973297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.145,
      "step": 24061
    },
    {
      "epoch": 0.8237589866484081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1486,
      "step": 24062
    },
    {
      "epoch": 0.8237932214994865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1441,
      "step": 24063
    },
    {
      "epoch": 0.8238274563505649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1437,
      "step": 24064
    },
    {
      "epoch": 0.8238616912016433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1423,
      "step": 24065
    },
    {
      "epoch": 0.8238959260527217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1411,
      "step": 24066
    },
    {
      "epoch": 0.8239301609038001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1375,
      "step": 24067
    },
    {
      "epoch": 0.8239643957548785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1327,
      "step": 24068
    },
    {
      "epoch": 0.8239986306059569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1286,
      "step": 24069
    },
    {
      "epoch": 0.8240328654570352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1616,
      "step": 24070
    },
    {
      "epoch": 0.8240671003081137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1523,
      "step": 24071
    },
    {
      "epoch": 0.824101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1397,
      "step": 24072
    },
    {
      "epoch": 0.8241355700102705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1307,
      "step": 24073
    },
    {
      "epoch": 0.8241698048613488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1286,
      "step": 24074
    },
    {
      "epoch": 0.8242040397124273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1401,
      "step": 24075
    },
    {
      "epoch": 0.8242382745635056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1472,
      "step": 24076
    },
    {
      "epoch": 0.8242725094145841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1505,
      "step": 24077
    },
    {
      "epoch": 0.8243067442656624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1432,
      "step": 24078
    },
    {
      "epoch": 0.8243409791167409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1576,
      "step": 24079
    },
    {
      "epoch": 0.8243752139678192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1341,
      "step": 24080
    },
    {
      "epoch": 0.8244094488188977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1415,
      "step": 24081
    },
    {
      "epoch": 0.824443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1417,
      "step": 24082
    },
    {
      "epoch": 0.8244779185210545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1373,
      "step": 24083
    },
    {
      "epoch": 0.8245121533721328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1343,
      "step": 24084
    },
    {
      "epoch": 0.8245463882232112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1425,
      "step": 24085
    },
    {
      "epoch": 0.8245806230742896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.137,
      "step": 24086
    },
    {
      "epoch": 0.824614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1357,
      "step": 24087
    },
    {
      "epoch": 0.8246490927764464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1401,
      "step": 24088
    },
    {
      "epoch": 0.8246833276275248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1395,
      "step": 24089
    },
    {
      "epoch": 0.8247175624786032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1455,
      "step": 24090
    },
    {
      "epoch": 0.8247517973296816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1466,
      "step": 24091
    },
    {
      "epoch": 0.82478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1342,
      "step": 24092
    },
    {
      "epoch": 0.8248202670318384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1531,
      "step": 24093
    },
    {
      "epoch": 0.8248545018829168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.159,
      "step": 24094
    },
    {
      "epoch": 0.8248887367339952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1331,
      "step": 24095
    },
    {
      "epoch": 0.8249229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1494,
      "step": 24096
    },
    {
      "epoch": 0.824957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1374,
      "step": 24097
    },
    {
      "epoch": 0.8249914412872305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1437,
      "step": 24098
    },
    {
      "epoch": 0.8250256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1576,
      "step": 24099
    },
    {
      "epoch": 0.8250599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1313,
      "step": 24100
    },
    {
      "epoch": 0.8250941458404656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1478,
      "step": 24101
    },
    {
      "epoch": 0.825128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1391,
      "step": 24102
    },
    {
      "epoch": 0.8251626155426224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1458,
      "step": 24103
    },
    {
      "epoch": 0.8251968503937008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.137,
      "step": 24104
    },
    {
      "epoch": 0.8252310852447792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1357,
      "step": 24105
    },
    {
      "epoch": 0.8252653200958576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1574,
      "step": 24106
    },
    {
      "epoch": 0.825299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1508,
      "step": 24107
    },
    {
      "epoch": 0.8253337897980144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1462,
      "step": 24108
    },
    {
      "epoch": 0.8253680246490928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1472,
      "step": 24109
    },
    {
      "epoch": 0.8254022595001712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1472,
      "step": 24110
    },
    {
      "epoch": 0.8254364943512496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 24111
    },
    {
      "epoch": 0.825470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1573,
      "step": 24112
    },
    {
      "epoch": 0.8255049640534063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1501,
      "step": 24113
    },
    {
      "epoch": 0.8255391989044848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1457,
      "step": 24114
    },
    {
      "epoch": 0.8255734337555631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1431,
      "step": 24115
    },
    {
      "epoch": 0.8256076686066416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1373,
      "step": 24116
    },
    {
      "epoch": 0.8256419034577199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1522,
      "step": 24117
    },
    {
      "epoch": 0.8256761383087984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1393,
      "step": 24118
    },
    {
      "epoch": 0.8257103731598767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1525,
      "step": 24119
    },
    {
      "epoch": 0.8257446080109552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 24120
    },
    {
      "epoch": 0.8257788428620335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1482,
      "step": 24121
    },
    {
      "epoch": 0.825813077713112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1601,
      "step": 24122
    },
    {
      "epoch": 0.8258473125641903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1427,
      "step": 24123
    },
    {
      "epoch": 0.8258815474152688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1432,
      "step": 24124
    },
    {
      "epoch": 0.8259157822663471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1471,
      "step": 24125
    },
    {
      "epoch": 0.8259500171174256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1348,
      "step": 24126
    },
    {
      "epoch": 0.8259842519685039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1293,
      "step": 24127
    },
    {
      "epoch": 0.8260184868195823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1531,
      "step": 24128
    },
    {
      "epoch": 0.8260527216706607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1331,
      "step": 24129
    },
    {
      "epoch": 0.8260869565217391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1375,
      "step": 24130
    },
    {
      "epoch": 0.8261211913728175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1393,
      "step": 24131
    },
    {
      "epoch": 0.8261554262238959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1345,
      "step": 24132
    },
    {
      "epoch": 0.8261896610749743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1064,
      "step": 24133
    },
    {
      "epoch": 0.8262238959260527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1404,
      "step": 24134
    },
    {
      "epoch": 0.8262581307771312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 24135
    },
    {
      "epoch": 0.8262923656282095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1269,
      "step": 24136
    },
    {
      "epoch": 0.826326600479288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1561,
      "step": 24137
    },
    {
      "epoch": 0.8263608353303663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1511,
      "step": 24138
    },
    {
      "epoch": 0.8263950701814448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.138,
      "step": 24139
    },
    {
      "epoch": 0.8264293050325231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1381,
      "step": 24140
    },
    {
      "epoch": 0.8264635398836016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1466,
      "step": 24141
    },
    {
      "epoch": 0.8264977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1368,
      "step": 24142
    },
    {
      "epoch": 0.8265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1251,
      "step": 24143
    },
    {
      "epoch": 0.8265662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.138,
      "step": 24144
    },
    {
      "epoch": 0.8266004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.151,
      "step": 24145
    },
    {
      "epoch": 0.8266347141389935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1436,
      "step": 24146
    },
    {
      "epoch": 0.8266689489900719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1558,
      "step": 24147
    },
    {
      "epoch": 0.8267031838411503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1331,
      "step": 24148
    },
    {
      "epoch": 0.8267374186922287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.124,
      "step": 24149
    },
    {
      "epoch": 0.8267716535433071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1569,
      "step": 24150
    },
    {
      "epoch": 0.8268058883943855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1247,
      "step": 24151
    },
    {
      "epoch": 0.8268401232454639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1296,
      "step": 24152
    },
    {
      "epoch": 0.8268743580965423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.129,
      "step": 24153
    },
    {
      "epoch": 0.8269085929476206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 24154
    },
    {
      "epoch": 0.8269428277986991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.125,
      "step": 24155
    },
    {
      "epoch": 0.8269770626497774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1234,
      "step": 24156
    },
    {
      "epoch": 0.8270112975008559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.149,
      "step": 24157
    },
    {
      "epoch": 0.8270455323519342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1405,
      "step": 24158
    },
    {
      "epoch": 0.8270797672030127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1284,
      "step": 24159
    },
    {
      "epoch": 0.827114002054091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1422,
      "step": 24160
    },
    {
      "epoch": 0.8271482369051695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1271,
      "step": 24161
    },
    {
      "epoch": 0.8271824717562478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1464,
      "step": 24162
    },
    {
      "epoch": 0.8272167066073263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1478,
      "step": 24163
    },
    {
      "epoch": 0.8272509414584046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1388,
      "step": 24164
    },
    {
      "epoch": 0.8272851763094831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1488,
      "step": 24165
    },
    {
      "epoch": 0.8273194111605614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1389,
      "step": 24166
    },
    {
      "epoch": 0.8273536460116399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1413,
      "step": 24167
    },
    {
      "epoch": 0.8273878808627182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1454,
      "step": 24168
    },
    {
      "epoch": 0.8274221157137966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1344,
      "step": 24169
    },
    {
      "epoch": 0.8274563505648751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.145,
      "step": 24170
    },
    {
      "epoch": 0.8274905854159534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1443,
      "step": 24171
    },
    {
      "epoch": 0.8275248202670319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1444,
      "step": 24172
    },
    {
      "epoch": 0.8275590551181102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1411,
      "step": 24173
    },
    {
      "epoch": 0.8275932899691887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.151,
      "step": 24174
    },
    {
      "epoch": 0.827627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1392,
      "step": 24175
    },
    {
      "epoch": 0.8276617596713455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1303,
      "step": 24176
    },
    {
      "epoch": 0.8276959945224238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1315,
      "step": 24177
    },
    {
      "epoch": 0.8277302293735023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1339,
      "step": 24178
    },
    {
      "epoch": 0.8277644642245806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1385,
      "step": 24179
    },
    {
      "epoch": 0.8277986990756591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1256,
      "step": 24180
    },
    {
      "epoch": 0.8278329339267374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1219,
      "step": 24181
    },
    {
      "epoch": 0.8278671687778159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.142,
      "step": 24182
    },
    {
      "epoch": 0.8279014036288942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1394,
      "step": 24183
    },
    {
      "epoch": 0.8279356384799726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1468,
      "step": 24184
    },
    {
      "epoch": 0.827969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1432,
      "step": 24185
    },
    {
      "epoch": 0.8280041081821294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1498,
      "step": 24186
    },
    {
      "epoch": 0.8280383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1472,
      "step": 24187
    },
    {
      "epoch": 0.8280725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1332,
      "step": 24188
    },
    {
      "epoch": 0.8281068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1356,
      "step": 24189
    },
    {
      "epoch": 0.828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1459,
      "step": 24190
    },
    {
      "epoch": 0.8281752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1386,
      "step": 24191
    },
    {
      "epoch": 0.8282095172885998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1224,
      "step": 24192
    },
    {
      "epoch": 0.8282437521396782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1535,
      "step": 24193
    },
    {
      "epoch": 0.8282779869907566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1322,
      "step": 24194
    },
    {
      "epoch": 0.828312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1368,
      "step": 24195
    },
    {
      "epoch": 0.8283464566929134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1398,
      "step": 24196
    },
    {
      "epoch": 0.8283806915439917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1438,
      "step": 24197
    },
    {
      "epoch": 0.8284149263950702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1329,
      "step": 24198
    },
    {
      "epoch": 0.8284491612461485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1397,
      "step": 24199
    },
    {
      "epoch": 0.828483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1511,
      "step": 24200
    },
    {
      "epoch": 0.8285176309483053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1361,
      "step": 24201
    },
    {
      "epoch": 0.8285518657993838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.145,
      "step": 24202
    },
    {
      "epoch": 0.8285861006504621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 24203
    },
    {
      "epoch": 0.8286203355015406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.145,
      "step": 24204
    },
    {
      "epoch": 0.8286545703526189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1443,
      "step": 24205
    },
    {
      "epoch": 0.8286888052036974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1464,
      "step": 24206
    },
    {
      "epoch": 0.8287230400547758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1541,
      "step": 24207
    },
    {
      "epoch": 0.8287572749058542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.124,
      "step": 24208
    },
    {
      "epoch": 0.8287915097569326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1426,
      "step": 24209
    },
    {
      "epoch": 0.828825744608011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1455,
      "step": 24210
    },
    {
      "epoch": 0.8288599794590894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1318,
      "step": 24211
    },
    {
      "epoch": 0.8288942143101677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1238,
      "step": 24212
    },
    {
      "epoch": 0.8289284491612462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 24213
    },
    {
      "epoch": 0.8289626840123245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1444,
      "step": 24214
    },
    {
      "epoch": 0.828996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1252,
      "step": 24215
    },
    {
      "epoch": 0.8290311537144813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1319,
      "step": 24216
    },
    {
      "epoch": 0.8290653885655598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1466,
      "step": 24217
    },
    {
      "epoch": 0.8290996234166381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 24218
    },
    {
      "epoch": 0.8291338582677166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1495,
      "step": 24219
    },
    {
      "epoch": 0.8291680931187949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1098,
      "step": 24220
    },
    {
      "epoch": 0.8292023279698734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.12,
      "step": 24221
    },
    {
      "epoch": 0.8292365628209517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.131,
      "step": 24222
    },
    {
      "epoch": 0.8292707976720302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1169,
      "step": 24223
    },
    {
      "epoch": 0.8293050325231085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 24224
    },
    {
      "epoch": 0.829339267374187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.149,
      "step": 24225
    },
    {
      "epoch": 0.8293735022252653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1475,
      "step": 24226
    },
    {
      "epoch": 0.8294077370763437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.148,
      "step": 24227
    },
    {
      "epoch": 0.8294419719274221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1298,
      "step": 24228
    },
    {
      "epoch": 0.8294762067785005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.138,
      "step": 24229
    },
    {
      "epoch": 0.8295104416295789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1462,
      "step": 24230
    },
    {
      "epoch": 0.8295446764806573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.143,
      "step": 24231
    },
    {
      "epoch": 0.8295789113317357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1428,
      "step": 24232
    },
    {
      "epoch": 0.8296131461828141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1173,
      "step": 24233
    },
    {
      "epoch": 0.8296473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1358,
      "step": 24234
    },
    {
      "epoch": 0.8296816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1419,
      "step": 24235
    },
    {
      "epoch": 0.8297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1407,
      "step": 24236
    },
    {
      "epoch": 0.8297500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1261,
      "step": 24237
    },
    {
      "epoch": 0.829784320438206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1319,
      "step": 24238
    },
    {
      "epoch": 0.8298185552892845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1547,
      "step": 24239
    },
    {
      "epoch": 0.8298527901403628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1414,
      "step": 24240
    },
    {
      "epoch": 0.8298870249914413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1399,
      "step": 24241
    },
    {
      "epoch": 0.8299212598425196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1329,
      "step": 24242
    },
    {
      "epoch": 0.8299554946935981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1255,
      "step": 24243
    },
    {
      "epoch": 0.8299897295446765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1306,
      "step": 24244
    },
    {
      "epoch": 0.8300239643957549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1353,
      "step": 24245
    },
    {
      "epoch": 0.8300581992468333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1578,
      "step": 24246
    },
    {
      "epoch": 0.8300924340979117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1454,
      "step": 24247
    },
    {
      "epoch": 0.8301266689489901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1358,
      "step": 24248
    },
    {
      "epoch": 0.8301609038000685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 24249
    },
    {
      "epoch": 0.8301951386511469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.129,
      "step": 24250
    },
    {
      "epoch": 0.8302293735022253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1446,
      "step": 24251
    },
    {
      "epoch": 0.8302636083533037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1313,
      "step": 24252
    },
    {
      "epoch": 0.830297843204382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1383,
      "step": 24253
    },
    {
      "epoch": 0.8303320780554605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1157,
      "step": 24254
    },
    {
      "epoch": 0.8303663129065388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1437,
      "step": 24255
    },
    {
      "epoch": 0.8304005477576173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1454,
      "step": 24256
    },
    {
      "epoch": 0.8304347826086956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1228,
      "step": 24257
    },
    {
      "epoch": 0.8304690174597741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1308,
      "step": 24258
    },
    {
      "epoch": 0.8305032523108524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1408,
      "step": 24259
    },
    {
      "epoch": 0.8305374871619309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1525,
      "step": 24260
    },
    {
      "epoch": 0.8305717220130092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1239,
      "step": 24261
    },
    {
      "epoch": 0.8306059568640877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1344,
      "step": 24262
    },
    {
      "epoch": 0.830640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.135,
      "step": 24263
    },
    {
      "epoch": 0.8306744265662445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1452,
      "step": 24264
    },
    {
      "epoch": 0.8307086614173228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1161,
      "step": 24265
    },
    {
      "epoch": 0.8307428962684013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1505,
      "step": 24266
    },
    {
      "epoch": 0.8307771311194796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1391,
      "step": 24267
    },
    {
      "epoch": 0.830811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1469,
      "step": 24268
    },
    {
      "epoch": 0.8308456008216364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1436,
      "step": 24269
    },
    {
      "epoch": 0.8308798356727148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1396,
      "step": 24270
    },
    {
      "epoch": 0.8309140705237932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1314,
      "step": 24271
    },
    {
      "epoch": 0.8309483053748716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1151,
      "step": 24272
    },
    {
      "epoch": 0.83098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1409,
      "step": 24273
    },
    {
      "epoch": 0.8310167750770284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1231,
      "step": 24274
    },
    {
      "epoch": 0.8310510099281068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.137,
      "step": 24275
    },
    {
      "epoch": 0.8310852447791852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1436,
      "step": 24276
    },
    {
      "epoch": 0.8311194796302636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1498,
      "step": 24277
    },
    {
      "epoch": 0.831153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1251,
      "step": 24278
    },
    {
      "epoch": 0.8311879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1278,
      "step": 24279
    },
    {
      "epoch": 0.8312221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1367,
      "step": 24280
    },
    {
      "epoch": 0.8312564190345773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1188,
      "step": 24281
    },
    {
      "epoch": 0.8312906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1253,
      "step": 24282
    },
    {
      "epoch": 0.831324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1546,
      "step": 24283
    },
    {
      "epoch": 0.8313591235878124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1107,
      "step": 24284
    },
    {
      "epoch": 0.8313933584388908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1383,
      "step": 24285
    },
    {
      "epoch": 0.8314275932899692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1529,
      "step": 24286
    },
    {
      "epoch": 0.8314618281410476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1303,
      "step": 24287
    },
    {
      "epoch": 0.831496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1412,
      "step": 24288
    },
    {
      "epoch": 0.8315302978432044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1371,
      "step": 24289
    },
    {
      "epoch": 0.8315645326942828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 24290
    },
    {
      "epoch": 0.8315987675453612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 24291
    },
    {
      "epoch": 0.8316330023964396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1514,
      "step": 24292
    },
    {
      "epoch": 0.831667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1318,
      "step": 24293
    },
    {
      "epoch": 0.8317014720985964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1442,
      "step": 24294
    },
    {
      "epoch": 0.8317357069496748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1456,
      "step": 24295
    },
    {
      "epoch": 0.8317699418007531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1287,
      "step": 24296
    },
    {
      "epoch": 0.8318041766518316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.141,
      "step": 24297
    },
    {
      "epoch": 0.8318384115029099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1317,
      "step": 24298
    },
    {
      "epoch": 0.8318726463539884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1406,
      "step": 24299
    },
    {
      "epoch": 0.8319068812050667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1169,
      "step": 24300
    },
    {
      "epoch": 0.8319411160561452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1306,
      "step": 24301
    },
    {
      "epoch": 0.8319753509072235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1277,
      "step": 24302
    },
    {
      "epoch": 0.832009585758302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1337,
      "step": 24303
    },
    {
      "epoch": 0.8320438206093803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1255,
      "step": 24304
    },
    {
      "epoch": 0.8320780554604588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 24305
    },
    {
      "epoch": 0.8321122903115371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1434,
      "step": 24306
    },
    {
      "epoch": 0.8321465251626156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1195,
      "step": 24307
    },
    {
      "epoch": 0.8321807600136939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1462,
      "step": 24308
    },
    {
      "epoch": 0.8322149948647724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1308,
      "step": 24309
    },
    {
      "epoch": 0.8322492297158507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1269,
      "step": 24310
    },
    {
      "epoch": 0.8322834645669291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1477,
      "step": 24311
    },
    {
      "epoch": 0.8323176994180075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1229,
      "step": 24312
    },
    {
      "epoch": 0.8323519342690859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1617,
      "step": 24313
    },
    {
      "epoch": 0.8323861691201643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1392,
      "step": 24314
    },
    {
      "epoch": 0.8324204039712427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1483,
      "step": 24315
    },
    {
      "epoch": 0.8324546388223211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1376,
      "step": 24316
    },
    {
      "epoch": 0.8324888736733995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 24317
    },
    {
      "epoch": 0.832523108524478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 24318
    },
    {
      "epoch": 0.8325573433755563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.148,
      "step": 24319
    },
    {
      "epoch": 0.8325915782266348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1454,
      "step": 24320
    },
    {
      "epoch": 0.8326258130777131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 24321
    },
    {
      "epoch": 0.8326600479287916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1298,
      "step": 24322
    },
    {
      "epoch": 0.8326942827798699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1372,
      "step": 24323
    },
    {
      "epoch": 0.8327285176309484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1342,
      "step": 24324
    },
    {
      "epoch": 0.8327627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1433,
      "step": 24325
    },
    {
      "epoch": 0.8327969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1274,
      "step": 24326
    },
    {
      "epoch": 0.8328312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1591,
      "step": 24327
    },
    {
      "epoch": 0.8328654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1376,
      "step": 24328
    },
    {
      "epoch": 0.8328996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1306,
      "step": 24329
    },
    {
      "epoch": 0.8329339267374187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1445,
      "step": 24330
    },
    {
      "epoch": 0.8329681615884971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1443,
      "step": 24331
    },
    {
      "epoch": 0.8330023964395755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1287,
      "step": 24332
    },
    {
      "epoch": 0.8330366312906539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1423,
      "step": 24333
    },
    {
      "epoch": 0.8330708661417323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1547,
      "step": 24334
    },
    {
      "epoch": 0.8331051009928107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1225,
      "step": 24335
    },
    {
      "epoch": 0.8331393358438891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1468,
      "step": 24336
    },
    {
      "epoch": 0.8331735706949674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1396,
      "step": 24337
    },
    {
      "epoch": 0.8332078055460459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1545,
      "step": 24338
    },
    {
      "epoch": 0.8332420403971242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1289,
      "step": 24339
    },
    {
      "epoch": 0.8332762752482027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1311,
      "step": 24340
    },
    {
      "epoch": 0.833310510099281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1377,
      "step": 24341
    },
    {
      "epoch": 0.8333447449503595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1385,
      "step": 24342
    },
    {
      "epoch": 0.8333789798014378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1322,
      "step": 24343
    },
    {
      "epoch": 0.8334132146525163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1265,
      "step": 24344
    },
    {
      "epoch": 0.8334474495035946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1327,
      "step": 24345
    },
    {
      "epoch": 0.8334816843546731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1259,
      "step": 24346
    },
    {
      "epoch": 0.8335159192057514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.125,
      "step": 24347
    },
    {
      "epoch": 0.8335501540568299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1352,
      "step": 24348
    },
    {
      "epoch": 0.8335843889079082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1424,
      "step": 24349
    },
    {
      "epoch": 0.8336186237589867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1335,
      "step": 24350
    },
    {
      "epoch": 0.833652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1388,
      "step": 24351
    },
    {
      "epoch": 0.8336870934611434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1407,
      "step": 24352
    },
    {
      "epoch": 0.8337213283122218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.129,
      "step": 24353
    },
    {
      "epoch": 0.8337555631633002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1423,
      "step": 24354
    },
    {
      "epoch": 0.8337897980143787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.134,
      "step": 24355
    },
    {
      "epoch": 0.833824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1383,
      "step": 24356
    },
    {
      "epoch": 0.8338582677165355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1334,
      "step": 24357
    },
    {
      "epoch": 0.8338925025676138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1398,
      "step": 24358
    },
    {
      "epoch": 0.8339267374186923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 24359
    },
    {
      "epoch": 0.8339609722697706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1461,
      "step": 24360
    },
    {
      "epoch": 0.8339952071208491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1469,
      "step": 24361
    },
    {
      "epoch": 0.8340294419719274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1249,
      "step": 24362
    },
    {
      "epoch": 0.8340636768230059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.14,
      "step": 24363
    },
    {
      "epoch": 0.8340979116740842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.135,
      "step": 24364
    },
    {
      "epoch": 0.8341321465251627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1454,
      "step": 24365
    },
    {
      "epoch": 0.834166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.136,
      "step": 24366
    },
    {
      "epoch": 0.8342006162273194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1378,
      "step": 24367
    },
    {
      "epoch": 0.8342348510783978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1252,
      "step": 24368
    },
    {
      "epoch": 0.8342690859294762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1441,
      "step": 24369
    },
    {
      "epoch": 0.8343033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1286,
      "step": 24370
    },
    {
      "epoch": 0.834337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.126,
      "step": 24371
    },
    {
      "epoch": 0.8343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1381,
      "step": 24372
    },
    {
      "epoch": 0.8344060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1294,
      "step": 24373
    },
    {
      "epoch": 0.8344402601848682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1362,
      "step": 24374
    },
    {
      "epoch": 0.8344744950359466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1381,
      "step": 24375
    },
    {
      "epoch": 0.834508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.148,
      "step": 24376
    },
    {
      "epoch": 0.8345429647381034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1248,
      "step": 24377
    },
    {
      "epoch": 0.8345771995891818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1287,
      "step": 24378
    },
    {
      "epoch": 0.8346114344402602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1193,
      "step": 24379
    },
    {
      "epoch": 0.8346456692913385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1161,
      "step": 24380
    },
    {
      "epoch": 0.834679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1361,
      "step": 24381
    },
    {
      "epoch": 0.8347141389934953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1502,
      "step": 24382
    },
    {
      "epoch": 0.8347483738445738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1428,
      "step": 24383
    },
    {
      "epoch": 0.8347826086956521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1248,
      "step": 24384
    },
    {
      "epoch": 0.8348168435467306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1379,
      "step": 24385
    },
    {
      "epoch": 0.8348510783978089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1329,
      "step": 24386
    },
    {
      "epoch": 0.8348853132488874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1417,
      "step": 24387
    },
    {
      "epoch": 0.8349195480999657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1333,
      "step": 24388
    },
    {
      "epoch": 0.8349537829510442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1285,
      "step": 24389
    },
    {
      "epoch": 0.8349880178021225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1272,
      "step": 24390
    },
    {
      "epoch": 0.835022252653201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1302,
      "step": 24391
    },
    {
      "epoch": 0.8350564875042794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1312,
      "step": 24392
    },
    {
      "epoch": 0.8350907223553578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1403,
      "step": 24393
    },
    {
      "epoch": 0.8351249572064362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1401,
      "step": 24394
    },
    {
      "epoch": 0.8351591920575145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.139,
      "step": 24395
    },
    {
      "epoch": 0.835193426908593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1389,
      "step": 24396
    },
    {
      "epoch": 0.8352276617596713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1472,
      "step": 24397
    },
    {
      "epoch": 0.8352618966107498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1289,
      "step": 24398
    },
    {
      "epoch": 0.8352961314618281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1472,
      "step": 24399
    },
    {
      "epoch": 0.8353303663129066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1268,
      "step": 24400
    },
    {
      "epoch": 0.8353646011639849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1237,
      "step": 24401
    },
    {
      "epoch": 0.8353988360150634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1349,
      "step": 24402
    },
    {
      "epoch": 0.8354330708661417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1274,
      "step": 24403
    },
    {
      "epoch": 0.8354673057172202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1131,
      "step": 24404
    },
    {
      "epoch": 0.8355015405682985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1356,
      "step": 24405
    },
    {
      "epoch": 0.835535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1314,
      "step": 24406
    },
    {
      "epoch": 0.8355700102704553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1261,
      "step": 24407
    },
    {
      "epoch": 0.8356042451215338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 24408
    },
    {
      "epoch": 0.8356384799726121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1274,
      "step": 24409
    },
    {
      "epoch": 0.8356727148236905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.142,
      "step": 24410
    },
    {
      "epoch": 0.8357069496747689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1516,
      "step": 24411
    },
    {
      "epoch": 0.8357411845258473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1524,
      "step": 24412
    },
    {
      "epoch": 0.8357754193769257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1492,
      "step": 24413
    },
    {
      "epoch": 0.8358096542280041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1285,
      "step": 24414
    },
    {
      "epoch": 0.8358438890790825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1203,
      "step": 24415
    },
    {
      "epoch": 0.8358781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1463,
      "step": 24416
    },
    {
      "epoch": 0.8359123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1378,
      "step": 24417
    },
    {
      "epoch": 0.8359465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1255,
      "step": 24418
    },
    {
      "epoch": 0.835980828483396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1397,
      "step": 24419
    },
    {
      "epoch": 0.8360150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.122,
      "step": 24420
    },
    {
      "epoch": 0.8360492981855528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1371,
      "step": 24421
    },
    {
      "epoch": 0.8360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.143,
      "step": 24422
    },
    {
      "epoch": 0.8361177678877096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.151,
      "step": 24423
    },
    {
      "epoch": 0.8361520027387881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1372,
      "step": 24424
    },
    {
      "epoch": 0.8361862375898664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1469,
      "step": 24425
    },
    {
      "epoch": 0.8362204724409449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.139,
      "step": 24426
    },
    {
      "epoch": 0.8362547072920232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1417,
      "step": 24427
    },
    {
      "epoch": 0.8362889421431017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1299,
      "step": 24428
    },
    {
      "epoch": 0.8363231769941801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1292,
      "step": 24429
    },
    {
      "epoch": 0.8363574118452585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1328,
      "step": 24430
    },
    {
      "epoch": 0.8363916466963369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1283,
      "step": 24431
    },
    {
      "epoch": 0.8364258815474153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1138,
      "step": 24432
    },
    {
      "epoch": 0.8364601163984937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1438,
      "step": 24433
    },
    {
      "epoch": 0.836494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1397,
      "step": 24434
    },
    {
      "epoch": 0.8365285861006505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1451,
      "step": 24435
    },
    {
      "epoch": 0.8365628209517288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1258,
      "step": 24436
    },
    {
      "epoch": 0.8365970558028073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1409,
      "step": 24437
    },
    {
      "epoch": 0.8366312906538856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1343,
      "step": 24438
    },
    {
      "epoch": 0.8366655255049641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1328,
      "step": 24439
    },
    {
      "epoch": 0.8366997603560424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1474,
      "step": 24440
    },
    {
      "epoch": 0.8367339952071209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1252,
      "step": 24441
    },
    {
      "epoch": 0.8367682300581992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1281,
      "step": 24442
    },
    {
      "epoch": 0.8368024649092777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1453,
      "step": 24443
    },
    {
      "epoch": 0.836836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1229,
      "step": 24444
    },
    {
      "epoch": 0.8368709346114345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1385,
      "step": 24445
    },
    {
      "epoch": 0.8369051694625128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1471,
      "step": 24446
    },
    {
      "epoch": 0.8369394043135913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1505,
      "step": 24447
    },
    {
      "epoch": 0.8369736391646696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.149,
      "step": 24448
    },
    {
      "epoch": 0.837007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1193,
      "step": 24449
    },
    {
      "epoch": 0.8370421088668264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1278,
      "step": 24450
    },
    {
      "epoch": 0.8370763437179048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1359,
      "step": 24451
    },
    {
      "epoch": 0.8371105785689832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1446,
      "step": 24452
    },
    {
      "epoch": 0.8371448134200616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1359,
      "step": 24453
    },
    {
      "epoch": 0.83717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1355,
      "step": 24454
    },
    {
      "epoch": 0.8372132831222184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.128,
      "step": 24455
    },
    {
      "epoch": 0.8372475179732968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 24456
    },
    {
      "epoch": 0.8372817528243752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1386,
      "step": 24457
    },
    {
      "epoch": 0.8373159876754536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 24458
    },
    {
      "epoch": 0.837350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1477,
      "step": 24459
    },
    {
      "epoch": 0.8373844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1095,
      "step": 24460
    },
    {
      "epoch": 0.8374186922286888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1333,
      "step": 24461
    },
    {
      "epoch": 0.8374529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1382,
      "step": 24462
    },
    {
      "epoch": 0.8374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 24463
    },
    {
      "epoch": 0.8375213967819239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1438,
      "step": 24464
    },
    {
      "epoch": 0.8375556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1277,
      "step": 24465
    },
    {
      "epoch": 0.8375898664840808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1303,
      "step": 24466
    },
    {
      "epoch": 0.8376241013351592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.145,
      "step": 24467
    },
    {
      "epoch": 0.8376583361862376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1316,
      "step": 24468
    },
    {
      "epoch": 0.837692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1342,
      "step": 24469
    },
    {
      "epoch": 0.8377268058883944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1248,
      "step": 24470
    },
    {
      "epoch": 0.8377610407394728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.138,
      "step": 24471
    },
    {
      "epoch": 0.8377952755905512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1246,
      "step": 24472
    },
    {
      "epoch": 0.8378295104416296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1457,
      "step": 24473
    },
    {
      "epoch": 0.837863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1335,
      "step": 24474
    },
    {
      "epoch": 0.8378979801437864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1345,
      "step": 24475
    },
    {
      "epoch": 0.8379322149948648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1454,
      "step": 24476
    },
    {
      "epoch": 0.8379664498459432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1434,
      "step": 24477
    },
    {
      "epoch": 0.8380006846970216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1328,
      "step": 24478
    },
    {
      "epoch": 0.8380349195480999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1456,
      "step": 24479
    },
    {
      "epoch": 0.8380691543991784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1301,
      "step": 24480
    },
    {
      "epoch": 0.8381033892502567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1361,
      "step": 24481
    },
    {
      "epoch": 0.8381376241013352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1375,
      "step": 24482
    },
    {
      "epoch": 0.8381718589524135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1267,
      "step": 24483
    },
    {
      "epoch": 0.838206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.154,
      "step": 24484
    },
    {
      "epoch": 0.8382403286545703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1132,
      "step": 24485
    },
    {
      "epoch": 0.8382745635056488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1397,
      "step": 24486
    },
    {
      "epoch": 0.8383087983567271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1281,
      "step": 24487
    },
    {
      "epoch": 0.8383430332078056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1512,
      "step": 24488
    },
    {
      "epoch": 0.8383772680588839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1498,
      "step": 24489
    },
    {
      "epoch": 0.8384115029099624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1398,
      "step": 24490
    },
    {
      "epoch": 0.8384457377610407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1114,
      "step": 24491
    },
    {
      "epoch": 0.8384799726121192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1295,
      "step": 24492
    },
    {
      "epoch": 0.8385142074631975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1473,
      "step": 24493
    },
    {
      "epoch": 0.8385484423142759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.14,
      "step": 24494
    },
    {
      "epoch": 0.8385826771653543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.144,
      "step": 24495
    },
    {
      "epoch": 0.8386169120164327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1483,
      "step": 24496
    },
    {
      "epoch": 0.8386511468675111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1301,
      "step": 24497
    },
    {
      "epoch": 0.8386853817185895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1277,
      "step": 24498
    },
    {
      "epoch": 0.8387196165696679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1382,
      "step": 24499
    },
    {
      "epoch": 0.8387538514207463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1305,
      "step": 24500
    },
    {
      "epoch": 0.8387880862718247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1415,
      "step": 24501
    },
    {
      "epoch": 0.8388223211229031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1351,
      "step": 24502
    },
    {
      "epoch": 0.8388565559739816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1224,
      "step": 24503
    },
    {
      "epoch": 0.8388907908250599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1319,
      "step": 24504
    },
    {
      "epoch": 0.8389250256761384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1341,
      "step": 24505
    },
    {
      "epoch": 0.8389592605272167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1216,
      "step": 24506
    },
    {
      "epoch": 0.8389934953782952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1371,
      "step": 24507
    },
    {
      "epoch": 0.8390277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1418,
      "step": 24508
    },
    {
      "epoch": 0.8390619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1294,
      "step": 24509
    },
    {
      "epoch": 0.8390961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1277,
      "step": 24510
    },
    {
      "epoch": 0.8391304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1536,
      "step": 24511
    },
    {
      "epoch": 0.8391646696336871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1494,
      "step": 24512
    },
    {
      "epoch": 0.8391989044847655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1363,
      "step": 24513
    },
    {
      "epoch": 0.8392331393358439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.123,
      "step": 24514
    },
    {
      "epoch": 0.8392673741869223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1299,
      "step": 24515
    },
    {
      "epoch": 0.8393016090380007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1526,
      "step": 24516
    },
    {
      "epoch": 0.8393358438890791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1268,
      "step": 24517
    },
    {
      "epoch": 0.8393700787401575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1518,
      "step": 24518
    },
    {
      "epoch": 0.8394043135912359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1185,
      "step": 24519
    },
    {
      "epoch": 0.8394385484423142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1359,
      "step": 24520
    },
    {
      "epoch": 0.8394727832933927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1317,
      "step": 24521
    },
    {
      "epoch": 0.839507018144471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1337,
      "step": 24522
    },
    {
      "epoch": 0.8395412529955495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1186,
      "step": 24523
    },
    {
      "epoch": 0.8395754878466278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1207,
      "step": 24524
    },
    {
      "epoch": 0.8396097226977063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1356,
      "step": 24525
    },
    {
      "epoch": 0.8396439575487846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1056,
      "step": 24526
    },
    {
      "epoch": 0.8396781923998631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1409,
      "step": 24527
    },
    {
      "epoch": 0.8397124272509414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1408,
      "step": 24528
    },
    {
      "epoch": 0.8397466621020199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1269,
      "step": 24529
    },
    {
      "epoch": 0.8397808969530982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1204,
      "step": 24530
    },
    {
      "epoch": 0.8398151318041767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1329,
      "step": 24531
    },
    {
      "epoch": 0.839849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1315,
      "step": 24532
    },
    {
      "epoch": 0.8398836015063335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1336,
      "step": 24533
    },
    {
      "epoch": 0.8399178363574118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1281,
      "step": 24534
    },
    {
      "epoch": 0.8399520712084902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1331,
      "step": 24535
    },
    {
      "epoch": 0.8399863060595686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1158,
      "step": 24536
    },
    {
      "epoch": 0.840020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1304,
      "step": 24537
    },
    {
      "epoch": 0.8400547757617255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1298,
      "step": 24538
    },
    {
      "epoch": 0.8400890106128038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1381,
      "step": 24539
    },
    {
      "epoch": 0.8401232454638823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1227,
      "step": 24540
    },
    {
      "epoch": 0.8401574803149606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1112,
      "step": 24541
    },
    {
      "epoch": 0.8401917151660391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1278,
      "step": 24542
    },
    {
      "epoch": 0.8402259500171174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1191,
      "step": 24543
    },
    {
      "epoch": 0.8402601848681959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1481,
      "step": 24544
    },
    {
      "epoch": 0.8402944197192742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1359,
      "step": 24545
    },
    {
      "epoch": 0.8403286545703527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.128,
      "step": 24546
    },
    {
      "epoch": 0.840362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.136,
      "step": 24547
    },
    {
      "epoch": 0.8403971242725095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1445,
      "step": 24548
    },
    {
      "epoch": 0.8404313591235878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 24549
    },
    {
      "epoch": 0.8404655939746662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1359,
      "step": 24550
    },
    {
      "epoch": 0.8404998288257446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1262,
      "step": 24551
    },
    {
      "epoch": 0.840534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1322,
      "step": 24552
    },
    {
      "epoch": 0.8405682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1355,
      "step": 24553
    },
    {
      "epoch": 0.8406025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.133,
      "step": 24554
    },
    {
      "epoch": 0.8406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1454,
      "step": 24555
    },
    {
      "epoch": 0.8406710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1516,
      "step": 24556
    },
    {
      "epoch": 0.840705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.143,
      "step": 24557
    },
    {
      "epoch": 0.8407394727832934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1315,
      "step": 24558
    },
    {
      "epoch": 0.8407737076343718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1237,
      "step": 24559
    },
    {
      "epoch": 0.8408079424854502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 24560
    },
    {
      "epoch": 0.8408421773365286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1342,
      "step": 24561
    },
    {
      "epoch": 0.840876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1245,
      "step": 24562
    },
    {
      "epoch": 0.8409106470386853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1351,
      "step": 24563
    },
    {
      "epoch": 0.8409448818897638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1444,
      "step": 24564
    },
    {
      "epoch": 0.8409791167408421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1267,
      "step": 24565
    },
    {
      "epoch": 0.8410133515919206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1351,
      "step": 24566
    },
    {
      "epoch": 0.8410475864429989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1281,
      "step": 24567
    },
    {
      "epoch": 0.8410818212940774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 24568
    },
    {
      "epoch": 0.8411160561451557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1501,
      "step": 24569
    },
    {
      "epoch": 0.8411502909962342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1315,
      "step": 24570
    },
    {
      "epoch": 0.8411845258473125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.137,
      "step": 24571
    },
    {
      "epoch": 0.841218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1148,
      "step": 24572
    },
    {
      "epoch": 0.8412529955494693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.132,
      "step": 24573
    },
    {
      "epoch": 0.8412872304005478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1291,
      "step": 24574
    },
    {
      "epoch": 0.8413214652516262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 24575
    },
    {
      "epoch": 0.8413557001027046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1188,
      "step": 24576
    },
    {
      "epoch": 0.841389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1393,
      "step": 24577
    },
    {
      "epoch": 0.8414241698048613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1373,
      "step": 24578
    },
    {
      "epoch": 0.8414584046559398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1381,
      "step": 24579
    },
    {
      "epoch": 0.8414926395070181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1463,
      "step": 24580
    },
    {
      "epoch": 0.8415268743580966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1562,
      "step": 24581
    },
    {
      "epoch": 0.8415611092091749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1331,
      "step": 24582
    },
    {
      "epoch": 0.8415953440602534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1314,
      "step": 24583
    },
    {
      "epoch": 0.8416295789113317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1341,
      "step": 24584
    },
    {
      "epoch": 0.8416638137624102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.131,
      "step": 24585
    },
    {
      "epoch": 0.8416980486134885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1404,
      "step": 24586
    },
    {
      "epoch": 0.841732283464567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1217,
      "step": 24587
    },
    {
      "epoch": 0.8417665183156453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1418,
      "step": 24588
    },
    {
      "epoch": 0.8418007531667238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1239,
      "step": 24589
    },
    {
      "epoch": 0.8418349880178021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1321,
      "step": 24590
    },
    {
      "epoch": 0.8418692228688806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1243,
      "step": 24591
    },
    {
      "epoch": 0.8419034577199589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1174,
      "step": 24592
    },
    {
      "epoch": 0.8419376925710373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.132,
      "step": 24593
    },
    {
      "epoch": 0.8419719274221157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1311,
      "step": 24594
    },
    {
      "epoch": 0.8420061622731941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.128,
      "step": 24595
    },
    {
      "epoch": 0.8420403971242725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1402,
      "step": 24596
    },
    {
      "epoch": 0.8420746319753509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1356,
      "step": 24597
    },
    {
      "epoch": 0.8421088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1494,
      "step": 24598
    },
    {
      "epoch": 0.8421431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1425,
      "step": 24599
    },
    {
      "epoch": 0.8421773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1398,
      "step": 24600
    },
    {
      "epoch": 0.8422115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1285,
      "step": 24601
    },
    {
      "epoch": 0.8422458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1224,
      "step": 24602
    },
    {
      "epoch": 0.8422800410818213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1298,
      "step": 24603
    },
    {
      "epoch": 0.8423142759328996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1289,
      "step": 24604
    },
    {
      "epoch": 0.8423485107839781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1324,
      "step": 24605
    },
    {
      "epoch": 0.8423827456350564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.137,
      "step": 24606
    },
    {
      "epoch": 0.8424169804861349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1405,
      "step": 24607
    },
    {
      "epoch": 0.8424512153372132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1272,
      "step": 24608
    },
    {
      "epoch": 0.8424854501882917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1152,
      "step": 24609
    },
    {
      "epoch": 0.84251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1411,
      "step": 24610
    },
    {
      "epoch": 0.8425539198904485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.127,
      "step": 24611
    },
    {
      "epoch": 0.8425881547415269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1186,
      "step": 24612
    },
    {
      "epoch": 0.8426223895926053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1281,
      "step": 24613
    },
    {
      "epoch": 0.8426566244436837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1311,
      "step": 24614
    },
    {
      "epoch": 0.8426908592947621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.128,
      "step": 24615
    },
    {
      "epoch": 0.8427250941458405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1289,
      "step": 24616
    },
    {
      "epoch": 0.8427593289969189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1425,
      "step": 24617
    },
    {
      "epoch": 0.8427935638479973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.112,
      "step": 24618
    },
    {
      "epoch": 0.8428277986990756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1277,
      "step": 24619
    },
    {
      "epoch": 0.8428620335501541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1468,
      "step": 24620
    },
    {
      "epoch": 0.8428962684012324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1208,
      "step": 24621
    },
    {
      "epoch": 0.8429305032523109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1402,
      "step": 24622
    },
    {
      "epoch": 0.8429647381033892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1379,
      "step": 24623
    },
    {
      "epoch": 0.8429989729544677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.142,
      "step": 24624
    },
    {
      "epoch": 0.843033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1339,
      "step": 24625
    },
    {
      "epoch": 0.8430674426566245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1383,
      "step": 24626
    },
    {
      "epoch": 0.8431016775077028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1267,
      "step": 24627
    },
    {
      "epoch": 0.8431359123587813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.133,
      "step": 24628
    },
    {
      "epoch": 0.8431701472098596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1235,
      "step": 24629
    },
    {
      "epoch": 0.8432043820609381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1339,
      "step": 24630
    },
    {
      "epoch": 0.8432386169120164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1234,
      "step": 24631
    },
    {
      "epoch": 0.8432728517630949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1175,
      "step": 24632
    },
    {
      "epoch": 0.8433070866141732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1515,
      "step": 24633
    },
    {
      "epoch": 0.8433413214652516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1288,
      "step": 24634
    },
    {
      "epoch": 0.84337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1297,
      "step": 24635
    },
    {
      "epoch": 0.8434097911674084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.14,
      "step": 24636
    },
    {
      "epoch": 0.8434440260184868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.138,
      "step": 24637
    },
    {
      "epoch": 0.8434782608695652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1247,
      "step": 24638
    },
    {
      "epoch": 0.8435124957206436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1356,
      "step": 24639
    },
    {
      "epoch": 0.843546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1302,
      "step": 24640
    },
    {
      "epoch": 0.8435809654228004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1366,
      "step": 24641
    },
    {
      "epoch": 0.8436152002738788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1282,
      "step": 24642
    },
    {
      "epoch": 0.8436494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1363,
      "step": 24643
    },
    {
      "epoch": 0.8436836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1197,
      "step": 24644
    },
    {
      "epoch": 0.843717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1367,
      "step": 24645
    },
    {
      "epoch": 0.8437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 24646
    },
    {
      "epoch": 0.8437863745292707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1072,
      "step": 24647
    },
    {
      "epoch": 0.8438206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1363,
      "step": 24648
    },
    {
      "epoch": 0.8438548442314276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1323,
      "step": 24649
    },
    {
      "epoch": 0.843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1403,
      "step": 24650
    },
    {
      "epoch": 0.8439233139335844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1359,
      "step": 24651
    },
    {
      "epoch": 0.8439575487846628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1299,
      "step": 24652
    },
    {
      "epoch": 0.8439917836357412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1255,
      "step": 24653
    },
    {
      "epoch": 0.8440260184868196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1381,
      "step": 24654
    },
    {
      "epoch": 0.844060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.128,
      "step": 24655
    },
    {
      "epoch": 0.8440944881889764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1315,
      "step": 24656
    },
    {
      "epoch": 0.8441287230400548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1305,
      "step": 24657
    },
    {
      "epoch": 0.8441629578911332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1297,
      "step": 24658
    },
    {
      "epoch": 0.8441971927422116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1208,
      "step": 24659
    },
    {
      "epoch": 0.84423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1152,
      "step": 24660
    },
    {
      "epoch": 0.8442656624443684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1301,
      "step": 24661
    },
    {
      "epoch": 0.8442998972954467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.141,
      "step": 24662
    },
    {
      "epoch": 0.8443341321465252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1579,
      "step": 24663
    },
    {
      "epoch": 0.8443683669976035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1308,
      "step": 24664
    },
    {
      "epoch": 0.844402601848682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1204,
      "step": 24665
    },
    {
      "epoch": 0.8444368366997603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1296,
      "step": 24666
    },
    {
      "epoch": 0.8444710715508388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1207,
      "step": 24667
    },
    {
      "epoch": 0.8445053064019171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1278,
      "step": 24668
    },
    {
      "epoch": 0.8445395412529956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1462,
      "step": 24669
    },
    {
      "epoch": 0.8445737761040739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1388,
      "step": 24670
    },
    {
      "epoch": 0.8446080109551524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1362,
      "step": 24671
    },
    {
      "epoch": 0.8446422458062307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1335,
      "step": 24672
    },
    {
      "epoch": 0.8446764806573092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1293,
      "step": 24673
    },
    {
      "epoch": 0.8447107155083875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1384,
      "step": 24674
    },
    {
      "epoch": 0.844744950359466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 24675
    },
    {
      "epoch": 0.8447791852105443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1254,
      "step": 24676
    },
    {
      "epoch": 0.8448134200616227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1357,
      "step": 24677
    },
    {
      "epoch": 0.8448476549127011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 24678
    },
    {
      "epoch": 0.8448818897637795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1501,
      "step": 24679
    },
    {
      "epoch": 0.8449161246148579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1392,
      "step": 24680
    },
    {
      "epoch": 0.8449503594659363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1307,
      "step": 24681
    },
    {
      "epoch": 0.8449845943170147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1378,
      "step": 24682
    },
    {
      "epoch": 0.8450188291680931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.112,
      "step": 24683
    },
    {
      "epoch": 0.8450530640191715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1384,
      "step": 24684
    },
    {
      "epoch": 0.8450872988702499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1331,
      "step": 24685
    },
    {
      "epoch": 0.8451215337213284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1218,
      "step": 24686
    },
    {
      "epoch": 0.8451557685724067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1333,
      "step": 24687
    },
    {
      "epoch": 0.8451900034234852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1295,
      "step": 24688
    },
    {
      "epoch": 0.8452242382745635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1257,
      "step": 24689
    },
    {
      "epoch": 0.845258473125642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1286,
      "step": 24690
    },
    {
      "epoch": 0.8452927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1502,
      "step": 24691
    },
    {
      "epoch": 0.8453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1187,
      "step": 24692
    },
    {
      "epoch": 0.8453611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1161,
      "step": 24693
    },
    {
      "epoch": 0.8453954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 24694
    },
    {
      "epoch": 0.8454296473810339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1547,
      "step": 24695
    },
    {
      "epoch": 0.8454638822321123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1212,
      "step": 24696
    },
    {
      "epoch": 0.8454981170831907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1098,
      "step": 24697
    },
    {
      "epoch": 0.8455323519342691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1146,
      "step": 24698
    },
    {
      "epoch": 0.8455665867853475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1465,
      "step": 24699
    },
    {
      "epoch": 0.8456008216364259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1235,
      "step": 24700
    },
    {
      "epoch": 0.8456350564875043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1422,
      "step": 24701
    },
    {
      "epoch": 0.8456692913385827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1483,
      "step": 24702
    },
    {
      "epoch": 0.845703526189661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.122,
      "step": 24703
    },
    {
      "epoch": 0.8457377610407395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1346,
      "step": 24704
    },
    {
      "epoch": 0.8457719958918178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1337,
      "step": 24705
    },
    {
      "epoch": 0.8458062307428963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1525,
      "step": 24706
    },
    {
      "epoch": 0.8458404655939746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1303,
      "step": 24707
    },
    {
      "epoch": 0.8458747004450531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1342,
      "step": 24708
    },
    {
      "epoch": 0.8459089352961314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1395,
      "step": 24709
    },
    {
      "epoch": 0.8459431701472099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1249,
      "step": 24710
    },
    {
      "epoch": 0.8459774049982882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1457,
      "step": 24711
    },
    {
      "epoch": 0.8460116398493667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1418,
      "step": 24712
    },
    {
      "epoch": 0.846045874700445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1236,
      "step": 24713
    },
    {
      "epoch": 0.8460801095515235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.144,
      "step": 24714
    },
    {
      "epoch": 0.8461143444026018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1331,
      "step": 24715
    },
    {
      "epoch": 0.8461485792536803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 24716
    },
    {
      "epoch": 0.8461828141047586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 24717
    },
    {
      "epoch": 0.846217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1175,
      "step": 24718
    },
    {
      "epoch": 0.8462512838069154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1387,
      "step": 24719
    },
    {
      "epoch": 0.8462855186579938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1433,
      "step": 24720
    },
    {
      "epoch": 0.8463197535090722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1405,
      "step": 24721
    },
    {
      "epoch": 0.8463539883601506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1317,
      "step": 24722
    },
    {
      "epoch": 0.8463882232112291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1101,
      "step": 24723
    },
    {
      "epoch": 0.8464224580623074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1479,
      "step": 24724
    },
    {
      "epoch": 0.8464566929133859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1314,
      "step": 24725
    },
    {
      "epoch": 0.8464909277644642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1247,
      "step": 24726
    },
    {
      "epoch": 0.8465251626155427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.134,
      "step": 24727
    },
    {
      "epoch": 0.846559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1282,
      "step": 24728
    },
    {
      "epoch": 0.8465936323176995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1253,
      "step": 24729
    },
    {
      "epoch": 0.8466278671687778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1281,
      "step": 24730
    },
    {
      "epoch": 0.8466621020198563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1313,
      "step": 24731
    },
    {
      "epoch": 0.8466963368709346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1205,
      "step": 24732
    },
    {
      "epoch": 0.846730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1178,
      "step": 24733
    },
    {
      "epoch": 0.8467648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1407,
      "step": 24734
    },
    {
      "epoch": 0.8467990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1192,
      "step": 24735
    },
    {
      "epoch": 0.8468332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1341,
      "step": 24736
    },
    {
      "epoch": 0.8468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1451,
      "step": 24737
    },
    {
      "epoch": 0.846901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1115,
      "step": 24738
    },
    {
      "epoch": 0.8469359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1214,
      "step": 24739
    },
    {
      "epoch": 0.8469702156795618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1528,
      "step": 24740
    },
    {
      "epoch": 0.8470044505306402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1383,
      "step": 24741
    },
    {
      "epoch": 0.8470386853817186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1393,
      "step": 24742
    },
    {
      "epoch": 0.847072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1171,
      "step": 24743
    },
    {
      "epoch": 0.8471071550838754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1382,
      "step": 24744
    },
    {
      "epoch": 0.8471413899349538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1157,
      "step": 24745
    },
    {
      "epoch": 0.8471756247860321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1266,
      "step": 24746
    },
    {
      "epoch": 0.8472098596371106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1349,
      "step": 24747
    },
    {
      "epoch": 0.8472440944881889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1268,
      "step": 24748
    },
    {
      "epoch": 0.8472783293392674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1168,
      "step": 24749
    },
    {
      "epoch": 0.8473125641903457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1283,
      "step": 24750
    },
    {
      "epoch": 0.8473467990414242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1166,
      "step": 24751
    },
    {
      "epoch": 0.8473810338925025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1291,
      "step": 24752
    },
    {
      "epoch": 0.847415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1224,
      "step": 24753
    },
    {
      "epoch": 0.8474495035946593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1225,
      "step": 24754
    },
    {
      "epoch": 0.8474837384457378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1245,
      "step": 24755
    },
    {
      "epoch": 0.8475179732968161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1118,
      "step": 24756
    },
    {
      "epoch": 0.8475522081478946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1353,
      "step": 24757
    },
    {
      "epoch": 0.8475864429989729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1352,
      "step": 24758
    },
    {
      "epoch": 0.8476206778500514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1363,
      "step": 24759
    },
    {
      "epoch": 0.8476549127011298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1256,
      "step": 24760
    },
    {
      "epoch": 0.8476891475522081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1325,
      "step": 24761
    },
    {
      "epoch": 0.8477233824032866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1201,
      "step": 24762
    },
    {
      "epoch": 0.8477576172543649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1296,
      "step": 24763
    },
    {
      "epoch": 0.8477918521054434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1313,
      "step": 24764
    },
    {
      "epoch": 0.8478260869565217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1442,
      "step": 24765
    },
    {
      "epoch": 0.8478603218076002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1114,
      "step": 24766
    },
    {
      "epoch": 0.8478945566586785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1426,
      "step": 24767
    },
    {
      "epoch": 0.847928791509757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1341,
      "step": 24768
    },
    {
      "epoch": 0.8479630263608353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1254,
      "step": 24769
    },
    {
      "epoch": 0.8479972612119138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1254,
      "step": 24770
    },
    {
      "epoch": 0.8480314960629921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1153,
      "step": 24771
    },
    {
      "epoch": 0.8480657309140706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1369,
      "step": 24772
    },
    {
      "epoch": 0.8480999657651489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1542,
      "step": 24773
    },
    {
      "epoch": 0.8481342006162274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1431,
      "step": 24774
    },
    {
      "epoch": 0.8481684354673057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1211,
      "step": 24775
    },
    {
      "epoch": 0.8482026703183841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1351,
      "step": 24776
    },
    {
      "epoch": 0.8482369051694625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1426,
      "step": 24777
    },
    {
      "epoch": 0.8482711400205409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1252,
      "step": 24778
    },
    {
      "epoch": 0.8483053748716193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1432,
      "step": 24779
    },
    {
      "epoch": 0.8483396097226977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1268,
      "step": 24780
    },
    {
      "epoch": 0.8483738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1323,
      "step": 24781
    },
    {
      "epoch": 0.8484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1369,
      "step": 24782
    },
    {
      "epoch": 0.8484423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1258,
      "step": 24783
    },
    {
      "epoch": 0.8484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1409,
      "step": 24784
    },
    {
      "epoch": 0.8485107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1031,
      "step": 24785
    },
    {
      "epoch": 0.8485450188291681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.121,
      "step": 24786
    },
    {
      "epoch": 0.8485792536802464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1385,
      "step": 24787
    },
    {
      "epoch": 0.8486134885313249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1214,
      "step": 24788
    },
    {
      "epoch": 0.8486477233824032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1485,
      "step": 24789
    },
    {
      "epoch": 0.8486819582334817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1292,
      "step": 24790
    },
    {
      "epoch": 0.84871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1254,
      "step": 24791
    },
    {
      "epoch": 0.8487504279356385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1135,
      "step": 24792
    },
    {
      "epoch": 0.8487846627867168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1295,
      "step": 24793
    },
    {
      "epoch": 0.8488188976377953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 24794
    },
    {
      "epoch": 0.8488531324888736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 24795
    },
    {
      "epoch": 0.8488873673399521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1245,
      "step": 24796
    },
    {
      "epoch": 0.8489216021910305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1324,
      "step": 24797
    },
    {
      "epoch": 0.8489558370421089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1213,
      "step": 24798
    },
    {
      "epoch": 0.8489900718931873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 24799
    },
    {
      "epoch": 0.8490243067442657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.122,
      "step": 24800
    },
    {
      "epoch": 0.8490585415953441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1222,
      "step": 24801
    },
    {
      "epoch": 0.8490927764464224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1264,
      "step": 24802
    },
    {
      "epoch": 0.8491270112975009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.115,
      "step": 24803
    },
    {
      "epoch": 0.8491612461485792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1119,
      "step": 24804
    },
    {
      "epoch": 0.8491954809996577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 24805
    },
    {
      "epoch": 0.849229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1359,
      "step": 24806
    },
    {
      "epoch": 0.8492639507018145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1385,
      "step": 24807
    },
    {
      "epoch": 0.8492981855528928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1282,
      "step": 24808
    },
    {
      "epoch": 0.8493324204039713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.126,
      "step": 24809
    },
    {
      "epoch": 0.8493666552550496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1318,
      "step": 24810
    },
    {
      "epoch": 0.8494008901061281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1212,
      "step": 24811
    },
    {
      "epoch": 0.8494351249572064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1501,
      "step": 24812
    },
    {
      "epoch": 0.8494693598082849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1418,
      "step": 24813
    },
    {
      "epoch": 0.8495035946593632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1387,
      "step": 24814
    },
    {
      "epoch": 0.8495378295104417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1065,
      "step": 24815
    },
    {
      "epoch": 0.84957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1198,
      "step": 24816
    },
    {
      "epoch": 0.8496062992125984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1397,
      "step": 24817
    },
    {
      "epoch": 0.8496405340636768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1272,
      "step": 24818
    },
    {
      "epoch": 0.8496747689147552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1082,
      "step": 24819
    },
    {
      "epoch": 0.8497090037658336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1276,
      "step": 24820
    },
    {
      "epoch": 0.849743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.132,
      "step": 24821
    },
    {
      "epoch": 0.8497774734679904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1422,
      "step": 24822
    },
    {
      "epoch": 0.8498117083190688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1303,
      "step": 24823
    },
    {
      "epoch": 0.8498459431701472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1341,
      "step": 24824
    },
    {
      "epoch": 0.8498801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.118,
      "step": 24825
    },
    {
      "epoch": 0.849914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1288,
      "step": 24826
    },
    {
      "epoch": 0.8499486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.126,
      "step": 24827
    },
    {
      "epoch": 0.8499828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1286,
      "step": 24828
    },
    {
      "epoch": 0.8500171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1393,
      "step": 24829
    },
    {
      "epoch": 0.8500513522766175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1396,
      "step": 24830
    },
    {
      "epoch": 0.850085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1447,
      "step": 24831
    },
    {
      "epoch": 0.8501198219787743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1322,
      "step": 24832
    },
    {
      "epoch": 0.8501540568298528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1372,
      "step": 24833
    },
    {
      "epoch": 0.8501882916809312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1293,
      "step": 24834
    },
    {
      "epoch": 0.8502225265320096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.123,
      "step": 24835
    },
    {
      "epoch": 0.850256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1114,
      "step": 24836
    },
    {
      "epoch": 0.8502909962341664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1257,
      "step": 24837
    },
    {
      "epoch": 0.8503252310852448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1373,
      "step": 24838
    },
    {
      "epoch": 0.8503594659363232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1285,
      "step": 24839
    },
    {
      "epoch": 0.8503937007874016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1246,
      "step": 24840
    },
    {
      "epoch": 0.85042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1239,
      "step": 24841
    },
    {
      "epoch": 0.8504621704895584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1504,
      "step": 24842
    },
    {
      "epoch": 0.8504964053406368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.128,
      "step": 24843
    },
    {
      "epoch": 0.8505306401917152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1128,
      "step": 24844
    },
    {
      "epoch": 0.8505648750427935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1205,
      "step": 24845
    },
    {
      "epoch": 0.850599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1266,
      "step": 24846
    },
    {
      "epoch": 0.8506333447449503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1334,
      "step": 24847
    },
    {
      "epoch": 0.8506675795960288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1343,
      "step": 24848
    },
    {
      "epoch": 0.8507018144471071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1471,
      "step": 24849
    },
    {
      "epoch": 0.8507360492981856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1284,
      "step": 24850
    },
    {
      "epoch": 0.8507702841492639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1356,
      "step": 24851
    },
    {
      "epoch": 0.8508045190003424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1381,
      "step": 24852
    },
    {
      "epoch": 0.8508387538514207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1188,
      "step": 24853
    },
    {
      "epoch": 0.8508729887024992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1324,
      "step": 24854
    },
    {
      "epoch": 0.8509072235535775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1245,
      "step": 24855
    },
    {
      "epoch": 0.850941458404656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1476,
      "step": 24856
    },
    {
      "epoch": 0.8509756932557343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.123,
      "step": 24857
    },
    {
      "epoch": 0.8510099281068128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1077,
      "step": 24858
    },
    {
      "epoch": 0.8510441629578911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1201,
      "step": 24859
    },
    {
      "epoch": 0.8510783978089695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1228,
      "step": 24860
    },
    {
      "epoch": 0.8511126326600479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1373,
      "step": 24861
    },
    {
      "epoch": 0.8511468675111263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1296,
      "step": 24862
    },
    {
      "epoch": 0.8511811023622047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1138,
      "step": 24863
    },
    {
      "epoch": 0.8512153372132831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1328,
      "step": 24864
    },
    {
      "epoch": 0.8512495720643615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.133,
      "step": 24865
    },
    {
      "epoch": 0.8512838069154399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1279,
      "step": 24866
    },
    {
      "epoch": 0.8513180417665183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1282,
      "step": 24867
    },
    {
      "epoch": 0.8513522766175967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1397,
      "step": 24868
    },
    {
      "epoch": 0.8513865114686752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1492,
      "step": 24869
    },
    {
      "epoch": 0.8514207463197535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1131,
      "step": 24870
    },
    {
      "epoch": 0.851454981170832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1264,
      "step": 24871
    },
    {
      "epoch": 0.8514892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1346,
      "step": 24872
    },
    {
      "epoch": 0.8515234508729888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1222,
      "step": 24873
    },
    {
      "epoch": 0.8515576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1316,
      "step": 24874
    },
    {
      "epoch": 0.8515919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1301,
      "step": 24875
    },
    {
      "epoch": 0.8516261554262239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1254,
      "step": 24876
    },
    {
      "epoch": 0.8516603902773023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1173,
      "step": 24877
    },
    {
      "epoch": 0.8516946251283807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1387,
      "step": 24878
    },
    {
      "epoch": 0.8517288599794591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1316,
      "step": 24879
    },
    {
      "epoch": 0.8517630948305375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1252,
      "step": 24880
    },
    {
      "epoch": 0.8517973296816159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1384,
      "step": 24881
    },
    {
      "epoch": 0.8518315645326943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1324,
      "step": 24882
    },
    {
      "epoch": 0.8518657993837727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1452,
      "step": 24883
    },
    {
      "epoch": 0.8519000342348511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1387,
      "step": 24884
    },
    {
      "epoch": 0.8519342690859295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1207,
      "step": 24885
    },
    {
      "epoch": 0.8519685039370078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1426,
      "step": 24886
    },
    {
      "epoch": 0.8520027387880863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1169,
      "step": 24887
    },
    {
      "epoch": 0.8520369736391646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1188,
      "step": 24888
    },
    {
      "epoch": 0.8520712084902431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 24889
    },
    {
      "epoch": 0.8521054433413214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1378,
      "step": 24890
    },
    {
      "epoch": 0.8521396781923999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1303,
      "step": 24891
    },
    {
      "epoch": 0.8521739130434782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1205,
      "step": 24892
    },
    {
      "epoch": 0.8522081478945567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1285,
      "step": 24893
    },
    {
      "epoch": 0.852242382745635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1311,
      "step": 24894
    },
    {
      "epoch": 0.8522766175967135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1304,
      "step": 24895
    },
    {
      "epoch": 0.8523108524477918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.118,
      "step": 24896
    },
    {
      "epoch": 0.8523450872988703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1497,
      "step": 24897
    },
    {
      "epoch": 0.8523793221499486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1359,
      "step": 24898
    },
    {
      "epoch": 0.8524135570010271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1259,
      "step": 24899
    },
    {
      "epoch": 0.8524477918521054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1294,
      "step": 24900
    },
    {
      "epoch": 0.8524820267031838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1382,
      "step": 24901
    },
    {
      "epoch": 0.8525162615542622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1331,
      "step": 24902
    },
    {
      "epoch": 0.8525504964053406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1187,
      "step": 24903
    },
    {
      "epoch": 0.852584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.149,
      "step": 24904
    },
    {
      "epoch": 0.8526189661074974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1236,
      "step": 24905
    },
    {
      "epoch": 0.8526532009585759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1249,
      "step": 24906
    },
    {
      "epoch": 0.8526874358096542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1058,
      "step": 24907
    },
    {
      "epoch": 0.8527216706607327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1327,
      "step": 24908
    },
    {
      "epoch": 0.852755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.119,
      "step": 24909
    },
    {
      "epoch": 0.8527901403628895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1265,
      "step": 24910
    },
    {
      "epoch": 0.8528243752139678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1157,
      "step": 24911
    },
    {
      "epoch": 0.8528586100650463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1233,
      "step": 24912
    },
    {
      "epoch": 0.8528928449161246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1242,
      "step": 24913
    },
    {
      "epoch": 0.8529270797672031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1237,
      "step": 24914
    },
    {
      "epoch": 0.8529613146182814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1422,
      "step": 24915
    },
    {
      "epoch": 0.8529955494693598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1315,
      "step": 24916
    },
    {
      "epoch": 0.8530297843204382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.142,
      "step": 24917
    },
    {
      "epoch": 0.8530640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1244,
      "step": 24918
    },
    {
      "epoch": 0.853098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1271,
      "step": 24919
    },
    {
      "epoch": 0.8531324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1315,
      "step": 24920
    },
    {
      "epoch": 0.8531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1341,
      "step": 24921
    },
    {
      "epoch": 0.8532009585758302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1285,
      "step": 24922
    },
    {
      "epoch": 0.8532351934269086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1261,
      "step": 24923
    },
    {
      "epoch": 0.853269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1215,
      "step": 24924
    },
    {
      "epoch": 0.8533036631290654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1201,
      "step": 24925
    },
    {
      "epoch": 0.8533378979801438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1214,
      "step": 24926
    },
    {
      "epoch": 0.8533721328312222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 24927
    },
    {
      "epoch": 0.8534063676823006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1319,
      "step": 24928
    },
    {
      "epoch": 0.853440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1288,
      "step": 24929
    },
    {
      "epoch": 0.8534748373844574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1126,
      "step": 24930
    },
    {
      "epoch": 0.8535090722355357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1227,
      "step": 24931
    },
    {
      "epoch": 0.8535433070866142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1387,
      "step": 24932
    },
    {
      "epoch": 0.8535775419376925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1375,
      "step": 24933
    },
    {
      "epoch": 0.853611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1019,
      "step": 24934
    },
    {
      "epoch": 0.8536460116398493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1433,
      "step": 24935
    },
    {
      "epoch": 0.8536802464909278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1198,
      "step": 24936
    },
    {
      "epoch": 0.8537144813420061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.13,
      "step": 24937
    },
    {
      "epoch": 0.8537487161930846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1167,
      "step": 24938
    },
    {
      "epoch": 0.8537829510441629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1244,
      "step": 24939
    },
    {
      "epoch": 0.8538171858952414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1443,
      "step": 24940
    },
    {
      "epoch": 0.8538514207463197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1122,
      "step": 24941
    },
    {
      "epoch": 0.8538856555973982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1316,
      "step": 24942
    },
    {
      "epoch": 0.8539198904484766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1329,
      "step": 24943
    },
    {
      "epoch": 0.853954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1295,
      "step": 24944
    },
    {
      "epoch": 0.8539883601506334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1179,
      "step": 24945
    },
    {
      "epoch": 0.8540225950017117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1198,
      "step": 24946
    },
    {
      "epoch": 0.8540568298527902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1461,
      "step": 24947
    },
    {
      "epoch": 0.8540910647038685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1269,
      "step": 24948
    },
    {
      "epoch": 0.854125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1286,
      "step": 24949
    },
    {
      "epoch": 0.8541595344060253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1216,
      "step": 24950
    },
    {
      "epoch": 0.8541937692571038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1247,
      "step": 24951
    },
    {
      "epoch": 0.8542280041081821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.123,
      "step": 24952
    },
    {
      "epoch": 0.8542622389592606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1525,
      "step": 24953
    },
    {
      "epoch": 0.8542964738103389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 24954
    },
    {
      "epoch": 0.8543307086614174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1404,
      "step": 24955
    },
    {
      "epoch": 0.8543649435124957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1227,
      "step": 24956
    },
    {
      "epoch": 0.8543991783635742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1345,
      "step": 24957
    },
    {
      "epoch": 0.8544334132146525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1255,
      "step": 24958
    },
    {
      "epoch": 0.854467648065731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1364,
      "step": 24959
    },
    {
      "epoch": 0.8545018829168093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1388,
      "step": 24960
    },
    {
      "epoch": 0.8545361177678877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1352,
      "step": 24961
    },
    {
      "epoch": 0.8545703526189661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.145,
      "step": 24962
    },
    {
      "epoch": 0.8546045874700445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1185,
      "step": 24963
    },
    {
      "epoch": 0.8546388223211229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1283,
      "step": 24964
    },
    {
      "epoch": 0.8546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1358,
      "step": 24965
    },
    {
      "epoch": 0.8547072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0992,
      "step": 24966
    },
    {
      "epoch": 0.8547415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1179,
      "step": 24967
    },
    {
      "epoch": 0.8547757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1247,
      "step": 24968
    },
    {
      "epoch": 0.8548099965765149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1079,
      "step": 24969
    },
    {
      "epoch": 0.8548442314275932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1366,
      "step": 24970
    },
    {
      "epoch": 0.8548784662786717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1345,
      "step": 24971
    },
    {
      "epoch": 0.85491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1142,
      "step": 24972
    },
    {
      "epoch": 0.8549469359808285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1152,
      "step": 24973
    },
    {
      "epoch": 0.8549811708319068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1319,
      "step": 24974
    },
    {
      "epoch": 0.8550154056829853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1265,
      "step": 24975
    },
    {
      "epoch": 0.8550496405340636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1232,
      "step": 24976
    },
    {
      "epoch": 0.8550838753851421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1375,
      "step": 24977
    },
    {
      "epoch": 0.8551181102362204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 24978
    },
    {
      "epoch": 0.8551523450872989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1172,
      "step": 24979
    },
    {
      "epoch": 0.8551865799383773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1255,
      "step": 24980
    },
    {
      "epoch": 0.8552208147894557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1115,
      "step": 24981
    },
    {
      "epoch": 0.8552550496405341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 24982
    },
    {
      "epoch": 0.8552892844916125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1255,
      "step": 24983
    },
    {
      "epoch": 0.8553235193426909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1187,
      "step": 24984
    },
    {
      "epoch": 0.8553577541937692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1293,
      "step": 24985
    },
    {
      "epoch": 0.8553919890448477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1306,
      "step": 24986
    },
    {
      "epoch": 0.855426223895926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1308,
      "step": 24987
    },
    {
      "epoch": 0.8554604587470045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.134,
      "step": 24988
    },
    {
      "epoch": 0.8554946935980828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1392,
      "step": 24989
    },
    {
      "epoch": 0.8555289284491613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1226,
      "step": 24990
    },
    {
      "epoch": 0.8555631633002396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1489,
      "step": 24991
    },
    {
      "epoch": 0.8555973981513181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1357,
      "step": 24992
    },
    {
      "epoch": 0.8556316330023964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1315,
      "step": 24993
    },
    {
      "epoch": 0.8556658678534749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1172,
      "step": 24994
    },
    {
      "epoch": 0.8557001027045532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1368,
      "step": 24995
    },
    {
      "epoch": 0.8557343375556317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1266,
      "step": 24996
    },
    {
      "epoch": 0.85576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1223,
      "step": 24997
    },
    {
      "epoch": 0.8558028072577885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.122,
      "step": 24998
    },
    {
      "epoch": 0.8558370421088668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1244,
      "step": 24999
    },
    {
      "epoch": 0.8558712769599452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1348,
      "step": 25000
    },
    {
      "epoch": 0.8559055118110236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1287,
      "step": 25001
    },
    {
      "epoch": 0.855939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.103,
      "step": 25002
    },
    {
      "epoch": 0.8559739815131804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1286,
      "step": 25003
    },
    {
      "epoch": 0.8560082163642588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1181,
      "step": 25004
    },
    {
      "epoch": 0.8560424512153372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1299,
      "step": 25005
    },
    {
      "epoch": 0.8560766860664156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.131,
      "step": 25006
    },
    {
      "epoch": 0.856110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1184,
      "step": 25007
    },
    {
      "epoch": 0.8561451557685724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1346,
      "step": 25008
    },
    {
      "epoch": 0.8561793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1179,
      "step": 25009
    },
    {
      "epoch": 0.8562136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1283,
      "step": 25010
    },
    {
      "epoch": 0.8562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1222,
      "step": 25011
    },
    {
      "epoch": 0.856282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1336,
      "step": 25012
    },
    {
      "epoch": 0.8563163300239643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1229,
      "step": 25013
    },
    {
      "epoch": 0.8563505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1295,
      "step": 25014
    },
    {
      "epoch": 0.8563847997261211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.122,
      "step": 25015
    },
    {
      "epoch": 0.8564190345771996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1279,
      "step": 25016
    },
    {
      "epoch": 0.856453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1194,
      "step": 25017
    },
    {
      "epoch": 0.8564875042793564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1176,
      "step": 25018
    },
    {
      "epoch": 0.8565217391304348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1152,
      "step": 25019
    },
    {
      "epoch": 0.8565559739815132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1083,
      "step": 25020
    },
    {
      "epoch": 0.8565902088325916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1121,
      "step": 25021
    },
    {
      "epoch": 0.85662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1161,
      "step": 25022
    },
    {
      "epoch": 0.8566586785347484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1378,
      "step": 25023
    },
    {
      "epoch": 0.8566929133858268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1245,
      "step": 25024
    },
    {
      "epoch": 0.8567271482369052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1347,
      "step": 25025
    },
    {
      "epoch": 0.8567613830879836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.118,
      "step": 25026
    },
    {
      "epoch": 0.856795617939062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1105,
      "step": 25027
    },
    {
      "epoch": 0.8568298527901403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1291,
      "step": 25028
    },
    {
      "epoch": 0.8568640876412188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1312,
      "step": 25029
    },
    {
      "epoch": 0.8568983224922971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1234,
      "step": 25030
    },
    {
      "epoch": 0.8569325573433756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1133,
      "step": 25031
    },
    {
      "epoch": 0.8569667921944539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1182,
      "step": 25032
    },
    {
      "epoch": 0.8570010270455324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.117,
      "step": 25033
    },
    {
      "epoch": 0.8570352618966107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.127,
      "step": 25034
    },
    {
      "epoch": 0.8570694967476892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.122,
      "step": 25035
    },
    {
      "epoch": 0.8571037315987675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 25036
    },
    {
      "epoch": 0.857137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1334,
      "step": 25037
    },
    {
      "epoch": 0.8571722013009243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1216,
      "step": 25038
    },
    {
      "epoch": 0.8572064361520028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1135,
      "step": 25039
    },
    {
      "epoch": 0.8572406710030811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1386,
      "step": 25040
    },
    {
      "epoch": 0.8572749058541596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1196,
      "step": 25041
    },
    {
      "epoch": 0.8573091407052379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1219,
      "step": 25042
    },
    {
      "epoch": 0.8573433755563163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1362,
      "step": 25043
    },
    {
      "epoch": 0.8573776104073947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1393,
      "step": 25044
    },
    {
      "epoch": 0.8574118452584731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1183,
      "step": 25045
    },
    {
      "epoch": 0.8574460801095515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1256,
      "step": 25046
    },
    {
      "epoch": 0.8574803149606299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1355,
      "step": 25047
    },
    {
      "epoch": 0.8575145498117083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1404,
      "step": 25048
    },
    {
      "epoch": 0.8575487846627867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1266,
      "step": 25049
    },
    {
      "epoch": 0.8575830195138651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1263,
      "step": 25050
    },
    {
      "epoch": 0.8576172543649435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1134,
      "step": 25051
    },
    {
      "epoch": 0.8576514892160219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.122,
      "step": 25052
    },
    {
      "epoch": 0.8576857240671003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1158,
      "step": 25053
    },
    {
      "epoch": 0.8577199589181788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1345,
      "step": 25054
    },
    {
      "epoch": 0.8577541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1214,
      "step": 25055
    },
    {
      "epoch": 0.8577884286203356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1438,
      "step": 25056
    },
    {
      "epoch": 0.8578226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1106,
      "step": 25057
    },
    {
      "epoch": 0.8578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1299,
      "step": 25058
    },
    {
      "epoch": 0.8578911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1257,
      "step": 25059
    },
    {
      "epoch": 0.8579253680246491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1163,
      "step": 25060
    },
    {
      "epoch": 0.8579596028757275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1234,
      "step": 25061
    },
    {
      "epoch": 0.8579938377268059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.135,
      "step": 25062
    },
    {
      "epoch": 0.8580280725778843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1192,
      "step": 25063
    },
    {
      "epoch": 0.8580623074289627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1205,
      "step": 25064
    },
    {
      "epoch": 0.8580965422800411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1281,
      "step": 25065
    },
    {
      "epoch": 0.8581307771311195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1346,
      "step": 25066
    },
    {
      "epoch": 0.8581650119821979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1319,
      "step": 25067
    },
    {
      "epoch": 0.8581992468332763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1242,
      "step": 25068
    },
    {
      "epoch": 0.8582334816843546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1212,
      "step": 25069
    },
    {
      "epoch": 0.8582677165354331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1438,
      "step": 25070
    },
    {
      "epoch": 0.8583019513865114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1252,
      "step": 25071
    },
    {
      "epoch": 0.8583361862375899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.122,
      "step": 25072
    },
    {
      "epoch": 0.8583704210886682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1234,
      "step": 25073
    },
    {
      "epoch": 0.8584046559397467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1364,
      "step": 25074
    },
    {
      "epoch": 0.858438890790825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1204,
      "step": 25075
    },
    {
      "epoch": 0.8584731256419035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1145,
      "step": 25076
    },
    {
      "epoch": 0.8585073604929818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.128,
      "step": 25077
    },
    {
      "epoch": 0.8585415953440603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1328,
      "step": 25078
    },
    {
      "epoch": 0.8585758301951386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1193,
      "step": 25079
    },
    {
      "epoch": 0.8586100650462171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1258,
      "step": 25080
    },
    {
      "epoch": 0.8586442998972954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1302,
      "step": 25081
    },
    {
      "epoch": 0.8586785347483739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1253,
      "step": 25082
    },
    {
      "epoch": 0.8587127695994522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1235,
      "step": 25083
    },
    {
      "epoch": 0.8587470044505306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1223,
      "step": 25084
    },
    {
      "epoch": 0.858781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1481,
      "step": 25085
    },
    {
      "epoch": 0.8588154741526874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 25086
    },
    {
      "epoch": 0.8588497090037658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1202,
      "step": 25087
    },
    {
      "epoch": 0.8588839438548442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1342,
      "step": 25088
    },
    {
      "epoch": 0.8589181787059226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1279,
      "step": 25089
    },
    {
      "epoch": 0.858952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1223,
      "step": 25090
    },
    {
      "epoch": 0.8589866484080795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1191,
      "step": 25091
    },
    {
      "epoch": 0.8590208832591578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1268,
      "step": 25092
    },
    {
      "epoch": 0.8590551181102363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1337,
      "step": 25093
    },
    {
      "epoch": 0.8590893529613146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1328,
      "step": 25094
    },
    {
      "epoch": 0.8591235878123931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1237,
      "step": 25095
    },
    {
      "epoch": 0.8591578226634714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1155,
      "step": 25096
    },
    {
      "epoch": 0.8591920575145499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.138,
      "step": 25097
    },
    {
      "epoch": 0.8592262923656282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1235,
      "step": 25098
    },
    {
      "epoch": 0.8592605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1192,
      "step": 25099
    },
    {
      "epoch": 0.859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1244,
      "step": 25100
    },
    {
      "epoch": 0.8593289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1454,
      "step": 25101
    },
    {
      "epoch": 0.8593632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1329,
      "step": 25102
    },
    {
      "epoch": 0.8593974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1297,
      "step": 25103
    },
    {
      "epoch": 0.8594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1489,
      "step": 25104
    },
    {
      "epoch": 0.859465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1262,
      "step": 25105
    },
    {
      "epoch": 0.8595001711742554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1113,
      "step": 25106
    },
    {
      "epoch": 0.8595344060253338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.12,
      "step": 25107
    },
    {
      "epoch": 0.8595686408764122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1153,
      "step": 25108
    },
    {
      "epoch": 0.8596028757274906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1313,
      "step": 25109
    },
    {
      "epoch": 0.859637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1279,
      "step": 25110
    },
    {
      "epoch": 0.8596713454296474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1414,
      "step": 25111
    },
    {
      "epoch": 0.8597055802807257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1176,
      "step": 25112
    },
    {
      "epoch": 0.8597398151318042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1097,
      "step": 25113
    },
    {
      "epoch": 0.8597740499828825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1219,
      "step": 25114
    },
    {
      "epoch": 0.859808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1257,
      "step": 25115
    },
    {
      "epoch": 0.8598425196850393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1436,
      "step": 25116
    },
    {
      "epoch": 0.8598767545361178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1386,
      "step": 25117
    },
    {
      "epoch": 0.8599109893871961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.14,
      "step": 25118
    },
    {
      "epoch": 0.8599452242382746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1204,
      "step": 25119
    },
    {
      "epoch": 0.8599794590893529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1224,
      "step": 25120
    },
    {
      "epoch": 0.8600136939404314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1458,
      "step": 25121
    },
    {
      "epoch": 0.8600479287915097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.137,
      "step": 25122
    },
    {
      "epoch": 0.8600821636425882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1368,
      "step": 25123
    },
    {
      "epoch": 0.8601163984936665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1323,
      "step": 25124
    },
    {
      "epoch": 0.860150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1226,
      "step": 25125
    },
    {
      "epoch": 0.8601848681958233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1144,
      "step": 25126
    },
    {
      "epoch": 0.8602191030469017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1257,
      "step": 25127
    },
    {
      "epoch": 0.8602533378979802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1292,
      "step": 25128
    },
    {
      "epoch": 0.8602875727490585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1348,
      "step": 25129
    },
    {
      "epoch": 0.860321807600137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1265,
      "step": 25130
    },
    {
      "epoch": 0.8603560424512153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1173,
      "step": 25131
    },
    {
      "epoch": 0.8603902773022938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1115,
      "step": 25132
    },
    {
      "epoch": 0.8604245121533721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1291,
      "step": 25133
    },
    {
      "epoch": 0.8604587470044506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1238,
      "step": 25134
    },
    {
      "epoch": 0.8604929818555289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1157,
      "step": 25135
    },
    {
      "epoch": 0.8605272167066074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1211,
      "step": 25136
    },
    {
      "epoch": 0.8605614515576857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1287,
      "step": 25137
    },
    {
      "epoch": 0.8605956864087642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1159,
      "step": 25138
    },
    {
      "epoch": 0.8606299212598425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1108,
      "step": 25139
    },
    {
      "epoch": 0.860664156110921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1406,
      "step": 25140
    },
    {
      "epoch": 0.8606983909619993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1061,
      "step": 25141
    },
    {
      "epoch": 0.8607326258130777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1369,
      "step": 25142
    },
    {
      "epoch": 0.8607668606641561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1301,
      "step": 25143
    },
    {
      "epoch": 0.8608010955152345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1204,
      "step": 25144
    },
    {
      "epoch": 0.8608353303663129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1077,
      "step": 25145
    },
    {
      "epoch": 0.8608695652173913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1327,
      "step": 25146
    },
    {
      "epoch": 0.8609038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1237,
      "step": 25147
    },
    {
      "epoch": 0.8609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1115,
      "step": 25148
    },
    {
      "epoch": 0.8609722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.132,
      "step": 25149
    },
    {
      "epoch": 0.8610065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1436,
      "step": 25150
    },
    {
      "epoch": 0.8610407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1229,
      "step": 25151
    },
    {
      "epoch": 0.8610749743238617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1302,
      "step": 25152
    },
    {
      "epoch": 0.86110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1245,
      "step": 25153
    },
    {
      "epoch": 0.8611434440260185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 25154
    },
    {
      "epoch": 0.8611776788770968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1274,
      "step": 25155
    },
    {
      "epoch": 0.8612119137281753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1259,
      "step": 25156
    },
    {
      "epoch": 0.8612461485792536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1402,
      "step": 25157
    },
    {
      "epoch": 0.8612803834303321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1258,
      "step": 25158
    },
    {
      "epoch": 0.8613146182814104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1239,
      "step": 25159
    },
    {
      "epoch": 0.8613488531324889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1154,
      "step": 25160
    },
    {
      "epoch": 0.8613830879835672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1172,
      "step": 25161
    },
    {
      "epoch": 0.8614173228346457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1414,
      "step": 25162
    },
    {
      "epoch": 0.861451557685724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1193,
      "step": 25163
    },
    {
      "epoch": 0.8614857925368025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1335,
      "step": 25164
    },
    {
      "epoch": 0.8615200273878809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1334,
      "step": 25165
    },
    {
      "epoch": 0.8615542622389593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1259,
      "step": 25166
    },
    {
      "epoch": 0.8615884970900377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.112,
      "step": 25167
    },
    {
      "epoch": 0.861622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1416,
      "step": 25168
    },
    {
      "epoch": 0.8616569667921945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1315,
      "step": 25169
    },
    {
      "epoch": 0.8616912016432728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1325,
      "step": 25170
    },
    {
      "epoch": 0.8617254364943513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1376,
      "step": 25171
    },
    {
      "epoch": 0.8617596713454296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1195,
      "step": 25172
    },
    {
      "epoch": 0.8617939061965081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1173,
      "step": 25173
    },
    {
      "epoch": 0.8618281410475864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1341,
      "step": 25174
    },
    {
      "epoch": 0.8618623758986649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1171,
      "step": 25175
    },
    {
      "epoch": 0.8618966107497432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1336,
      "step": 25176
    },
    {
      "epoch": 0.8619308456008217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.128,
      "step": 25177
    },
    {
      "epoch": 0.8619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1332,
      "step": 25178
    },
    {
      "epoch": 0.8619993153029785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1255,
      "step": 25179
    },
    {
      "epoch": 0.8620335501540568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1225,
      "step": 25180
    },
    {
      "epoch": 0.8620677850051353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1219,
      "step": 25181
    },
    {
      "epoch": 0.8621020198562136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1212,
      "step": 25182
    },
    {
      "epoch": 0.862136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1139,
      "step": 25183
    },
    {
      "epoch": 0.8621704895583704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1291,
      "step": 25184
    },
    {
      "epoch": 0.8622047244094488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1302,
      "step": 25185
    },
    {
      "epoch": 0.8622389592605272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1308,
      "step": 25186
    },
    {
      "epoch": 0.8622731941116056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1146,
      "step": 25187
    },
    {
      "epoch": 0.862307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1219,
      "step": 25188
    },
    {
      "epoch": 0.8623416638137624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1398,
      "step": 25189
    },
    {
      "epoch": 0.8623758986648408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1259,
      "step": 25190
    },
    {
      "epoch": 0.8624101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1275,
      "step": 25191
    },
    {
      "epoch": 0.8624443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1545,
      "step": 25192
    },
    {
      "epoch": 0.862478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1233,
      "step": 25193
    },
    {
      "epoch": 0.8625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1262,
      "step": 25194
    },
    {
      "epoch": 0.8625470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1175,
      "step": 25195
    },
    {
      "epoch": 0.8625813077713111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.126,
      "step": 25196
    },
    {
      "epoch": 0.8626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1251,
      "step": 25197
    },
    {
      "epoch": 0.8626497774734679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1209,
      "step": 25198
    },
    {
      "epoch": 0.8626840123245464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1182,
      "step": 25199
    },
    {
      "epoch": 0.8627182471756247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1121,
      "step": 25200
    },
    {
      "epoch": 0.8627524820267032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1211,
      "step": 25201
    },
    {
      "epoch": 0.8627867168777816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1291,
      "step": 25202
    },
    {
      "epoch": 0.86282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1055,
      "step": 25203
    },
    {
      "epoch": 0.8628551865799384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1062,
      "step": 25204
    },
    {
      "epoch": 0.8628894214310168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1254,
      "step": 25205
    },
    {
      "epoch": 0.8629236562820952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1287,
      "step": 25206
    },
    {
      "epoch": 0.8629578911331736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1163,
      "step": 25207
    },
    {
      "epoch": 0.862992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1132,
      "step": 25208
    },
    {
      "epoch": 0.8630263608353304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1269,
      "step": 25209
    },
    {
      "epoch": 0.8630605956864088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1217,
      "step": 25210
    },
    {
      "epoch": 0.8630948305374871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1236,
      "step": 25211
    },
    {
      "epoch": 0.8631290653885656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1429,
      "step": 25212
    },
    {
      "epoch": 0.8631633002396439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1279,
      "step": 25213
    },
    {
      "epoch": 0.8631975350907224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1258,
      "step": 25214
    },
    {
      "epoch": 0.8632317699418007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1201,
      "step": 25215
    },
    {
      "epoch": 0.8632660047928792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.119,
      "step": 25216
    },
    {
      "epoch": 0.8633002396439575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1234,
      "step": 25217
    },
    {
      "epoch": 0.863334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1253,
      "step": 25218
    },
    {
      "epoch": 0.8633687093461143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1134,
      "step": 25219
    },
    {
      "epoch": 0.8634029441971928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1285,
      "step": 25220
    },
    {
      "epoch": 0.8634371790482711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1298,
      "step": 25221
    },
    {
      "epoch": 0.8634714138993496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1243,
      "step": 25222
    },
    {
      "epoch": 0.8635056487504279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1131,
      "step": 25223
    },
    {
      "epoch": 0.8635398836015064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1113,
      "step": 25224
    },
    {
      "epoch": 0.8635741184525847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.126,
      "step": 25225
    },
    {
      "epoch": 0.8636083533036631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1165,
      "step": 25226
    },
    {
      "epoch": 0.8636425881547415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1281,
      "step": 25227
    },
    {
      "epoch": 0.8636768230058199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1315,
      "step": 25228
    },
    {
      "epoch": 0.8637110578568983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1249,
      "step": 25229
    },
    {
      "epoch": 0.8637452927079767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1193,
      "step": 25230
    },
    {
      "epoch": 0.8637795275590551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1043,
      "step": 25231
    },
    {
      "epoch": 0.8638137624101335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 25232
    },
    {
      "epoch": 0.8638479972612119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1159,
      "step": 25233
    },
    {
      "epoch": 0.8638822321122903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1366,
      "step": 25234
    },
    {
      "epoch": 0.8639164669633687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1364,
      "step": 25235
    },
    {
      "epoch": 0.8639507018144471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1104,
      "step": 25236
    },
    {
      "epoch": 0.8639849366655256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1279,
      "step": 25237
    },
    {
      "epoch": 0.8640191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1184,
      "step": 25238
    },
    {
      "epoch": 0.8640534063676824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1188,
      "step": 25239
    },
    {
      "epoch": 0.8640876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1328,
      "step": 25240
    },
    {
      "epoch": 0.8641218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1195,
      "step": 25241
    },
    {
      "epoch": 0.8641561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1181,
      "step": 25242
    },
    {
      "epoch": 0.8641903457719959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1192,
      "step": 25243
    },
    {
      "epoch": 0.8642245806230743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1217,
      "step": 25244
    },
    {
      "epoch": 0.8642588154741527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1154,
      "step": 25245
    },
    {
      "epoch": 0.8642930503252311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1142,
      "step": 25246
    },
    {
      "epoch": 0.8643272851763095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1328,
      "step": 25247
    },
    {
      "epoch": 0.8643615200273879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1268,
      "step": 25248
    },
    {
      "epoch": 0.8643957548784663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1426,
      "step": 25249
    },
    {
      "epoch": 0.8644299897295447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1318,
      "step": 25250
    },
    {
      "epoch": 0.8644642245806231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1311,
      "step": 25251
    },
    {
      "epoch": 0.8644984594317014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1119,
      "step": 25252
    },
    {
      "epoch": 0.8645326942827799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1124,
      "step": 25253
    },
    {
      "epoch": 0.8645669291338582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1361,
      "step": 25254
    },
    {
      "epoch": 0.8646011639849367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.127,
      "step": 25255
    },
    {
      "epoch": 0.864635398836015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1158,
      "step": 25256
    },
    {
      "epoch": 0.8646696336870935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1231,
      "step": 25257
    },
    {
      "epoch": 0.8647038685381718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1239,
      "step": 25258
    },
    {
      "epoch": 0.8647381033892503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1063,
      "step": 25259
    },
    {
      "epoch": 0.8647723382403286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1175,
      "step": 25260
    },
    {
      "epoch": 0.8648065730914071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1362,
      "step": 25261
    },
    {
      "epoch": 0.8648408079424854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1487,
      "step": 25262
    },
    {
      "epoch": 0.8648750427935639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1287,
      "step": 25263
    },
    {
      "epoch": 0.8649092776446422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1189,
      "step": 25264
    },
    {
      "epoch": 0.8649435124957207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1079,
      "step": 25265
    },
    {
      "epoch": 0.864977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1015,
      "step": 25266
    },
    {
      "epoch": 0.8650119821978774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1075,
      "step": 25267
    },
    {
      "epoch": 0.8650462170489558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1243,
      "step": 25268
    },
    {
      "epoch": 0.8650804519000342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1207,
      "step": 25269
    },
    {
      "epoch": 0.8651146867511126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1192,
      "step": 25270
    },
    {
      "epoch": 0.865148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1111,
      "step": 25271
    },
    {
      "epoch": 0.8651831564532694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1133,
      "step": 25272
    },
    {
      "epoch": 0.8652173913043478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1323,
      "step": 25273
    },
    {
      "epoch": 0.8652516261554263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1258,
      "step": 25274
    },
    {
      "epoch": 0.8652858610065046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1205,
      "step": 25275
    },
    {
      "epoch": 0.8653200958575831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1046,
      "step": 25276
    },
    {
      "epoch": 0.8653543307086614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1255,
      "step": 25277
    },
    {
      "epoch": 0.8653885655597399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1129,
      "step": 25278
    },
    {
      "epoch": 0.8654228004108182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1155,
      "step": 25279
    },
    {
      "epoch": 0.8654570352618967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1128,
      "step": 25280
    },
    {
      "epoch": 0.865491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1305,
      "step": 25281
    },
    {
      "epoch": 0.8655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1052,
      "step": 25282
    },
    {
      "epoch": 0.8655597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.109,
      "step": 25283
    },
    {
      "epoch": 0.8655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1185,
      "step": 25284
    },
    {
      "epoch": 0.8656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1228,
      "step": 25285
    },
    {
      "epoch": 0.865662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1192,
      "step": 25286
    },
    {
      "epoch": 0.8656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1305,
      "step": 25287
    },
    {
      "epoch": 0.8657309140705238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1225,
      "step": 25288
    },
    {
      "epoch": 0.8657651489216022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1047,
      "step": 25289
    },
    {
      "epoch": 0.8657993837726806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1135,
      "step": 25290
    },
    {
      "epoch": 0.865833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1244,
      "step": 25291
    },
    {
      "epoch": 0.8658678534748374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.144,
      "step": 25292
    },
    {
      "epoch": 0.8659020883259158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1252,
      "step": 25293
    },
    {
      "epoch": 0.8659363231769942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1064,
      "step": 25294
    },
    {
      "epoch": 0.8659705580280725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1025,
      "step": 25295
    },
    {
      "epoch": 0.866004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 25296
    },
    {
      "epoch": 0.8660390277302293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1359,
      "step": 25297
    },
    {
      "epoch": 0.8660732625813078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1238,
      "step": 25298
    },
    {
      "epoch": 0.8661074974323861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 25299
    },
    {
      "epoch": 0.8661417322834646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1274,
      "step": 25300
    },
    {
      "epoch": 0.8661759671345429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1276,
      "step": 25301
    },
    {
      "epoch": 0.8662102019856214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.133,
      "step": 25302
    },
    {
      "epoch": 0.8662444368366997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1222,
      "step": 25303
    },
    {
      "epoch": 0.8662786716877782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1059,
      "step": 25304
    },
    {
      "epoch": 0.8663129065388565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1288,
      "step": 25305
    },
    {
      "epoch": 0.866347141389935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1174,
      "step": 25306
    },
    {
      "epoch": 0.8663813762410133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1109,
      "step": 25307
    },
    {
      "epoch": 0.8664156110920918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1147,
      "step": 25308
    },
    {
      "epoch": 0.8664498459431701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1137,
      "step": 25309
    },
    {
      "epoch": 0.8664840807942485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1154,
      "step": 25310
    },
    {
      "epoch": 0.866518315645327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1376,
      "step": 25311
    },
    {
      "epoch": 0.8665525504964053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1285,
      "step": 25312
    },
    {
      "epoch": 0.8665867853474838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1364,
      "step": 25313
    },
    {
      "epoch": 0.8666210201985621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1371,
      "step": 25314
    },
    {
      "epoch": 0.8666552550496406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.123,
      "step": 25315
    },
    {
      "epoch": 0.8666894899007189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1334,
      "step": 25316
    },
    {
      "epoch": 0.8667237247517974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1213,
      "step": 25317
    },
    {
      "epoch": 0.8667579596028757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1286,
      "step": 25318
    },
    {
      "epoch": 0.8667921944539542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1362,
      "step": 25319
    },
    {
      "epoch": 0.8668264293050325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1042,
      "step": 25320
    },
    {
      "epoch": 0.866860664156111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1166,
      "step": 25321
    },
    {
      "epoch": 0.8668948990071893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1118,
      "step": 25322
    },
    {
      "epoch": 0.8669291338582678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1123,
      "step": 25323
    },
    {
      "epoch": 0.8669633687093461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1179,
      "step": 25324
    },
    {
      "epoch": 0.8669976035604245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1227,
      "step": 25325
    },
    {
      "epoch": 0.8670318384115029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1266,
      "step": 25326
    },
    {
      "epoch": 0.8670660732625813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1243,
      "step": 25327
    },
    {
      "epoch": 0.8671003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1238,
      "step": 25328
    },
    {
      "epoch": 0.8671345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1124,
      "step": 25329
    },
    {
      "epoch": 0.8671687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1086,
      "step": 25330
    },
    {
      "epoch": 0.8672030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1238,
      "step": 25331
    },
    {
      "epoch": 0.8672372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1242,
      "step": 25332
    },
    {
      "epoch": 0.8672714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1165,
      "step": 25333
    },
    {
      "epoch": 0.8673057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.124,
      "step": 25334
    },
    {
      "epoch": 0.8673399520712085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1233,
      "step": 25335
    },
    {
      "epoch": 0.8673741869222868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1322,
      "step": 25336
    },
    {
      "epoch": 0.8674084217733653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1318,
      "step": 25337
    },
    {
      "epoch": 0.8674426566244436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1105,
      "step": 25338
    },
    {
      "epoch": 0.8674768914755221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1305,
      "step": 25339
    },
    {
      "epoch": 0.8675111263266004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1267,
      "step": 25340
    },
    {
      "epoch": 0.8675453611776789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1229,
      "step": 25341
    },
    {
      "epoch": 0.8675795960287572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.119,
      "step": 25342
    },
    {
      "epoch": 0.8676138308798357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1077,
      "step": 25343
    },
    {
      "epoch": 0.867648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1323,
      "step": 25344
    },
    {
      "epoch": 0.8676823005819925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1161,
      "step": 25345
    },
    {
      "epoch": 0.8677165354330708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1227,
      "step": 25346
    },
    {
      "epoch": 0.8677507702841493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1176,
      "step": 25347
    },
    {
      "epoch": 0.8677850051352277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.084,
      "step": 25348
    },
    {
      "epoch": 0.8678192399863061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1203,
      "step": 25349
    },
    {
      "epoch": 0.8678534748373845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1173,
      "step": 25350
    },
    {
      "epoch": 0.8678877096884628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 25351
    },
    {
      "epoch": 0.8679219445395413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1231,
      "step": 25352
    },
    {
      "epoch": 0.8679561793906196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1248,
      "step": 25353
    },
    {
      "epoch": 0.8679904142416981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1192,
      "step": 25354
    },
    {
      "epoch": 0.8680246490927764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1262,
      "step": 25355
    },
    {
      "epoch": 0.8680588839438549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1091,
      "step": 25356
    },
    {
      "epoch": 0.8680931187949332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1255,
      "step": 25357
    },
    {
      "epoch": 0.8681273536460117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.131,
      "step": 25358
    },
    {
      "epoch": 0.86816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1232,
      "step": 25359
    },
    {
      "epoch": 0.8681958233481685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1171,
      "step": 25360
    },
    {
      "epoch": 0.8682300581992468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1302,
      "step": 25361
    },
    {
      "epoch": 0.8682642930503253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.124,
      "step": 25362
    },
    {
      "epoch": 0.8682985279014036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1132,
      "step": 25363
    },
    {
      "epoch": 0.8683327627524821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1161,
      "step": 25364
    },
    {
      "epoch": 0.8683669976035604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1237,
      "step": 25365
    },
    {
      "epoch": 0.8684012324546388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1314,
      "step": 25366
    },
    {
      "epoch": 0.8684354673057172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1146,
      "step": 25367
    },
    {
      "epoch": 0.8684697021567956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1073,
      "step": 25368
    },
    {
      "epoch": 0.868503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1333,
      "step": 25369
    },
    {
      "epoch": 0.8685381718589524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1155,
      "step": 25370
    },
    {
      "epoch": 0.8685724067100308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1196,
      "step": 25371
    },
    {
      "epoch": 0.8686066415611092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1311,
      "step": 25372
    },
    {
      "epoch": 0.8686408764121876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1382,
      "step": 25373
    },
    {
      "epoch": 0.868675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 25374
    },
    {
      "epoch": 0.8687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1219,
      "step": 25375
    },
    {
      "epoch": 0.8687435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.123,
      "step": 25376
    },
    {
      "epoch": 0.8687778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.127,
      "step": 25377
    },
    {
      "epoch": 0.8688120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1457,
      "step": 25378
    },
    {
      "epoch": 0.868846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1129,
      "step": 25379
    },
    {
      "epoch": 0.8688805203697364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1156,
      "step": 25380
    },
    {
      "epoch": 0.8689147552208147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1262,
      "step": 25381
    },
    {
      "epoch": 0.8689489900718932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.106,
      "step": 25382
    },
    {
      "epoch": 0.8689832249229715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1406,
      "step": 25383
    },
    {
      "epoch": 0.86901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1221,
      "step": 25384
    },
    {
      "epoch": 0.8690516946251284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1338,
      "step": 25385
    },
    {
      "epoch": 0.8690859294762068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.131,
      "step": 25386
    },
    {
      "epoch": 0.8691201643272852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1258,
      "step": 25387
    },
    {
      "epoch": 0.8691543991783636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1156,
      "step": 25388
    },
    {
      "epoch": 0.869188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1251,
      "step": 25389
    },
    {
      "epoch": 0.8692228688805204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1189,
      "step": 25390
    },
    {
      "epoch": 0.8692571037315988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1025,
      "step": 25391
    },
    {
      "epoch": 0.8692913385826772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1233,
      "step": 25392
    },
    {
      "epoch": 0.8693255734337556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1416,
      "step": 25393
    },
    {
      "epoch": 0.869359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1292,
      "step": 25394
    },
    {
      "epoch": 0.8693940431359124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1099,
      "step": 25395
    },
    {
      "epoch": 0.8694282779869907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1176,
      "step": 25396
    },
    {
      "epoch": 0.8694625128380692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1285,
      "step": 25397
    },
    {
      "epoch": 0.8694967476891475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1329,
      "step": 25398
    },
    {
      "epoch": 0.869530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.112,
      "step": 25399
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1283,
      "step": 25400
    },
    {
      "epoch": 0.8695994522423828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1222,
      "step": 25401
    },
    {
      "epoch": 0.8696336870934611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1119,
      "step": 25402
    },
    {
      "epoch": 0.8696679219445396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1145,
      "step": 25403
    },
    {
      "epoch": 0.8697021567956179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1203,
      "step": 25404
    },
    {
      "epoch": 0.8697363916466964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1049,
      "step": 25405
    },
    {
      "epoch": 0.8697706264977747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1321,
      "step": 25406
    },
    {
      "epoch": 0.8698048613488532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1072,
      "step": 25407
    },
    {
      "epoch": 0.8698390961999315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1233,
      "step": 25408
    },
    {
      "epoch": 0.86987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.114,
      "step": 25409
    },
    {
      "epoch": 0.8699075659020883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1321,
      "step": 25410
    },
    {
      "epoch": 0.8699418007531667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1351,
      "step": 25411
    },
    {
      "epoch": 0.8699760356042451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1285,
      "step": 25412
    },
    {
      "epoch": 0.8700102704553235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1223,
      "step": 25413
    },
    {
      "epoch": 0.8700445053064019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1214,
      "step": 25414
    },
    {
      "epoch": 0.8700787401574803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1161,
      "step": 25415
    },
    {
      "epoch": 0.8701129750085587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1078,
      "step": 25416
    },
    {
      "epoch": 0.8701472098596371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1095,
      "step": 25417
    },
    {
      "epoch": 0.8701814447107155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.113,
      "step": 25418
    },
    {
      "epoch": 0.8702156795617939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1107,
      "step": 25419
    },
    {
      "epoch": 0.8702499144128722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 25420
    },
    {
      "epoch": 0.8702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1226,
      "step": 25421
    },
    {
      "epoch": 0.8703183841150292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1353,
      "step": 25422
    },
    {
      "epoch": 0.8703526189661075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1245,
      "step": 25423
    },
    {
      "epoch": 0.870386853817186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1166,
      "step": 25424
    },
    {
      "epoch": 0.8704210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1185,
      "step": 25425
    },
    {
      "epoch": 0.8704553235193427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1249,
      "step": 25426
    },
    {
      "epoch": 0.8704895583704211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0956,
      "step": 25427
    },
    {
      "epoch": 0.8705237932214995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1346,
      "step": 25428
    },
    {
      "epoch": 0.8705580280725779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1128,
      "step": 25429
    },
    {
      "epoch": 0.8705922629236563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 25430
    },
    {
      "epoch": 0.8706264977747347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1366,
      "step": 25431
    },
    {
      "epoch": 0.8706607326258131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1199,
      "step": 25432
    },
    {
      "epoch": 0.8706949674768915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.112,
      "step": 25433
    },
    {
      "epoch": 0.8707292023279699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1259,
      "step": 25434
    },
    {
      "epoch": 0.8707634371790482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1157,
      "step": 25435
    },
    {
      "epoch": 0.8707976720301267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1154,
      "step": 25436
    },
    {
      "epoch": 0.870831906881205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1079,
      "step": 25437
    },
    {
      "epoch": 0.8708661417322835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1121,
      "step": 25438
    },
    {
      "epoch": 0.8709003765833618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1058,
      "step": 25439
    },
    {
      "epoch": 0.8709346114344403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1144,
      "step": 25440
    },
    {
      "epoch": 0.8709688462855186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1207,
      "step": 25441
    },
    {
      "epoch": 0.8710030811365971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1184,
      "step": 25442
    },
    {
      "epoch": 0.8710373159876754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1372,
      "step": 25443
    },
    {
      "epoch": 0.8710715508387539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.125,
      "step": 25444
    },
    {
      "epoch": 0.8711057856898322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1205,
      "step": 25445
    },
    {
      "epoch": 0.8711400205409107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1183,
      "step": 25446
    },
    {
      "epoch": 0.871174255391989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1136,
      "step": 25447
    },
    {
      "epoch": 0.8712084902430675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1176,
      "step": 25448
    },
    {
      "epoch": 0.8712427250941458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1204,
      "step": 25449
    },
    {
      "epoch": 0.8712769599452242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1323,
      "step": 25450
    },
    {
      "epoch": 0.8713111947963026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1149,
      "step": 25451
    },
    {
      "epoch": 0.871345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1116,
      "step": 25452
    },
    {
      "epoch": 0.8713796644984594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.12,
      "step": 25453
    },
    {
      "epoch": 0.8714138993495378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.116,
      "step": 25454
    },
    {
      "epoch": 0.8714481342006162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1234,
      "step": 25455
    },
    {
      "epoch": 0.8714823690516946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1391,
      "step": 25456
    },
    {
      "epoch": 0.871516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1143,
      "step": 25457
    },
    {
      "epoch": 0.8715508387538514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1068,
      "step": 25458
    },
    {
      "epoch": 0.8715850736049299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1204,
      "step": 25459
    },
    {
      "epoch": 0.8716193084560082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1113,
      "step": 25460
    },
    {
      "epoch": 0.8716535433070867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1078,
      "step": 25461
    },
    {
      "epoch": 0.871687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1271,
      "step": 25462
    },
    {
      "epoch": 0.8717220130092435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 25463
    },
    {
      "epoch": 0.8717562478603218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 25464
    },
    {
      "epoch": 0.8717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1256,
      "step": 25465
    },
    {
      "epoch": 0.8718247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1201,
      "step": 25466
    },
    {
      "epoch": 0.871858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0977,
      "step": 25467
    },
    {
      "epoch": 0.8718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.12,
      "step": 25468
    },
    {
      "epoch": 0.8719274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1154,
      "step": 25469
    },
    {
      "epoch": 0.8719616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1282,
      "step": 25470
    },
    {
      "epoch": 0.8719958918178706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1132,
      "step": 25471
    },
    {
      "epoch": 0.872030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1168,
      "step": 25472
    },
    {
      "epoch": 0.8720643615200274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1289,
      "step": 25473
    },
    {
      "epoch": 0.8720985963711058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1086,
      "step": 25474
    },
    {
      "epoch": 0.8721328312221842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1284,
      "step": 25475
    },
    {
      "epoch": 0.8721670660732626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1231,
      "step": 25476
    },
    {
      "epoch": 0.872201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1096,
      "step": 25477
    },
    {
      "epoch": 0.8722355357754193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1093,
      "step": 25478
    },
    {
      "epoch": 0.8722697706264978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1104,
      "step": 25479
    },
    {
      "epoch": 0.8723040054775761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1165,
      "step": 25480
    },
    {
      "epoch": 0.8723382403286546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1245,
      "step": 25481
    },
    {
      "epoch": 0.8723724751797329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1267,
      "step": 25482
    },
    {
      "epoch": 0.8724067100308114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1226,
      "step": 25483
    },
    {
      "epoch": 0.8724409448818897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.109,
      "step": 25484
    },
    {
      "epoch": 0.8724751797329682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1195,
      "step": 25485
    },
    {
      "epoch": 0.8725094145840465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1122,
      "step": 25486
    },
    {
      "epoch": 0.872543649435125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1162,
      "step": 25487
    },
    {
      "epoch": 0.8725778842862033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1349,
      "step": 25488
    },
    {
      "epoch": 0.8726121191372818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 25489
    },
    {
      "epoch": 0.8726463539883601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1303,
      "step": 25490
    },
    {
      "epoch": 0.8726805888394386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.109,
      "step": 25491
    },
    {
      "epoch": 0.8727148236905169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.128,
      "step": 25492
    },
    {
      "epoch": 0.8727490585415953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1084,
      "step": 25493
    },
    {
      "epoch": 0.8727832933926737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1135,
      "step": 25494
    },
    {
      "epoch": 0.8728175282437521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1306,
      "step": 25495
    },
    {
      "epoch": 0.8728517630948306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1349,
      "step": 25496
    },
    {
      "epoch": 0.8728859979459089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1314,
      "step": 25497
    },
    {
      "epoch": 0.8729202327969874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.106,
      "step": 25498
    },
    {
      "epoch": 0.8729544676480657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1158,
      "step": 25499
    },
    {
      "epoch": 0.8729887024991442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1148,
      "step": 25500
    },
    {
      "epoch": 0.8730229373502225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 25501
    },
    {
      "epoch": 0.873057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1142,
      "step": 25502
    },
    {
      "epoch": 0.8730914070523793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1432,
      "step": 25503
    },
    {
      "epoch": 0.8731256419034578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1202,
      "step": 25504
    },
    {
      "epoch": 0.8731598767545361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1137,
      "step": 25505
    },
    {
      "epoch": 0.8731941116056146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.11,
      "step": 25506
    },
    {
      "epoch": 0.8732283464566929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1172,
      "step": 25507
    },
    {
      "epoch": 0.8732625813077713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1277,
      "step": 25508
    },
    {
      "epoch": 0.8732968161588497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.117,
      "step": 25509
    },
    {
      "epoch": 0.8733310510099281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1066,
      "step": 25510
    },
    {
      "epoch": 0.8733652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1145,
      "step": 25511
    },
    {
      "epoch": 0.8733995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1146,
      "step": 25512
    },
    {
      "epoch": 0.8734337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1369,
      "step": 25513
    },
    {
      "epoch": 0.8734679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1262,
      "step": 25514
    },
    {
      "epoch": 0.8735022252653201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1148,
      "step": 25515
    },
    {
      "epoch": 0.8735364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1142,
      "step": 25516
    },
    {
      "epoch": 0.8735706949674769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1,
      "step": 25517
    },
    {
      "epoch": 0.8736049298185553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1264,
      "step": 25518
    },
    {
      "epoch": 0.8736391646696336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1242,
      "step": 25519
    },
    {
      "epoch": 0.8736733995207121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.117,
      "step": 25520
    },
    {
      "epoch": 0.8737076343717904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1129,
      "step": 25521
    },
    {
      "epoch": 0.8737418692228689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1204,
      "step": 25522
    },
    {
      "epoch": 0.8737761040739472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1208,
      "step": 25523
    },
    {
      "epoch": 0.8738103389250257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1065,
      "step": 25524
    },
    {
      "epoch": 0.873844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1173,
      "step": 25525
    },
    {
      "epoch": 0.8738788086271825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.127,
      "step": 25526
    },
    {
      "epoch": 0.8739130434782608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1112,
      "step": 25527
    },
    {
      "epoch": 0.8739472783293393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1094,
      "step": 25528
    },
    {
      "epoch": 0.8739815131804176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1208,
      "step": 25529
    },
    {
      "epoch": 0.8740157480314961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.121,
      "step": 25530
    },
    {
      "epoch": 0.8740499828825744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1174,
      "step": 25531
    },
    {
      "epoch": 0.8740842177336529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1012,
      "step": 25532
    },
    {
      "epoch": 0.8741184525847313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1108,
      "step": 25533
    },
    {
      "epoch": 0.8741526874358096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1086,
      "step": 25534
    },
    {
      "epoch": 0.8741869222868881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1351,
      "step": 25535
    },
    {
      "epoch": 0.8742211571379664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.121,
      "step": 25536
    },
    {
      "epoch": 0.8742553919890449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1331,
      "step": 25537
    },
    {
      "epoch": 0.8742896268401232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1241,
      "step": 25538
    },
    {
      "epoch": 0.8743238616912017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.112,
      "step": 25539
    },
    {
      "epoch": 0.87435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1137,
      "step": 25540
    },
    {
      "epoch": 0.8743923313933585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1224,
      "step": 25541
    },
    {
      "epoch": 0.8744265662444368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1183,
      "step": 25542
    },
    {
      "epoch": 0.8744608010955153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1002,
      "step": 25543
    },
    {
      "epoch": 0.8744950359465936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1317,
      "step": 25544
    },
    {
      "epoch": 0.8745292707976721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1185,
      "step": 25545
    },
    {
      "epoch": 0.8745635056487504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1315,
      "step": 25546
    },
    {
      "epoch": 0.8745977404998289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1191,
      "step": 25547
    },
    {
      "epoch": 0.8746319753509072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.112,
      "step": 25548
    },
    {
      "epoch": 0.8746662102019856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1184,
      "step": 25549
    },
    {
      "epoch": 0.874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1193,
      "step": 25550
    },
    {
      "epoch": 0.8747346799041424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.114,
      "step": 25551
    },
    {
      "epoch": 0.8747689147552208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1177,
      "step": 25552
    },
    {
      "epoch": 0.8748031496062992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1015,
      "step": 25553
    },
    {
      "epoch": 0.8748373844573776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1291,
      "step": 25554
    },
    {
      "epoch": 0.874871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1194,
      "step": 25555
    },
    {
      "epoch": 0.8749058541595344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1054,
      "step": 25556
    },
    {
      "epoch": 0.8749400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1177,
      "step": 25557
    },
    {
      "epoch": 0.8749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1089,
      "step": 25558
    },
    {
      "epoch": 0.8750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1222,
      "step": 25559
    },
    {
      "epoch": 0.875042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1138,
      "step": 25560
    },
    {
      "epoch": 0.8750770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1062,
      "step": 25561
    },
    {
      "epoch": 0.8751112632660047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1227,
      "step": 25562
    },
    {
      "epoch": 0.8751454981170832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1102,
      "step": 25563
    },
    {
      "epoch": 0.8751797329681615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1105,
      "step": 25564
    },
    {
      "epoch": 0.87521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 25565
    },
    {
      "epoch": 0.8752482026703183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1063,
      "step": 25566
    },
    {
      "epoch": 0.8752824375213968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1197,
      "step": 25567
    },
    {
      "epoch": 0.8753166723724752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.116,
      "step": 25568
    },
    {
      "epoch": 0.8753509072235536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0983,
      "step": 25569
    },
    {
      "epoch": 0.875385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0954,
      "step": 25570
    },
    {
      "epoch": 0.8754193769257104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1327,
      "step": 25571
    },
    {
      "epoch": 0.8754536117767888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1051,
      "step": 25572
    },
    {
      "epoch": 0.8754878466278672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1009,
      "step": 25573
    },
    {
      "epoch": 0.8755220814789456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1329,
      "step": 25574
    },
    {
      "epoch": 0.875556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1017,
      "step": 25575
    },
    {
      "epoch": 0.8755905511811024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.128,
      "step": 25576
    },
    {
      "epoch": 0.8756247860321807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1151,
      "step": 25577
    },
    {
      "epoch": 0.8756590208832592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1252,
      "step": 25578
    },
    {
      "epoch": 0.8756932557343375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1229,
      "step": 25579
    },
    {
      "epoch": 0.875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1357,
      "step": 25580
    },
    {
      "epoch": 0.8757617254364943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1248,
      "step": 25581
    },
    {
      "epoch": 0.8757959602875728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1126,
      "step": 25582
    },
    {
      "epoch": 0.8758301951386511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1179,
      "step": 25583
    },
    {
      "epoch": 0.8758644299897296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1183,
      "step": 25584
    },
    {
      "epoch": 0.8758986648408079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1,
      "step": 25585
    },
    {
      "epoch": 0.8759328996918864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1081,
      "step": 25586
    },
    {
      "epoch": 0.8759671345429647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1242,
      "step": 25587
    },
    {
      "epoch": 0.8760013693940432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1205,
      "step": 25588
    },
    {
      "epoch": 0.8760356042451215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1242,
      "step": 25589
    },
    {
      "epoch": 0.8760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1203,
      "step": 25590
    },
    {
      "epoch": 0.8761040739472783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1184,
      "step": 25591
    },
    {
      "epoch": 0.8761383087983567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1096,
      "step": 25592
    },
    {
      "epoch": 0.8761725436494351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1164,
      "step": 25593
    },
    {
      "epoch": 0.8762067785005135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 25594
    },
    {
      "epoch": 0.8762410133515919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1237,
      "step": 25595
    },
    {
      "epoch": 0.8762752482026703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1191,
      "step": 25596
    },
    {
      "epoch": 0.8763094830537487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1229,
      "step": 25597
    },
    {
      "epoch": 0.8763437179048271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1099,
      "step": 25598
    },
    {
      "epoch": 0.8763779527559055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 25599
    },
    {
      "epoch": 0.8764121876069839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1194,
      "step": 25600
    },
    {
      "epoch": 0.8764464224580623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.116,
      "step": 25601
    },
    {
      "epoch": 0.8764806573091407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1304,
      "step": 25602
    },
    {
      "epoch": 0.876514892160219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1144,
      "step": 25603
    },
    {
      "epoch": 0.8765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.129,
      "step": 25604
    },
    {
      "epoch": 0.876583361862376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1265,
      "step": 25605
    },
    {
      "epoch": 0.8766175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1132,
      "step": 25606
    },
    {
      "epoch": 0.8766518315645327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0983,
      "step": 25607
    },
    {
      "epoch": 0.8766860664156111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1223,
      "step": 25608
    },
    {
      "epoch": 0.8767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.116,
      "step": 25609
    },
    {
      "epoch": 0.8767545361177679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1229,
      "step": 25610
    },
    {
      "epoch": 0.8767887709688463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1137,
      "step": 25611
    },
    {
      "epoch": 0.8768230058199247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1223,
      "step": 25612
    },
    {
      "epoch": 0.8768572406710031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1029,
      "step": 25613
    },
    {
      "epoch": 0.8768914755220815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1234,
      "step": 25614
    },
    {
      "epoch": 0.8769257103731599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1165,
      "step": 25615
    },
    {
      "epoch": 0.8769599452242383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1094,
      "step": 25616
    },
    {
      "epoch": 0.8769941800753167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1227,
      "step": 25617
    },
    {
      "epoch": 0.877028414926395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1224,
      "step": 25618
    },
    {
      "epoch": 0.8770626497774735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1191,
      "step": 25619
    },
    {
      "epoch": 0.8770968846285518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1106,
      "step": 25620
    },
    {
      "epoch": 0.8771311194796303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1157,
      "step": 25621
    },
    {
      "epoch": 0.8771653543307086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1466,
      "step": 25622
    },
    {
      "epoch": 0.8771995891817871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1161,
      "step": 25623
    },
    {
      "epoch": 0.8772338240328654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1193,
      "step": 25624
    },
    {
      "epoch": 0.8772680588839439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 25625
    },
    {
      "epoch": 0.8773022937350222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1095,
      "step": 25626
    },
    {
      "epoch": 0.8773365285861007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1118,
      "step": 25627
    },
    {
      "epoch": 0.877370763437179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 25628
    },
    {
      "epoch": 0.8774049982882575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1128,
      "step": 25629
    },
    {
      "epoch": 0.8774392331393358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1147,
      "step": 25630
    },
    {
      "epoch": 0.8774734679904143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1171,
      "step": 25631
    },
    {
      "epoch": 0.8775077028414926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.121,
      "step": 25632
    },
    {
      "epoch": 0.877541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1084,
      "step": 25633
    },
    {
      "epoch": 0.8775761725436494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1128,
      "step": 25634
    },
    {
      "epoch": 0.8776104073947278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1145,
      "step": 25635
    },
    {
      "epoch": 0.8776446422458062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1286,
      "step": 25636
    },
    {
      "epoch": 0.8776788770968846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1279,
      "step": 25637
    },
    {
      "epoch": 0.877713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1177,
      "step": 25638
    },
    {
      "epoch": 0.8777473467990414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1036,
      "step": 25639
    },
    {
      "epoch": 0.8777815816501198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1232,
      "step": 25640
    },
    {
      "epoch": 0.8778158165011982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.116,
      "step": 25641
    },
    {
      "epoch": 0.8778500513522767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1155,
      "step": 25642
    },
    {
      "epoch": 0.877884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.111,
      "step": 25643
    },
    {
      "epoch": 0.8779185210544335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1044,
      "step": 25644
    },
    {
      "epoch": 0.8779527559055118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1039,
      "step": 25645
    },
    {
      "epoch": 0.8779869907565903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1004,
      "step": 25646
    },
    {
      "epoch": 0.8780212256076686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1134,
      "step": 25647
    },
    {
      "epoch": 0.878055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1272,
      "step": 25648
    },
    {
      "epoch": 0.8780896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1202,
      "step": 25649
    },
    {
      "epoch": 0.8781239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1134,
      "step": 25650
    },
    {
      "epoch": 0.8781581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0992,
      "step": 25651
    },
    {
      "epoch": 0.8781923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1132,
      "step": 25652
    },
    {
      "epoch": 0.878226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1331,
      "step": 25653
    },
    {
      "epoch": 0.8782608695652174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1114,
      "step": 25654
    },
    {
      "epoch": 0.8782951044162958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 25655
    },
    {
      "epoch": 0.8783293392673742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0944,
      "step": 25656
    },
    {
      "epoch": 0.8783635741184526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.106,
      "step": 25657
    },
    {
      "epoch": 0.878397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1131,
      "step": 25658
    },
    {
      "epoch": 0.8784320438206094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1359,
      "step": 25659
    },
    {
      "epoch": 0.8784662786716878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.121,
      "step": 25660
    },
    {
      "epoch": 0.8785005135227661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1128,
      "step": 25661
    },
    {
      "epoch": 0.8785347483738446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1191,
      "step": 25662
    },
    {
      "epoch": 0.8785689832249229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1168,
      "step": 25663
    },
    {
      "epoch": 0.8786032180760014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1201,
      "step": 25664
    },
    {
      "epoch": 0.8786374529270797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1141,
      "step": 25665
    },
    {
      "epoch": 0.8786716877781582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1257,
      "step": 25666
    },
    {
      "epoch": 0.8787059226292365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1139,
      "step": 25667
    },
    {
      "epoch": 0.878740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.11,
      "step": 25668
    },
    {
      "epoch": 0.8787743923313933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1164,
      "step": 25669
    },
    {
      "epoch": 0.8788086271824718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1245,
      "step": 25670
    },
    {
      "epoch": 0.8788428620335501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1259,
      "step": 25671
    },
    {
      "epoch": 0.8788770968846286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1211,
      "step": 25672
    },
    {
      "epoch": 0.8789113317357069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1144,
      "step": 25673
    },
    {
      "epoch": 0.8789455665867854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.114,
      "step": 25674
    },
    {
      "epoch": 0.8789798014378637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1268,
      "step": 25675
    },
    {
      "epoch": 0.8790140362889421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1191,
      "step": 25676
    },
    {
      "epoch": 0.8790482711400205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1176,
      "step": 25677
    },
    {
      "epoch": 0.8790825059910989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1133,
      "step": 25678
    },
    {
      "epoch": 0.8791167408421774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1126,
      "step": 25679
    },
    {
      "epoch": 0.8791509756932557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1098,
      "step": 25680
    },
    {
      "epoch": 0.8791852105443342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1158,
      "step": 25681
    },
    {
      "epoch": 0.8792194453954125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1121,
      "step": 25682
    },
    {
      "epoch": 0.879253680246491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1037,
      "step": 25683
    },
    {
      "epoch": 0.8792879150975693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1078,
      "step": 25684
    },
    {
      "epoch": 0.8793221499486478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1335,
      "step": 25685
    },
    {
      "epoch": 0.8793563847997261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1113,
      "step": 25686
    },
    {
      "epoch": 0.8793906196508046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.113,
      "step": 25687
    },
    {
      "epoch": 0.8794248545018829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 25688
    },
    {
      "epoch": 0.8794590893529614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1322,
      "step": 25689
    },
    {
      "epoch": 0.8794933242040397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1195,
      "step": 25690
    },
    {
      "epoch": 0.8795275590551181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1122,
      "step": 25691
    },
    {
      "epoch": 0.8795617939061965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.11,
      "step": 25692
    },
    {
      "epoch": 0.8795960287572749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1346,
      "step": 25693
    },
    {
      "epoch": 0.8796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1214,
      "step": 25694
    },
    {
      "epoch": 0.8796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1246,
      "step": 25695
    },
    {
      "epoch": 0.8796987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.121,
      "step": 25696
    },
    {
      "epoch": 0.8797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.111,
      "step": 25697
    },
    {
      "epoch": 0.8797672030126669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1066,
      "step": 25698
    },
    {
      "epoch": 0.8798014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1133,
      "step": 25699
    },
    {
      "epoch": 0.8798356727148237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1107,
      "step": 25700
    },
    {
      "epoch": 0.8798699075659021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1057,
      "step": 25701
    },
    {
      "epoch": 0.8799041424169805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1173,
      "step": 25702
    },
    {
      "epoch": 0.8799383772680589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1222,
      "step": 25703
    },
    {
      "epoch": 0.8799726121191372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1075,
      "step": 25704
    },
    {
      "epoch": 0.8800068469702157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1357,
      "step": 25705
    },
    {
      "epoch": 0.880041081821294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1229,
      "step": 25706
    },
    {
      "epoch": 0.8800753166723725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1169,
      "step": 25707
    },
    {
      "epoch": 0.8801095515234508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1189,
      "step": 25708
    },
    {
      "epoch": 0.8801437863745293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1213,
      "step": 25709
    },
    {
      "epoch": 0.8801780212256076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1083,
      "step": 25710
    },
    {
      "epoch": 0.8802122560766861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1227,
      "step": 25711
    },
    {
      "epoch": 0.8802464909277644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1039,
      "step": 25712
    },
    {
      "epoch": 0.8802807257788429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1195,
      "step": 25713
    },
    {
      "epoch": 0.8803149606299212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 25714
    },
    {
      "epoch": 0.8803491954809997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 25715
    },
    {
      "epoch": 0.8803834303320781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.118,
      "step": 25716
    },
    {
      "epoch": 0.8804176651831565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1051,
      "step": 25717
    },
    {
      "epoch": 0.8804519000342349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1094,
      "step": 25718
    },
    {
      "epoch": 0.8804861348853132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.126,
      "step": 25719
    },
    {
      "epoch": 0.8805203697363917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1216,
      "step": 25720
    },
    {
      "epoch": 0.88055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1165,
      "step": 25721
    },
    {
      "epoch": 0.8805888394385485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1207,
      "step": 25722
    },
    {
      "epoch": 0.8806230742896268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1157,
      "step": 25723
    },
    {
      "epoch": 0.8806573091407053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1151,
      "step": 25724
    },
    {
      "epoch": 0.8806915439917836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1271,
      "step": 25725
    },
    {
      "epoch": 0.8807257788428621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1121,
      "step": 25726
    },
    {
      "epoch": 0.8807600136939404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1081,
      "step": 25727
    },
    {
      "epoch": 0.8807942485450189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1129,
      "step": 25728
    },
    {
      "epoch": 0.8808284833960972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1297,
      "step": 25729
    },
    {
      "epoch": 0.8808627182471757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1123,
      "step": 25730
    },
    {
      "epoch": 0.880896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0965,
      "step": 25731
    },
    {
      "epoch": 0.8809311879493325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1093,
      "step": 25732
    },
    {
      "epoch": 0.8809654228004108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1328,
      "step": 25733
    },
    {
      "epoch": 0.8809996576514892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.125,
      "step": 25734
    },
    {
      "epoch": 0.8810338925025676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1122,
      "step": 25735
    },
    {
      "epoch": 0.881068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1035,
      "step": 25736
    },
    {
      "epoch": 0.8811023622047244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1149,
      "step": 25737
    },
    {
      "epoch": 0.8811365970558028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1068,
      "step": 25738
    },
    {
      "epoch": 0.8811708319068812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1123,
      "step": 25739
    },
    {
      "epoch": 0.8812050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0991,
      "step": 25740
    },
    {
      "epoch": 0.881239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1391,
      "step": 25741
    },
    {
      "epoch": 0.8812735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1108,
      "step": 25742
    },
    {
      "epoch": 0.8813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1171,
      "step": 25743
    },
    {
      "epoch": 0.8813420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1179,
      "step": 25744
    },
    {
      "epoch": 0.8813762410133515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1068,
      "step": 25745
    },
    {
      "epoch": 0.88141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1264,
      "step": 25746
    },
    {
      "epoch": 0.8814447107155083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1205,
      "step": 25747
    },
    {
      "epoch": 0.8814789455665868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1082,
      "step": 25748
    },
    {
      "epoch": 0.8815131804176651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1185,
      "step": 25749
    },
    {
      "epoch": 0.8815474152687436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1089,
      "step": 25750
    },
    {
      "epoch": 0.8815816501198219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1154,
      "step": 25751
    },
    {
      "epoch": 0.8816158849709004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1144,
      "step": 25752
    },
    {
      "epoch": 0.8816501198219788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1151,
      "step": 25753
    },
    {
      "epoch": 0.8816843546730572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1164,
      "step": 25754
    },
    {
      "epoch": 0.8817185895241356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1141,
      "step": 25755
    },
    {
      "epoch": 0.881752824375214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1037,
      "step": 25756
    },
    {
      "epoch": 0.8817870592262924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1024,
      "step": 25757
    },
    {
      "epoch": 0.8818212940773708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1338,
      "step": 25758
    },
    {
      "epoch": 0.8818555289284492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1298,
      "step": 25759
    },
    {
      "epoch": 0.8818897637795275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1145,
      "step": 25760
    },
    {
      "epoch": 0.881923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1061,
      "step": 25761
    },
    {
      "epoch": 0.8819582334816843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1155,
      "step": 25762
    },
    {
      "epoch": 0.8819924683327628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1228,
      "step": 25763
    },
    {
      "epoch": 0.8820267031838411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1259,
      "step": 25764
    },
    {
      "epoch": 0.8820609380349196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1179,
      "step": 25765
    },
    {
      "epoch": 0.8820951728859979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1019,
      "step": 25766
    },
    {
      "epoch": 0.8821294077370764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1144,
      "step": 25767
    },
    {
      "epoch": 0.8821636425881547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1237,
      "step": 25768
    },
    {
      "epoch": 0.8821978774392332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1027,
      "step": 25769
    },
    {
      "epoch": 0.8822321122903115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.13,
      "step": 25770
    },
    {
      "epoch": 0.88226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1293,
      "step": 25771
    },
    {
      "epoch": 0.8823005819924683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1135,
      "step": 25772
    },
    {
      "epoch": 0.8823348168435468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1195,
      "step": 25773
    },
    {
      "epoch": 0.8823690516946251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.102,
      "step": 25774
    },
    {
      "epoch": 0.8824032865457035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1107,
      "step": 25775
    },
    {
      "epoch": 0.8824375213967819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.118,
      "step": 25776
    },
    {
      "epoch": 0.8824717562478603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.108,
      "step": 25777
    },
    {
      "epoch": 0.8825059910989387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1269,
      "step": 25778
    },
    {
      "epoch": 0.8825402259500171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1033,
      "step": 25779
    },
    {
      "epoch": 0.8825744608010955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1195,
      "step": 25780
    },
    {
      "epoch": 0.8826086956521739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1139,
      "step": 25781
    },
    {
      "epoch": 0.8826429305032523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1036,
      "step": 25782
    },
    {
      "epoch": 0.8826771653543307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1244,
      "step": 25783
    },
    {
      "epoch": 0.8827114002054091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1051,
      "step": 25784
    },
    {
      "epoch": 0.8827456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1108,
      "step": 25785
    },
    {
      "epoch": 0.8827798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0957,
      "step": 25786
    },
    {
      "epoch": 0.8828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1289,
      "step": 25787
    },
    {
      "epoch": 0.8828483396097226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1313,
      "step": 25788
    },
    {
      "epoch": 0.8828825744608011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1029,
      "step": 25789
    },
    {
      "epoch": 0.8829168093118795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1261,
      "step": 25790
    },
    {
      "epoch": 0.8829510441629579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1025,
      "step": 25791
    },
    {
      "epoch": 0.8829852790140363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0939,
      "step": 25792
    },
    {
      "epoch": 0.8830195138651147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1122,
      "step": 25793
    },
    {
      "epoch": 0.8830537487161931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1163,
      "step": 25794
    },
    {
      "epoch": 0.8830879835672715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1096,
      "step": 25795
    },
    {
      "epoch": 0.8831222184183499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1136,
      "step": 25796
    },
    {
      "epoch": 0.8831564532694283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1179,
      "step": 25797
    },
    {
      "epoch": 0.8831906881205067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0962,
      "step": 25798
    },
    {
      "epoch": 0.8832249229715851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1181,
      "step": 25799
    },
    {
      "epoch": 0.8832591578226635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1098,
      "step": 25800
    },
    {
      "epoch": 0.8832933926737419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1139,
      "step": 25801
    },
    {
      "epoch": 0.8833276275248203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1259,
      "step": 25802
    },
    {
      "epoch": 0.8833618623758986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1257,
      "step": 25803
    },
    {
      "epoch": 0.8833960972269771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1042,
      "step": 25804
    },
    {
      "epoch": 0.8834303320780554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1154,
      "step": 25805
    },
    {
      "epoch": 0.8834645669291339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1233,
      "step": 25806
    },
    {
      "epoch": 0.8834988017802122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1098,
      "step": 25807
    },
    {
      "epoch": 0.8835330366312907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1139,
      "step": 25808
    },
    {
      "epoch": 0.883567271482369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 25809
    },
    {
      "epoch": 0.8836015063334475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1015,
      "step": 25810
    },
    {
      "epoch": 0.8836357411845258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1082,
      "step": 25811
    },
    {
      "epoch": 0.8836699760356043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1352,
      "step": 25812
    },
    {
      "epoch": 0.8837042108866826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1105,
      "step": 25813
    },
    {
      "epoch": 0.8837384457377611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1012,
      "step": 25814
    },
    {
      "epoch": 0.8837726805888394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1183,
      "step": 25815
    },
    {
      "epoch": 0.8838069154399179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1176,
      "step": 25816
    },
    {
      "epoch": 0.8838411502909962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1037,
      "step": 25817
    },
    {
      "epoch": 0.8838753851420746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1229,
      "step": 25818
    },
    {
      "epoch": 0.883909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1138,
      "step": 25819
    },
    {
      "epoch": 0.8839438548442314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1158,
      "step": 25820
    },
    {
      "epoch": 0.8839780896953098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1397,
      "step": 25821
    },
    {
      "epoch": 0.8840123245463882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1107,
      "step": 25822
    },
    {
      "epoch": 0.8840465593974666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1262,
      "step": 25823
    },
    {
      "epoch": 0.884080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1143,
      "step": 25824
    },
    {
      "epoch": 0.8841150290996234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.127,
      "step": 25825
    },
    {
      "epoch": 0.8841492639507018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1046,
      "step": 25826
    },
    {
      "epoch": 0.8841834988017803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1374,
      "step": 25827
    },
    {
      "epoch": 0.8842177336528586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1077,
      "step": 25828
    },
    {
      "epoch": 0.8842519685039371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1202,
      "step": 25829
    },
    {
      "epoch": 0.8842862033550154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1215,
      "step": 25830
    },
    {
      "epoch": 0.8843204382060939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1171,
      "step": 25831
    },
    {
      "epoch": 0.8843546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1153,
      "step": 25832
    },
    {
      "epoch": 0.8843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1009,
      "step": 25833
    },
    {
      "epoch": 0.884423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1167,
      "step": 25834
    },
    {
      "epoch": 0.8844573776104074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1143,
      "step": 25835
    },
    {
      "epoch": 0.8844916124614858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1115,
      "step": 25836
    },
    {
      "epoch": 0.8845258473125642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1046,
      "step": 25837
    },
    {
      "epoch": 0.8845600821636426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.109,
      "step": 25838
    },
    {
      "epoch": 0.884594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1372,
      "step": 25839
    },
    {
      "epoch": 0.8846285518657994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1223,
      "step": 25840
    },
    {
      "epoch": 0.8846627867168778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1137,
      "step": 25841
    },
    {
      "epoch": 0.8846970215679562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1079,
      "step": 25842
    },
    {
      "epoch": 0.8847312564190346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 25843
    },
    {
      "epoch": 0.884765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.122,
      "step": 25844
    },
    {
      "epoch": 0.8847997261211914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1226,
      "step": 25845
    },
    {
      "epoch": 0.8848339609722697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1109,
      "step": 25846
    },
    {
      "epoch": 0.8848681958233482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1145,
      "step": 25847
    },
    {
      "epoch": 0.8849024306744265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 25848
    },
    {
      "epoch": 0.884936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1089,
      "step": 25849
    },
    {
      "epoch": 0.8849709003765833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1025,
      "step": 25850
    },
    {
      "epoch": 0.8850051352276618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0932,
      "step": 25851
    },
    {
      "epoch": 0.8850393700787401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0968,
      "step": 25852
    },
    {
      "epoch": 0.8850736049298186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1066,
      "step": 25853
    },
    {
      "epoch": 0.8851078397808969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0933,
      "step": 25854
    },
    {
      "epoch": 0.8851420746319754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1121,
      "step": 25855
    },
    {
      "epoch": 0.8851763094830537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1115,
      "step": 25856
    },
    {
      "epoch": 0.8852105443341322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1167,
      "step": 25857
    },
    {
      "epoch": 0.8852447791852105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1103,
      "step": 25858
    },
    {
      "epoch": 0.885279014036289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1014,
      "step": 25859
    },
    {
      "epoch": 0.8853132488873673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1148,
      "step": 25860
    },
    {
      "epoch": 0.8853474837384457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1091,
      "step": 25861
    },
    {
      "epoch": 0.8853817185895241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1115,
      "step": 25862
    },
    {
      "epoch": 0.8854159534406025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0935,
      "step": 25863
    },
    {
      "epoch": 0.885450188291681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1051,
      "step": 25864
    },
    {
      "epoch": 0.8854844231427593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1154,
      "step": 25865
    },
    {
      "epoch": 0.8855186579938378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.12,
      "step": 25866
    },
    {
      "epoch": 0.8855528928449161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1366,
      "step": 25867
    },
    {
      "epoch": 0.8855871276959946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1144,
      "step": 25868
    },
    {
      "epoch": 0.8856213625470729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1148,
      "step": 25869
    },
    {
      "epoch": 0.8856555973981514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1152,
      "step": 25870
    },
    {
      "epoch": 0.8856898322492297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1176,
      "step": 25871
    },
    {
      "epoch": 0.8857240671003082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1237,
      "step": 25872
    },
    {
      "epoch": 0.8857583019513865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1191,
      "step": 25873
    },
    {
      "epoch": 0.885792536802465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.124,
      "step": 25874
    },
    {
      "epoch": 0.8858267716535433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1068,
      "step": 25875
    },
    {
      "epoch": 0.8858610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1257,
      "step": 25876
    },
    {
      "epoch": 0.8858952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1189,
      "step": 25877
    },
    {
      "epoch": 0.8859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1186,
      "step": 25878
    },
    {
      "epoch": 0.8859637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1091,
      "step": 25879
    },
    {
      "epoch": 0.8859979459089353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1222,
      "step": 25880
    },
    {
      "epoch": 0.8860321807600137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1237,
      "step": 25881
    },
    {
      "epoch": 0.8860664156110921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1003,
      "step": 25882
    },
    {
      "epoch": 0.8861006504621705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1102,
      "step": 25883
    },
    {
      "epoch": 0.8861348853132489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1143,
      "step": 25884
    },
    {
      "epoch": 0.8861691201643273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.117,
      "step": 25885
    },
    {
      "epoch": 0.8862033550154057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1302,
      "step": 25886
    },
    {
      "epoch": 0.886237589866484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.12,
      "step": 25887
    },
    {
      "epoch": 0.8862718247175625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1272,
      "step": 25888
    },
    {
      "epoch": 0.8863060595686408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.101,
      "step": 25889
    },
    {
      "epoch": 0.8863402944197193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1177,
      "step": 25890
    },
    {
      "epoch": 0.8863745292707976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 25891
    },
    {
      "epoch": 0.8864087641218761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1162,
      "step": 25892
    },
    {
      "epoch": 0.8864429989729544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1154,
      "step": 25893
    },
    {
      "epoch": 0.8864772338240329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1187,
      "step": 25894
    },
    {
      "epoch": 0.8865114686751112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1127,
      "step": 25895
    },
    {
      "epoch": 0.8865457035261897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1239,
      "step": 25896
    },
    {
      "epoch": 0.886579938377268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1002,
      "step": 25897
    },
    {
      "epoch": 0.8866141732283465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.113,
      "step": 25898
    },
    {
      "epoch": 0.8866484080794248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1247,
      "step": 25899
    },
    {
      "epoch": 0.8866826429305033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1195,
      "step": 25900
    },
    {
      "epoch": 0.8867168777815817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1266,
      "step": 25901
    },
    {
      "epoch": 0.88675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1339,
      "step": 25902
    },
    {
      "epoch": 0.8867853474837385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1189,
      "step": 25903
    },
    {
      "epoch": 0.8868195823348168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1197,
      "step": 25904
    },
    {
      "epoch": 0.8868538171858953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 25905
    },
    {
      "epoch": 0.8868880520369736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.091,
      "step": 25906
    },
    {
      "epoch": 0.8869222868880521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1169,
      "step": 25907
    },
    {
      "epoch": 0.8869565217391304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0989,
      "step": 25908
    },
    {
      "epoch": 0.8869907565902089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.115,
      "step": 25909
    },
    {
      "epoch": 0.8870249914412872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1145,
      "step": 25910
    },
    {
      "epoch": 0.8870592262923657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1251,
      "step": 25911
    },
    {
      "epoch": 0.887093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1183,
      "step": 25912
    },
    {
      "epoch": 0.8871276959945225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1123,
      "step": 25913
    },
    {
      "epoch": 0.8871619308456008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1191,
      "step": 25914
    },
    {
      "epoch": 0.8871961656966793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0997,
      "step": 25915
    },
    {
      "epoch": 0.8872304005477576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1149,
      "step": 25916
    },
    {
      "epoch": 0.887264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1208,
      "step": 25917
    },
    {
      "epoch": 0.8872988702499144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1069,
      "step": 25918
    },
    {
      "epoch": 0.8873331051009928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.102,
      "step": 25919
    },
    {
      "epoch": 0.8873673399520712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.111,
      "step": 25920
    },
    {
      "epoch": 0.8874015748031496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 25921
    },
    {
      "epoch": 0.887435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1153,
      "step": 25922
    },
    {
      "epoch": 0.8874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1203,
      "step": 25923
    },
    {
      "epoch": 0.8875042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1204,
      "step": 25924
    },
    {
      "epoch": 0.8875385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0988,
      "step": 25925
    },
    {
      "epoch": 0.8875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1169,
      "step": 25926
    },
    {
      "epoch": 0.88760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1077,
      "step": 25927
    },
    {
      "epoch": 0.8876412187606983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1152,
      "step": 25928
    },
    {
      "epoch": 0.8876754536117768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1159,
      "step": 25929
    },
    {
      "epoch": 0.8877096884628551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1124,
      "step": 25930
    },
    {
      "epoch": 0.8877439233139336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1065,
      "step": 25931
    },
    {
      "epoch": 0.8877781581650119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1166,
      "step": 25932
    },
    {
      "epoch": 0.8878123930160904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1235,
      "step": 25933
    },
    {
      "epoch": 0.8878466278671687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.112,
      "step": 25934
    },
    {
      "epoch": 0.8878808627182472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1313,
      "step": 25935
    },
    {
      "epoch": 0.8879150975693256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1047,
      "step": 25936
    },
    {
      "epoch": 0.887949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1135,
      "step": 25937
    },
    {
      "epoch": 0.8879835672714824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1266,
      "step": 25938
    },
    {
      "epoch": 0.8880178021225608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0912,
      "step": 25939
    },
    {
      "epoch": 0.8880520369736392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1047,
      "step": 25940
    },
    {
      "epoch": 0.8880862718247176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1174,
      "step": 25941
    },
    {
      "epoch": 0.888120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1195,
      "step": 25942
    },
    {
      "epoch": 0.8881547415268743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1015,
      "step": 25943
    },
    {
      "epoch": 0.8881889763779528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1057,
      "step": 25944
    },
    {
      "epoch": 0.8882232112290311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1371,
      "step": 25945
    },
    {
      "epoch": 0.8882574460801096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1184,
      "step": 25946
    },
    {
      "epoch": 0.8882916809311879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1075,
      "step": 25947
    },
    {
      "epoch": 0.8883259157822664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1025,
      "step": 25948
    },
    {
      "epoch": 0.8883601506333447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1128,
      "step": 25949
    },
    {
      "epoch": 0.8883943854844232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1297,
      "step": 25950
    },
    {
      "epoch": 0.8884286203355015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1115,
      "step": 25951
    },
    {
      "epoch": 0.88846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 25952
    },
    {
      "epoch": 0.8884970900376583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1067,
      "step": 25953
    },
    {
      "epoch": 0.8885313248887368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1103,
      "step": 25954
    },
    {
      "epoch": 0.8885655597398151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1153,
      "step": 25955
    },
    {
      "epoch": 0.8885997945908936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1081,
      "step": 25956
    },
    {
      "epoch": 0.8886340294419719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1061,
      "step": 25957
    },
    {
      "epoch": 0.8886682642930503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1241,
      "step": 25958
    },
    {
      "epoch": 0.8887024991441287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1148,
      "step": 25959
    },
    {
      "epoch": 0.8887367339952071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1131,
      "step": 25960
    },
    {
      "epoch": 0.8887709688462855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1057,
      "step": 25961
    },
    {
      "epoch": 0.8888052036973639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1058,
      "step": 25962
    },
    {
      "epoch": 0.8888394385484423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1136,
      "step": 25963
    },
    {
      "epoch": 0.8888736733995207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1002,
      "step": 25964
    },
    {
      "epoch": 0.8889079082505991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.125,
      "step": 25965
    },
    {
      "epoch": 0.8889421431016775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1102,
      "step": 25966
    },
    {
      "epoch": 0.8889763779527559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1129,
      "step": 25967
    },
    {
      "epoch": 0.8890106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1171,
      "step": 25968
    },
    {
      "epoch": 0.8890448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1111,
      "step": 25969
    },
    {
      "epoch": 0.8890790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1087,
      "step": 25970
    },
    {
      "epoch": 0.8891133173570694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1239,
      "step": 25971
    },
    {
      "epoch": 0.8891475522081479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1164,
      "step": 25972
    },
    {
      "epoch": 0.8891817870592263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1144,
      "step": 25973
    },
    {
      "epoch": 0.8892160219103047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.097,
      "step": 25974
    },
    {
      "epoch": 0.8892502567613831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1124,
      "step": 25975
    },
    {
      "epoch": 0.8892844916124615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1115,
      "step": 25976
    },
    {
      "epoch": 0.8893187264635399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1251,
      "step": 25977
    },
    {
      "epoch": 0.8893529613146183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0974,
      "step": 25978
    },
    {
      "epoch": 0.8893871961656967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0957,
      "step": 25979
    },
    {
      "epoch": 0.8894214310167751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 25980
    },
    {
      "epoch": 0.8894556658678535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0962,
      "step": 25981
    },
    {
      "epoch": 0.8894899007189319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1177,
      "step": 25982
    },
    {
      "epoch": 0.8895241355700103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1004,
      "step": 25983
    },
    {
      "epoch": 0.8895583704210887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1207,
      "step": 25984
    },
    {
      "epoch": 0.8895926052721671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1185,
      "step": 25985
    },
    {
      "epoch": 0.8896268401232454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0952,
      "step": 25986
    },
    {
      "epoch": 0.8896610749743239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.122,
      "step": 25987
    },
    {
      "epoch": 0.8896953098254022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1044,
      "step": 25988
    },
    {
      "epoch": 0.8897295446764807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1083,
      "step": 25989
    },
    {
      "epoch": 0.889763779527559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1153,
      "step": 25990
    },
    {
      "epoch": 0.8897980143786375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1238,
      "step": 25991
    },
    {
      "epoch": 0.8898322492297158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.101,
      "step": 25992
    },
    {
      "epoch": 0.8898664840807943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 25993
    },
    {
      "epoch": 0.8899007189318726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1129,
      "step": 25994
    },
    {
      "epoch": 0.8899349537829511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1177,
      "step": 25995
    },
    {
      "epoch": 0.8899691886340294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1151,
      "step": 25996
    },
    {
      "epoch": 0.8900034234851079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1145,
      "step": 25997
    },
    {
      "epoch": 0.8900376583361862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.114,
      "step": 25998
    },
    {
      "epoch": 0.8900718931872647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 25999
    },
    {
      "epoch": 0.890106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1036,
      "step": 26000
    },
    {
      "epoch": 0.8901403628894214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1276,
      "step": 26001
    },
    {
      "epoch": 0.8901745977404998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1063,
      "step": 26002
    },
    {
      "epoch": 0.8902088325915782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1055,
      "step": 26003
    },
    {
      "epoch": 0.8902430674426566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1097,
      "step": 26004
    },
    {
      "epoch": 0.890277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1213,
      "step": 26005
    },
    {
      "epoch": 0.8903115371448134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1264,
      "step": 26006
    },
    {
      "epoch": 0.8903457719958918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1159,
      "step": 26007
    },
    {
      "epoch": 0.8903800068469702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0981,
      "step": 26008
    },
    {
      "epoch": 0.8904142416980486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1293,
      "step": 26009
    },
    {
      "epoch": 0.8904484765491271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1268,
      "step": 26010
    },
    {
      "epoch": 0.8904827114002054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1187,
      "step": 26011
    },
    {
      "epoch": 0.8905169462512839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1207,
      "step": 26012
    },
    {
      "epoch": 0.8905511811023622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1163,
      "step": 26013
    },
    {
      "epoch": 0.8905854159534407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1078,
      "step": 26014
    },
    {
      "epoch": 0.890619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1128,
      "step": 26015
    },
    {
      "epoch": 0.8906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1064,
      "step": 26016
    },
    {
      "epoch": 0.8906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1053,
      "step": 26017
    },
    {
      "epoch": 0.8907223553577542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1105,
      "step": 26018
    },
    {
      "epoch": 0.8907565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1042,
      "step": 26019
    },
    {
      "epoch": 0.890790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1146,
      "step": 26020
    },
    {
      "epoch": 0.8908250599109894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1148,
      "step": 26021
    },
    {
      "epoch": 0.8908592947620678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1217,
      "step": 26022
    },
    {
      "epoch": 0.8908935296131462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1039,
      "step": 26023
    },
    {
      "epoch": 0.8909277644642246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1156,
      "step": 26024
    },
    {
      "epoch": 0.890961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0988,
      "step": 26025
    },
    {
      "epoch": 0.8909962341663814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1216,
      "step": 26026
    },
    {
      "epoch": 0.8910304690174597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1203,
      "step": 26027
    },
    {
      "epoch": 0.8910647038685382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0891,
      "step": 26028
    },
    {
      "epoch": 0.8910989387196165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1233,
      "step": 26029
    },
    {
      "epoch": 0.891133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0948,
      "step": 26030
    },
    {
      "epoch": 0.8911674084217733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1185,
      "step": 26031
    },
    {
      "epoch": 0.8912016432728518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1132,
      "step": 26032
    },
    {
      "epoch": 0.8912358781239301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.111,
      "step": 26033
    },
    {
      "epoch": 0.8912701129750086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1045,
      "step": 26034
    },
    {
      "epoch": 0.8913043478260869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0918,
      "step": 26035
    },
    {
      "epoch": 0.8913385826771654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1181,
      "step": 26036
    },
    {
      "epoch": 0.8913728175282437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1087,
      "step": 26037
    },
    {
      "epoch": 0.8914070523793222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1161,
      "step": 26038
    },
    {
      "epoch": 0.8914412872304005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1193,
      "step": 26039
    },
    {
      "epoch": 0.891475522081479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1133,
      "step": 26040
    },
    {
      "epoch": 0.8915097569325573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1194,
      "step": 26041
    },
    {
      "epoch": 0.8915439917836357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 26042
    },
    {
      "epoch": 0.8915782266347141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1169,
      "step": 26043
    },
    {
      "epoch": 0.8916124614857925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1086,
      "step": 26044
    },
    {
      "epoch": 0.8916466963368709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1062,
      "step": 26045
    },
    {
      "epoch": 0.8916809311879493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1264,
      "step": 26046
    },
    {
      "epoch": 0.8917151660390278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1122,
      "step": 26047
    },
    {
      "epoch": 0.8917494008901061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0979,
      "step": 26048
    },
    {
      "epoch": 0.8917836357411846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.108,
      "step": 26049
    },
    {
      "epoch": 0.8918178705922629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1224,
      "step": 26050
    },
    {
      "epoch": 0.8918521054433414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 26051
    },
    {
      "epoch": 0.8918863402944197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1153,
      "step": 26052
    },
    {
      "epoch": 0.8919205751454982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0971,
      "step": 26053
    },
    {
      "epoch": 0.8919548099965765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1103,
      "step": 26054
    },
    {
      "epoch": 0.891989044847655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1051,
      "step": 26055
    },
    {
      "epoch": 0.8920232796987333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1084,
      "step": 26056
    },
    {
      "epoch": 0.8920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.112,
      "step": 26057
    },
    {
      "epoch": 0.8920917494008901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1025,
      "step": 26058
    },
    {
      "epoch": 0.8921259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1142,
      "step": 26059
    },
    {
      "epoch": 0.8921602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1066,
      "step": 26060
    },
    {
      "epoch": 0.8921944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1005,
      "step": 26061
    },
    {
      "epoch": 0.8922286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.13,
      "step": 26062
    },
    {
      "epoch": 0.8922629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0973,
      "step": 26063
    },
    {
      "epoch": 0.8922971585073605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1062,
      "step": 26064
    },
    {
      "epoch": 0.8923313933584389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0997,
      "step": 26065
    },
    {
      "epoch": 0.8923656282095173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1059,
      "step": 26066
    },
    {
      "epoch": 0.8923998630605957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 26067
    },
    {
      "epoch": 0.892434097911674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1031,
      "step": 26068
    },
    {
      "epoch": 0.8924683327627525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1145,
      "step": 26069
    },
    {
      "epoch": 0.8925025676138308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1027,
      "step": 26070
    },
    {
      "epoch": 0.8925368024649093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1103,
      "step": 26071
    },
    {
      "epoch": 0.8925710373159876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 26072
    },
    {
      "epoch": 0.8926052721670661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1184,
      "step": 26073
    },
    {
      "epoch": 0.8926395070181444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1006,
      "step": 26074
    },
    {
      "epoch": 0.8926737418692229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.088,
      "step": 26075
    },
    {
      "epoch": 0.8927079767203012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1157,
      "step": 26076
    },
    {
      "epoch": 0.8927422115713797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1001,
      "step": 26077
    },
    {
      "epoch": 0.892776446422458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1032,
      "step": 26078
    },
    {
      "epoch": 0.8928106812735365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.111,
      "step": 26079
    },
    {
      "epoch": 0.8928449161246148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1221,
      "step": 26080
    },
    {
      "epoch": 0.8928791509756933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 26081
    },
    {
      "epoch": 0.8929133858267716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1109,
      "step": 26082
    },
    {
      "epoch": 0.89294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1035,
      "step": 26083
    },
    {
      "epoch": 0.8929818555289285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 26084
    },
    {
      "epoch": 0.8930160903800068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1065,
      "step": 26085
    },
    {
      "epoch": 0.8930503252310853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1095,
      "step": 26086
    },
    {
      "epoch": 0.8930845600821636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1022,
      "step": 26087
    },
    {
      "epoch": 0.8931187949332421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1252,
      "step": 26088
    },
    {
      "epoch": 0.8931530297843204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0979,
      "step": 26089
    },
    {
      "epoch": 0.8931872646353989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1143,
      "step": 26090
    },
    {
      "epoch": 0.8932214994864772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1052,
      "step": 26091
    },
    {
      "epoch": 0.8932557343375557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1094,
      "step": 26092
    },
    {
      "epoch": 0.893289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1083,
      "step": 26093
    },
    {
      "epoch": 0.8933242040397125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0983,
      "step": 26094
    },
    {
      "epoch": 0.8933584388907908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1196,
      "step": 26095
    },
    {
      "epoch": 0.8933926737418693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1028,
      "step": 26096
    },
    {
      "epoch": 0.8934269085929476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1075,
      "step": 26097
    },
    {
      "epoch": 0.893461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0966,
      "step": 26098
    },
    {
      "epoch": 0.8934953782951044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1233,
      "step": 26099
    },
    {
      "epoch": 0.8935296131461828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1174,
      "step": 26100
    },
    {
      "epoch": 0.8935638479972612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1067,
      "step": 26101
    },
    {
      "epoch": 0.8935980828483396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1227,
      "step": 26102
    },
    {
      "epoch": 0.893632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 26103
    },
    {
      "epoch": 0.8936665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1302,
      "step": 26104
    },
    {
      "epoch": 0.8937007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0992,
      "step": 26105
    },
    {
      "epoch": 0.8937350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.107,
      "step": 26106
    },
    {
      "epoch": 0.8937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1218,
      "step": 26107
    },
    {
      "epoch": 0.89380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1071,
      "step": 26108
    },
    {
      "epoch": 0.8938377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1238,
      "step": 26109
    },
    {
      "epoch": 0.8938719616569668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1094,
      "step": 26110
    },
    {
      "epoch": 0.8939061965080451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.103,
      "step": 26111
    },
    {
      "epoch": 0.8939404313591236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1121,
      "step": 26112
    },
    {
      "epoch": 0.8939746662102019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0958,
      "step": 26113
    },
    {
      "epoch": 0.8940089010612804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1198,
      "step": 26114
    },
    {
      "epoch": 0.8940431359123587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1103,
      "step": 26115
    },
    {
      "epoch": 0.8940773707634372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0989,
      "step": 26116
    },
    {
      "epoch": 0.8941116056145155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0965,
      "step": 26117
    },
    {
      "epoch": 0.894145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0978,
      "step": 26118
    },
    {
      "epoch": 0.8941800753166723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.122,
      "step": 26119
    },
    {
      "epoch": 0.8942143101677508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1145,
      "step": 26120
    },
    {
      "epoch": 0.8942485450188292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0988,
      "step": 26121
    },
    {
      "epoch": 0.8942827798699076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1162,
      "step": 26122
    },
    {
      "epoch": 0.894317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1087,
      "step": 26123
    },
    {
      "epoch": 0.8943512495720644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0916,
      "step": 26124
    },
    {
      "epoch": 0.8943854844231428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1078,
      "step": 26125
    },
    {
      "epoch": 0.8944197192742211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.101,
      "step": 26126
    },
    {
      "epoch": 0.8944539541252996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1102,
      "step": 26127
    },
    {
      "epoch": 0.8944881889763779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.085,
      "step": 26128
    },
    {
      "epoch": 0.8945224238274564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0999,
      "step": 26129
    },
    {
      "epoch": 0.8945566586785347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1065,
      "step": 26130
    },
    {
      "epoch": 0.8945908935296132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1126,
      "step": 26131
    },
    {
      "epoch": 0.8946251283806915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 26132
    },
    {
      "epoch": 0.89465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1328,
      "step": 26133
    },
    {
      "epoch": 0.8946935980828483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1208,
      "step": 26134
    },
    {
      "epoch": 0.8947278329339268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1016,
      "step": 26135
    },
    {
      "epoch": 0.8947620677850051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1107,
      "step": 26136
    },
    {
      "epoch": 0.8947963026360836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 26137
    },
    {
      "epoch": 0.8948305374871619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1233,
      "step": 26138
    },
    {
      "epoch": 0.8948647723382404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0992,
      "step": 26139
    },
    {
      "epoch": 0.8948990071893187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1136,
      "step": 26140
    },
    {
      "epoch": 0.8949332420403971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1089,
      "step": 26141
    },
    {
      "epoch": 0.8949674768914755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1092,
      "step": 26142
    },
    {
      "epoch": 0.8950017117425539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1029,
      "step": 26143
    },
    {
      "epoch": 0.8950359465936323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.108,
      "step": 26144
    },
    {
      "epoch": 0.8950701814447107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1143,
      "step": 26145
    },
    {
      "epoch": 0.8951044162957891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1011,
      "step": 26146
    },
    {
      "epoch": 0.8951386511468675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1144,
      "step": 26147
    },
    {
      "epoch": 0.8951728859979459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1181,
      "step": 26148
    },
    {
      "epoch": 0.8952071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1074,
      "step": 26149
    },
    {
      "epoch": 0.8952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1072,
      "step": 26150
    },
    {
      "epoch": 0.8952755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0994,
      "step": 26151
    },
    {
      "epoch": 0.8953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0985,
      "step": 26152
    },
    {
      "epoch": 0.8953440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0909,
      "step": 26153
    },
    {
      "epoch": 0.8953782951044162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1093,
      "step": 26154
    },
    {
      "epoch": 0.8954125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.108,
      "step": 26155
    },
    {
      "epoch": 0.895446764806573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1186,
      "step": 26156
    },
    {
      "epoch": 0.8954809996576515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1026,
      "step": 26157
    },
    {
      "epoch": 0.8955152345087299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1069,
      "step": 26158
    },
    {
      "epoch": 0.8955494693598083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0928,
      "step": 26159
    },
    {
      "epoch": 0.8955837042108867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1089,
      "step": 26160
    },
    {
      "epoch": 0.8956179390619651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1005,
      "step": 26161
    },
    {
      "epoch": 0.8956521739130435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1207,
      "step": 26162
    },
    {
      "epoch": 0.8956864087641219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0994,
      "step": 26163
    },
    {
      "epoch": 0.8957206436152003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0941,
      "step": 26164
    },
    {
      "epoch": 0.8957548784662787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1162,
      "step": 26165
    },
    {
      "epoch": 0.8957891133173571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0864,
      "step": 26166
    },
    {
      "epoch": 0.8958233481684355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1226,
      "step": 26167
    },
    {
      "epoch": 0.8958575830195139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1092,
      "step": 26168
    },
    {
      "epoch": 0.8958918178705922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1126,
      "step": 26169
    },
    {
      "epoch": 0.8959260527216707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1262,
      "step": 26170
    },
    {
      "epoch": 0.895960287572749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1038,
      "step": 26171
    },
    {
      "epoch": 0.8959945224238275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0978,
      "step": 26172
    },
    {
      "epoch": 0.8960287572749058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.101,
      "step": 26173
    },
    {
      "epoch": 0.8960629921259843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1039,
      "step": 26174
    },
    {
      "epoch": 0.8960972269770626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1334,
      "step": 26175
    },
    {
      "epoch": 0.8961314618281411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 26176
    },
    {
      "epoch": 0.8961656966792194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1137,
      "step": 26177
    },
    {
      "epoch": 0.8961999315302979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1121,
      "step": 26178
    },
    {
      "epoch": 0.8962341663813762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.106,
      "step": 26179
    },
    {
      "epoch": 0.8962684012324547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1183,
      "step": 26180
    },
    {
      "epoch": 0.896302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1102,
      "step": 26181
    },
    {
      "epoch": 0.8963368709346115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1202,
      "step": 26182
    },
    {
      "epoch": 0.8963711057856898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1097,
      "step": 26183
    },
    {
      "epoch": 0.8964053406367682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1085,
      "step": 26184
    },
    {
      "epoch": 0.8964395754878466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.108,
      "step": 26185
    },
    {
      "epoch": 0.896473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1263,
      "step": 26186
    },
    {
      "epoch": 0.8965080451900034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0971,
      "step": 26187
    },
    {
      "epoch": 0.8965422800410818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1231,
      "step": 26188
    },
    {
      "epoch": 0.8965765148921602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1147,
      "step": 26189
    },
    {
      "epoch": 0.8966107497432386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.084,
      "step": 26190
    },
    {
      "epoch": 0.896644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1138,
      "step": 26191
    },
    {
      "epoch": 0.8966792194453954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1077,
      "step": 26192
    },
    {
      "epoch": 0.8967134542964738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.115,
      "step": 26193
    },
    {
      "epoch": 0.8967476891475522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1068,
      "step": 26194
    },
    {
      "epoch": 0.8967819239986307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.103,
      "step": 26195
    },
    {
      "epoch": 0.896816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1197,
      "step": 26196
    },
    {
      "epoch": 0.8968503937007875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1074,
      "step": 26197
    },
    {
      "epoch": 0.8968846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1172,
      "step": 26198
    },
    {
      "epoch": 0.8969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1046,
      "step": 26199
    },
    {
      "epoch": 0.8969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1134,
      "step": 26200
    },
    {
      "epoch": 0.896987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1202,
      "step": 26201
    },
    {
      "epoch": 0.8970215679561794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1119,
      "step": 26202
    },
    {
      "epoch": 0.8970558028072578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1164,
      "step": 26203
    },
    {
      "epoch": 0.8970900376583362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1126,
      "step": 26204
    },
    {
      "epoch": 0.8971242725094146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1001,
      "step": 26205
    },
    {
      "epoch": 0.897158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1159,
      "step": 26206
    },
    {
      "epoch": 0.8971927422115714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1179,
      "step": 26207
    },
    {
      "epoch": 0.8972269770626498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1098,
      "step": 26208
    },
    {
      "epoch": 0.8972612119137282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 26209
    },
    {
      "epoch": 0.8972954467648065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1171,
      "step": 26210
    },
    {
      "epoch": 0.897329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.102,
      "step": 26211
    },
    {
      "epoch": 0.8973639164669633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0902,
      "step": 26212
    },
    {
      "epoch": 0.8973981513180418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.085,
      "step": 26213
    },
    {
      "epoch": 0.8974323861691201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1093,
      "step": 26214
    },
    {
      "epoch": 0.8974666210201986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1092,
      "step": 26215
    },
    {
      "epoch": 0.8975008558712769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.111,
      "step": 26216
    },
    {
      "epoch": 0.8975350907223554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1052,
      "step": 26217
    },
    {
      "epoch": 0.8975693255734337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.106,
      "step": 26218
    },
    {
      "epoch": 0.8976035604245122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1025,
      "step": 26219
    },
    {
      "epoch": 0.8976377952755905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1097,
      "step": 26220
    },
    {
      "epoch": 0.897672030126669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1131,
      "step": 26221
    },
    {
      "epoch": 0.8977062649777473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1089,
      "step": 26222
    },
    {
      "epoch": 0.8977404998288258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1084,
      "step": 26223
    },
    {
      "epoch": 0.8977747346799041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1057,
      "step": 26224
    },
    {
      "epoch": 0.8978089695309825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1195,
      "step": 26225
    },
    {
      "epoch": 0.8978432043820609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1042,
      "step": 26226
    },
    {
      "epoch": 0.8978774392331393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1035,
      "step": 26227
    },
    {
      "epoch": 0.8979116740842177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1185,
      "step": 26228
    },
    {
      "epoch": 0.8979459089352961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1107,
      "step": 26229
    },
    {
      "epoch": 0.8979801437863745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0974,
      "step": 26230
    },
    {
      "epoch": 0.8980143786374529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1181,
      "step": 26231
    },
    {
      "epoch": 0.8980486134885314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0816,
      "step": 26232
    },
    {
      "epoch": 0.8980828483396097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1063,
      "step": 26233
    },
    {
      "epoch": 0.8981170831906882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1072,
      "step": 26234
    },
    {
      "epoch": 0.8981513180417665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1103,
      "step": 26235
    },
    {
      "epoch": 0.898185552892845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1157,
      "step": 26236
    },
    {
      "epoch": 0.8982197877439233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1198,
      "step": 26237
    },
    {
      "epoch": 0.8982540225950018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0993,
      "step": 26238
    },
    {
      "epoch": 0.8982882574460801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1038,
      "step": 26239
    },
    {
      "epoch": 0.8983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1043,
      "step": 26240
    },
    {
      "epoch": 0.8983567271482369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1153,
      "step": 26241
    },
    {
      "epoch": 0.8983909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1135,
      "step": 26242
    },
    {
      "epoch": 0.8984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0988,
      "step": 26243
    },
    {
      "epoch": 0.8984594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1147,
      "step": 26244
    },
    {
      "epoch": 0.8984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0963,
      "step": 26245
    },
    {
      "epoch": 0.8985279014036289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0988,
      "step": 26246
    },
    {
      "epoch": 0.8985621362547073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0958,
      "step": 26247
    },
    {
      "epoch": 0.8985963711057857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1187,
      "step": 26248
    },
    {
      "epoch": 0.8986306059568641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1078,
      "step": 26249
    },
    {
      "epoch": 0.8986648408079425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1068,
      "step": 26250
    },
    {
      "epoch": 0.8986990756590209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1158,
      "step": 26251
    },
    {
      "epoch": 0.8987333105100993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0959,
      "step": 26252
    },
    {
      "epoch": 0.8987675453611776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0975,
      "step": 26253
    },
    {
      "epoch": 0.8988017802122561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1162,
      "step": 26254
    },
    {
      "epoch": 0.8988360150633344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1023,
      "step": 26255
    },
    {
      "epoch": 0.8988702499144129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1035,
      "step": 26256
    },
    {
      "epoch": 0.8989044847654912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1252,
      "step": 26257
    },
    {
      "epoch": 0.8989387196165697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0947,
      "step": 26258
    },
    {
      "epoch": 0.898972954467648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1106,
      "step": 26259
    },
    {
      "epoch": 0.8990071893187265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0967,
      "step": 26260
    },
    {
      "epoch": 0.8990414241698048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1066,
      "step": 26261
    },
    {
      "epoch": 0.8990756590208833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1202,
      "step": 26262
    },
    {
      "epoch": 0.8991098938719616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1049,
      "step": 26263
    },
    {
      "epoch": 0.8991441287230401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0949,
      "step": 26264
    },
    {
      "epoch": 0.8991783635741184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1178,
      "step": 26265
    },
    {
      "epoch": 0.8992125984251969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1136,
      "step": 26266
    },
    {
      "epoch": 0.8992468332762753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1143,
      "step": 26267
    },
    {
      "epoch": 0.8992810681273536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1091,
      "step": 26268
    },
    {
      "epoch": 0.8993153029784321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1055,
      "step": 26269
    },
    {
      "epoch": 0.8993495378295104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1013,
      "step": 26270
    },
    {
      "epoch": 0.8993837726805889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1031,
      "step": 26271
    },
    {
      "epoch": 0.8994180075316672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1132,
      "step": 26272
    },
    {
      "epoch": 0.8994522423827457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1063,
      "step": 26273
    },
    {
      "epoch": 0.899486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.101,
      "step": 26274
    },
    {
      "epoch": 0.8995207120849025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1141,
      "step": 26275
    },
    {
      "epoch": 0.8995549469359808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1024,
      "step": 26276
    },
    {
      "epoch": 0.8995891817870593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1035,
      "step": 26277
    },
    {
      "epoch": 0.8996234166381376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1048,
      "step": 26278
    },
    {
      "epoch": 0.8996576514892161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1047,
      "step": 26279
    },
    {
      "epoch": 0.8996918863402944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1164,
      "step": 26280
    },
    {
      "epoch": 0.8997261211913729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1096,
      "step": 26281
    },
    {
      "epoch": 0.8997603560424512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1003,
      "step": 26282
    },
    {
      "epoch": 0.8997945908935296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1162,
      "step": 26283
    },
    {
      "epoch": 0.899828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 26284
    },
    {
      "epoch": 0.8998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1218,
      "step": 26285
    },
    {
      "epoch": 0.8998972954467648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1191,
      "step": 26286
    },
    {
      "epoch": 0.8999315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1064,
      "step": 26287
    },
    {
      "epoch": 0.8999657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1145,
      "step": 26288
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1151,
      "step": 26289
    },
    {
      "epoch": 0.9000342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1061,
      "step": 26290
    },
    {
      "epoch": 0.9000684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1083,
      "step": 26291
    },
    {
      "epoch": 0.9001027045532352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1042,
      "step": 26292
    },
    {
      "epoch": 0.9001369394043136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1085,
      "step": 26293
    },
    {
      "epoch": 0.900171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0962,
      "step": 26294
    },
    {
      "epoch": 0.9002054091064704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1075,
      "step": 26295
    },
    {
      "epoch": 0.9002396439575487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1167,
      "step": 26296
    },
    {
      "epoch": 0.9002738788086272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0954,
      "step": 26297
    },
    {
      "epoch": 0.9003081136597055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1034,
      "step": 26298
    },
    {
      "epoch": 0.900342348510784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.11,
      "step": 26299
    },
    {
      "epoch": 0.9003765833618623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1042,
      "step": 26300
    },
    {
      "epoch": 0.9004108182129408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1066,
      "step": 26301
    },
    {
      "epoch": 0.9004450530640191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1082,
      "step": 26302
    },
    {
      "epoch": 0.9004792879150976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0977,
      "step": 26303
    },
    {
      "epoch": 0.900513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1,
      "step": 26304
    },
    {
      "epoch": 0.9005477576172544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1098,
      "step": 26305
    },
    {
      "epoch": 0.9005819924683328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1026,
      "step": 26306
    },
    {
      "epoch": 0.9006162273194112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.098,
      "step": 26307
    },
    {
      "epoch": 0.9006504621704896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1192,
      "step": 26308
    },
    {
      "epoch": 0.900684697021568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1134,
      "step": 26309
    },
    {
      "epoch": 0.9007189318726464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0795,
      "step": 26310
    },
    {
      "epoch": 0.9007531667237247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0949,
      "step": 26311
    },
    {
      "epoch": 0.9007874015748032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1168,
      "step": 26312
    },
    {
      "epoch": 0.9008216364258815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1091,
      "step": 26313
    },
    {
      "epoch": 0.90085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0987,
      "step": 26314
    },
    {
      "epoch": 0.9008901061280383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1104,
      "step": 26315
    },
    {
      "epoch": 0.9009243409791168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1079,
      "step": 26316
    },
    {
      "epoch": 0.9009585758301951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1129,
      "step": 26317
    },
    {
      "epoch": 0.9009928106812736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1042,
      "step": 26318
    },
    {
      "epoch": 0.9010270455323519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 26319
    },
    {
      "epoch": 0.9010612803834304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1025,
      "step": 26320
    },
    {
      "epoch": 0.9010955152345087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1005,
      "step": 26321
    },
    {
      "epoch": 0.9011297500855872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1049,
      "step": 26322
    },
    {
      "epoch": 0.9011639849366655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1105,
      "step": 26323
    },
    {
      "epoch": 0.901198219787744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1042,
      "step": 26324
    },
    {
      "epoch": 0.9012324546388223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.09,
      "step": 26325
    },
    {
      "epoch": 0.9012666894899007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1016,
      "step": 26326
    },
    {
      "epoch": 0.9013009243409791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1036,
      "step": 26327
    },
    {
      "epoch": 0.9013351591920575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1046,
      "step": 26328
    },
    {
      "epoch": 0.9013693940431359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1059,
      "step": 26329
    },
    {
      "epoch": 0.9014036288942143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1014,
      "step": 26330
    },
    {
      "epoch": 0.9014378637452927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1133,
      "step": 26331
    },
    {
      "epoch": 0.9014720985963711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1036,
      "step": 26332
    },
    {
      "epoch": 0.9015063334474495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1089,
      "step": 26333
    },
    {
      "epoch": 0.9015405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0969,
      "step": 26334
    },
    {
      "epoch": 0.9015748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 26335
    },
    {
      "epoch": 0.9016090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0999,
      "step": 26336
    },
    {
      "epoch": 0.901643272851763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1131,
      "step": 26337
    },
    {
      "epoch": 0.9016775077028415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1143,
      "step": 26338
    },
    {
      "epoch": 0.9017117425539198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 26339
    },
    {
      "epoch": 0.9017459774049983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1056,
      "step": 26340
    },
    {
      "epoch": 0.9017802122560767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1035,
      "step": 26341
    },
    {
      "epoch": 0.9018144471071551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1209,
      "step": 26342
    },
    {
      "epoch": 0.9018486819582335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1115,
      "step": 26343
    },
    {
      "epoch": 0.9018829168093119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1187,
      "step": 26344
    },
    {
      "epoch": 0.9019171516603903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1113,
      "step": 26345
    },
    {
      "epoch": 0.9019513865114687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1152,
      "step": 26346
    },
    {
      "epoch": 0.9019856213625471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1125,
      "step": 26347
    },
    {
      "epoch": 0.9020198562136255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1138,
      "step": 26348
    },
    {
      "epoch": 0.9020540910647039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1253,
      "step": 26349
    },
    {
      "epoch": 0.9020883259157823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1172,
      "step": 26350
    },
    {
      "epoch": 0.9021225607668607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1225,
      "step": 26351
    },
    {
      "epoch": 0.902156795617939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0964,
      "step": 26352
    },
    {
      "epoch": 0.9021910304690175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1122,
      "step": 26353
    },
    {
      "epoch": 0.9022252653200958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 26354
    },
    {
      "epoch": 0.9022595001711743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1133,
      "step": 26355
    },
    {
      "epoch": 0.9022937350222526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1004,
      "step": 26356
    },
    {
      "epoch": 0.9023279698733311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1226,
      "step": 26357
    },
    {
      "epoch": 0.9023622047244094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1089,
      "step": 26358
    },
    {
      "epoch": 0.9023964395754879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.11,
      "step": 26359
    },
    {
      "epoch": 0.9024306744265662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1221,
      "step": 26360
    },
    {
      "epoch": 0.9024649092776447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1003,
      "step": 26361
    },
    {
      "epoch": 0.902499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1198,
      "step": 26362
    },
    {
      "epoch": 0.9025333789798015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1094,
      "step": 26363
    },
    {
      "epoch": 0.9025676138308798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1149,
      "step": 26364
    },
    {
      "epoch": 0.9026018486819583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 26365
    },
    {
      "epoch": 0.9026360835330366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0966,
      "step": 26366
    },
    {
      "epoch": 0.902670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1035,
      "step": 26367
    },
    {
      "epoch": 0.9027045532351934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1206,
      "step": 26368
    },
    {
      "epoch": 0.9027387880862718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1106,
      "step": 26369
    },
    {
      "epoch": 0.9027730229373502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1123,
      "step": 26370
    },
    {
      "epoch": 0.9028072577884286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1068,
      "step": 26371
    },
    {
      "epoch": 0.902841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1238,
      "step": 26372
    },
    {
      "epoch": 0.9028757274905854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1023,
      "step": 26373
    },
    {
      "epoch": 0.9029099623416638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1003,
      "step": 26374
    },
    {
      "epoch": 0.9029441971927422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1207,
      "step": 26375
    },
    {
      "epoch": 0.9029784320438206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0944,
      "step": 26376
    },
    {
      "epoch": 0.903012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1057,
      "step": 26377
    },
    {
      "epoch": 0.9030469017459775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 26378
    },
    {
      "epoch": 0.9030811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1009,
      "step": 26379
    },
    {
      "epoch": 0.9031153714481343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1169,
      "step": 26380
    },
    {
      "epoch": 0.9031496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1235,
      "step": 26381
    },
    {
      "epoch": 0.903183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1356,
      "step": 26382
    },
    {
      "epoch": 0.9032180760013694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1056,
      "step": 26383
    },
    {
      "epoch": 0.9032523108524478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1184,
      "step": 26384
    },
    {
      "epoch": 0.9032865457035262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.107,
      "step": 26385
    },
    {
      "epoch": 0.9033207805546046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1109,
      "step": 26386
    },
    {
      "epoch": 0.903355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1068,
      "step": 26387
    },
    {
      "epoch": 0.9033892502567614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1002,
      "step": 26388
    },
    {
      "epoch": 0.9034234851078398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1046,
      "step": 26389
    },
    {
      "epoch": 0.9034577199589182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0996,
      "step": 26390
    },
    {
      "epoch": 0.9034919548099966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0977,
      "step": 26391
    },
    {
      "epoch": 0.903526189661075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1219,
      "step": 26392
    },
    {
      "epoch": 0.9035604245121533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1072,
      "step": 26393
    },
    {
      "epoch": 0.9035946593632318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1228,
      "step": 26394
    },
    {
      "epoch": 0.9036288942143101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1146,
      "step": 26395
    },
    {
      "epoch": 0.9036631290653886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 26396
    },
    {
      "epoch": 0.9036973639164669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1176,
      "step": 26397
    },
    {
      "epoch": 0.9037315987675454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 26398
    },
    {
      "epoch": 0.9037658336186237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1035,
      "step": 26399
    },
    {
      "epoch": 0.9038000684697022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1136,
      "step": 26400
    },
    {
      "epoch": 0.9038343033207805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1198,
      "step": 26401
    },
    {
      "epoch": 0.903868538171859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1012,
      "step": 26402
    },
    {
      "epoch": 0.9039027730229373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1064,
      "step": 26403
    },
    {
      "epoch": 0.9039370078740158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0964,
      "step": 26404
    },
    {
      "epoch": 0.9039712427250941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1089,
      "step": 26405
    },
    {
      "epoch": 0.9040054775761726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1256,
      "step": 26406
    },
    {
      "epoch": 0.9040397124272509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1023,
      "step": 26407
    },
    {
      "epoch": 0.9040739472783293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.113,
      "step": 26408
    },
    {
      "epoch": 0.9041081821294077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0996,
      "step": 26409
    },
    {
      "epoch": 0.9041424169804861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0896,
      "step": 26410
    },
    {
      "epoch": 0.9041766518315645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1017,
      "step": 26411
    },
    {
      "epoch": 0.9042108866826429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.108,
      "step": 26412
    },
    {
      "epoch": 0.9042451215337213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1087,
      "step": 26413
    },
    {
      "epoch": 0.9042793563847997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1046,
      "step": 26414
    },
    {
      "epoch": 0.9043135912358782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1078,
      "step": 26415
    },
    {
      "epoch": 0.9043478260869565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0976,
      "step": 26416
    },
    {
      "epoch": 0.904382060938035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0985,
      "step": 26417
    },
    {
      "epoch": 0.9044162957891133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1255,
      "step": 26418
    },
    {
      "epoch": 0.9044505306401918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1101,
      "step": 26419
    },
    {
      "epoch": 0.9044847654912701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.102,
      "step": 26420
    },
    {
      "epoch": 0.9045190003423486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0972,
      "step": 26421
    },
    {
      "epoch": 0.9045532351934269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1085,
      "step": 26422
    },
    {
      "epoch": 0.9045874700445053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1013,
      "step": 26423
    },
    {
      "epoch": 0.9046217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1184,
      "step": 26424
    },
    {
      "epoch": 0.9046559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1111,
      "step": 26425
    },
    {
      "epoch": 0.9046901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 26426
    },
    {
      "epoch": 0.9047244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1277,
      "step": 26427
    },
    {
      "epoch": 0.9047586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1136,
      "step": 26428
    },
    {
      "epoch": 0.9047928791509757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1112,
      "step": 26429
    },
    {
      "epoch": 0.9048271140020541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.109,
      "step": 26430
    },
    {
      "epoch": 0.9048613488531325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1075,
      "step": 26431
    },
    {
      "epoch": 0.9048955837042109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1062,
      "step": 26432
    },
    {
      "epoch": 0.9049298185552893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1068,
      "step": 26433
    },
    {
      "epoch": 0.9049640534063677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0984,
      "step": 26434
    },
    {
      "epoch": 0.9049982882574461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1107,
      "step": 26435
    },
    {
      "epoch": 0.9050325231085244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1147,
      "step": 26436
    },
    {
      "epoch": 0.9050667579596029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1038,
      "step": 26437
    },
    {
      "epoch": 0.9051009928106812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1075,
      "step": 26438
    },
    {
      "epoch": 0.9051352276617597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0974,
      "step": 26439
    },
    {
      "epoch": 0.905169462512838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0977,
      "step": 26440
    },
    {
      "epoch": 0.9052036973639165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0891,
      "step": 26441
    },
    {
      "epoch": 0.9052379322149948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0982,
      "step": 26442
    },
    {
      "epoch": 0.9052721670660733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1147,
      "step": 26443
    },
    {
      "epoch": 0.9053064019171516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1034,
      "step": 26444
    },
    {
      "epoch": 0.9053406367682301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1041,
      "step": 26445
    },
    {
      "epoch": 0.9053748716193084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0882,
      "step": 26446
    },
    {
      "epoch": 0.9054091064703869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1096,
      "step": 26447
    },
    {
      "epoch": 0.9054433413214652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0953,
      "step": 26448
    },
    {
      "epoch": 0.9054775761725437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1097,
      "step": 26449
    },
    {
      "epoch": 0.905511811023622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.106,
      "step": 26450
    },
    {
      "epoch": 0.9055460458747004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0898,
      "step": 26451
    },
    {
      "epoch": 0.9055802807257789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1024,
      "step": 26452
    },
    {
      "epoch": 0.9056145155768572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0904,
      "step": 26453
    },
    {
      "epoch": 0.9056487504279357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1228,
      "step": 26454
    },
    {
      "epoch": 0.905682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1126,
      "step": 26455
    },
    {
      "epoch": 0.9057172201300925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0978,
      "step": 26456
    },
    {
      "epoch": 0.9057514549811708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1133,
      "step": 26457
    },
    {
      "epoch": 0.9057856898322493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0961,
      "step": 26458
    },
    {
      "epoch": 0.9058199246833276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1128,
      "step": 26459
    },
    {
      "epoch": 0.9058541595344061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1121,
      "step": 26460
    },
    {
      "epoch": 0.9058883943854844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0977,
      "step": 26461
    },
    {
      "epoch": 0.9059226292365629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1183,
      "step": 26462
    },
    {
      "epoch": 0.9059568640876412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.103,
      "step": 26463
    },
    {
      "epoch": 0.9059910989387197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1041,
      "step": 26464
    },
    {
      "epoch": 0.906025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1028,
      "step": 26465
    },
    {
      "epoch": 0.9060595686408764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0927,
      "step": 26466
    },
    {
      "epoch": 0.9060938034919548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0909,
      "step": 26467
    },
    {
      "epoch": 0.9061280383430332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0943,
      "step": 26468
    },
    {
      "epoch": 0.9061622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0831,
      "step": 26469
    },
    {
      "epoch": 0.90619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 26470
    },
    {
      "epoch": 0.9062307428962684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0976,
      "step": 26471
    },
    {
      "epoch": 0.9062649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1211,
      "step": 26472
    },
    {
      "epoch": 0.9062992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1203,
      "step": 26473
    },
    {
      "epoch": 0.9063334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1011,
      "step": 26474
    },
    {
      "epoch": 0.906367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1082,
      "step": 26475
    },
    {
      "epoch": 0.9064019171516604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0966,
      "step": 26476
    },
    {
      "epoch": 0.9064361520027387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0965,
      "step": 26477
    },
    {
      "epoch": 0.9064703868538172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.11,
      "step": 26478
    },
    {
      "epoch": 0.9065046217048955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0916,
      "step": 26479
    },
    {
      "epoch": 0.906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1069,
      "step": 26480
    },
    {
      "epoch": 0.9065730914070523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0988,
      "step": 26481
    },
    {
      "epoch": 0.9066073262581308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0941,
      "step": 26482
    },
    {
      "epoch": 0.9066415611092091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0805,
      "step": 26483
    },
    {
      "epoch": 0.9066757959602876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0958,
      "step": 26484
    },
    {
      "epoch": 0.9067100308113659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1263,
      "step": 26485
    },
    {
      "epoch": 0.9067442656624444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1143,
      "step": 26486
    },
    {
      "epoch": 0.9067785005135227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1086,
      "step": 26487
    },
    {
      "epoch": 0.9068127353646012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1002,
      "step": 26488
    },
    {
      "epoch": 0.9068469702156796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1127,
      "step": 26489
    },
    {
      "epoch": 0.906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0788,
      "step": 26490
    },
    {
      "epoch": 0.9069154399178364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0909,
      "step": 26491
    },
    {
      "epoch": 0.9069496747689147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1078,
      "step": 26492
    },
    {
      "epoch": 0.9069839096199932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1045,
      "step": 26493
    },
    {
      "epoch": 0.9070181444710715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1028,
      "step": 26494
    },
    {
      "epoch": 0.90705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.123,
      "step": 26495
    },
    {
      "epoch": 0.9070866141732283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1043,
      "step": 26496
    },
    {
      "epoch": 0.9071208490243068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0964,
      "step": 26497
    },
    {
      "epoch": 0.9071550838753851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0987,
      "step": 26498
    },
    {
      "epoch": 0.9071893187264636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0988,
      "step": 26499
    },
    {
      "epoch": 0.9072235535775419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1128,
      "step": 26500
    },
    {
      "epoch": 0.9072577884286204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0899,
      "step": 26501
    },
    {
      "epoch": 0.9072920232796987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0957,
      "step": 26502
    },
    {
      "epoch": 0.9073262581307772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.113,
      "step": 26503
    },
    {
      "epoch": 0.9073604929818555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1047,
      "step": 26504
    },
    {
      "epoch": 0.907394727832934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1024,
      "step": 26505
    },
    {
      "epoch": 0.9074289626840123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1202,
      "step": 26506
    },
    {
      "epoch": 0.9074631975350907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0947,
      "step": 26507
    },
    {
      "epoch": 0.9074974323861691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1067,
      "step": 26508
    },
    {
      "epoch": 0.9075316672372475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 26509
    },
    {
      "epoch": 0.9075659020883259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1032,
      "step": 26510
    },
    {
      "epoch": 0.9076001369394043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1031,
      "step": 26511
    },
    {
      "epoch": 0.9076343717904827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.107,
      "step": 26512
    },
    {
      "epoch": 0.9076686066415611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1064,
      "step": 26513
    },
    {
      "epoch": 0.9077028414926395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1185,
      "step": 26514
    },
    {
      "epoch": 0.9077370763437179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0975,
      "step": 26515
    },
    {
      "epoch": 0.9077713111947963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1037,
      "step": 26516
    },
    {
      "epoch": 0.9078055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1177,
      "step": 26517
    },
    {
      "epoch": 0.907839780896953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1046,
      "step": 26518
    },
    {
      "epoch": 0.9078740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1065,
      "step": 26519
    },
    {
      "epoch": 0.9079082505991098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0977,
      "step": 26520
    },
    {
      "epoch": 0.9079424854501883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1052,
      "step": 26521
    },
    {
      "epoch": 0.9079767203012666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1076,
      "step": 26522
    },
    {
      "epoch": 0.9080109551523451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1059,
      "step": 26523
    },
    {
      "epoch": 0.9080451900034234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1109,
      "step": 26524
    },
    {
      "epoch": 0.9080794248545019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0967,
      "step": 26525
    },
    {
      "epoch": 0.9081136597055803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.098,
      "step": 26526
    },
    {
      "epoch": 0.9081478945566587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.107,
      "step": 26527
    },
    {
      "epoch": 0.9081821294077371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 26528
    },
    {
      "epoch": 0.9082163642588155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1065,
      "step": 26529
    },
    {
      "epoch": 0.9082505991098939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1108,
      "step": 26530
    },
    {
      "epoch": 0.9082848339609723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0983,
      "step": 26531
    },
    {
      "epoch": 0.9083190688120507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1148,
      "step": 26532
    },
    {
      "epoch": 0.908353303663129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0931,
      "step": 26533
    },
    {
      "epoch": 0.9083875385142075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0937,
      "step": 26534
    },
    {
      "epoch": 0.9084217733652858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.091,
      "step": 26535
    },
    {
      "epoch": 0.9084560082163643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1136,
      "step": 26536
    },
    {
      "epoch": 0.9084902430674426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1058,
      "step": 26537
    },
    {
      "epoch": 0.9085244779185211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.108,
      "step": 26538
    },
    {
      "epoch": 0.9085587127695994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1139,
      "step": 26539
    },
    {
      "epoch": 0.9085929476206779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1101,
      "step": 26540
    },
    {
      "epoch": 0.9086271824717562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1,
      "step": 26541
    },
    {
      "epoch": 0.9086614173228347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1003,
      "step": 26542
    },
    {
      "epoch": 0.908695652173913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1026,
      "step": 26543
    },
    {
      "epoch": 0.9087298870249915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1094,
      "step": 26544
    },
    {
      "epoch": 0.9087641218760698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0881,
      "step": 26545
    },
    {
      "epoch": 0.9087983567271483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0928,
      "step": 26546
    },
    {
      "epoch": 0.9088325915782266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1019,
      "step": 26547
    },
    {
      "epoch": 0.908866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0932,
      "step": 26548
    },
    {
      "epoch": 0.9089010612803834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1086,
      "step": 26549
    },
    {
      "epoch": 0.9089352961314618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1085,
      "step": 26550
    },
    {
      "epoch": 0.9089695309825402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1189,
      "step": 26551
    },
    {
      "epoch": 0.9090037658336186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 26552
    },
    {
      "epoch": 0.909038000684697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.096,
      "step": 26553
    },
    {
      "epoch": 0.9090722355357754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1041,
      "step": 26554
    },
    {
      "epoch": 0.9091064703868538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1146,
      "step": 26555
    },
    {
      "epoch": 0.9091407052379322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1176,
      "step": 26556
    },
    {
      "epoch": 0.9091749400890106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1028,
      "step": 26557
    },
    {
      "epoch": 0.909209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0869,
      "step": 26558
    },
    {
      "epoch": 0.9092434097911674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1059,
      "step": 26559
    },
    {
      "epoch": 0.9092776446422458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1118,
      "step": 26560
    },
    {
      "epoch": 0.9093118794933241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1063,
      "step": 26561
    },
    {
      "epoch": 0.9093461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0876,
      "step": 26562
    },
    {
      "epoch": 0.909380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0849,
      "step": 26563
    },
    {
      "epoch": 0.9094145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1129,
      "step": 26564
    },
    {
      "epoch": 0.9094488188976378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 26565
    },
    {
      "epoch": 0.9094830537487162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1036,
      "step": 26566
    },
    {
      "epoch": 0.9095172885997946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.113,
      "step": 26567
    },
    {
      "epoch": 0.909551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.113,
      "step": 26568
    },
    {
      "epoch": 0.9095857583019514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1157,
      "step": 26569
    },
    {
      "epoch": 0.9096199931530298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1078,
      "step": 26570
    },
    {
      "epoch": 0.9096542280041082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1026,
      "step": 26571
    },
    {
      "epoch": 0.9096884628551866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0952,
      "step": 26572
    },
    {
      "epoch": 0.909722697706265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1122,
      "step": 26573
    },
    {
      "epoch": 0.9097569325573434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1013,
      "step": 26574
    },
    {
      "epoch": 0.9097911674084218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0789,
      "step": 26575
    },
    {
      "epoch": 0.9098254022595001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0991,
      "step": 26576
    },
    {
      "epoch": 0.9098596371105786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1046,
      "step": 26577
    },
    {
      "epoch": 0.9098938719616569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1023,
      "step": 26578
    },
    {
      "epoch": 0.9099281068127354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0891,
      "step": 26579
    },
    {
      "epoch": 0.9099623416638137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1158,
      "step": 26580
    },
    {
      "epoch": 0.9099965765148922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1047,
      "step": 26581
    },
    {
      "epoch": 0.9100308113659705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0949,
      "step": 26582
    },
    {
      "epoch": 0.910065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1003,
      "step": 26583
    },
    {
      "epoch": 0.9100992810681273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1084,
      "step": 26584
    },
    {
      "epoch": 0.9101335159192058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0898,
      "step": 26585
    },
    {
      "epoch": 0.9101677507702841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1156,
      "step": 26586
    },
    {
      "epoch": 0.9102019856213626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1187,
      "step": 26587
    },
    {
      "epoch": 0.9102362204724409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1008,
      "step": 26588
    },
    {
      "epoch": 0.9102704553235194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.11,
      "step": 26589
    },
    {
      "epoch": 0.9103046901745977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1191,
      "step": 26590
    },
    {
      "epoch": 0.9103389250256761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0922,
      "step": 26591
    },
    {
      "epoch": 0.9103731598767545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1111,
      "step": 26592
    },
    {
      "epoch": 0.9104073947278329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1008,
      "step": 26593
    },
    {
      "epoch": 0.9104416295789113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0971,
      "step": 26594
    },
    {
      "epoch": 0.9104758644299897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1078,
      "step": 26595
    },
    {
      "epoch": 0.9105100992810681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1185,
      "step": 26596
    },
    {
      "epoch": 0.9105443341321465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1182,
      "step": 26597
    },
    {
      "epoch": 0.9105785689832249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 26598
    },
    {
      "epoch": 0.9106128038343033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0996,
      "step": 26599
    },
    {
      "epoch": 0.9106470386853818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 26600
    },
    {
      "epoch": 0.9106812735364601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0995,
      "step": 26601
    },
    {
      "epoch": 0.9107155083875386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1213,
      "step": 26602
    },
    {
      "epoch": 0.9107497432386169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1134,
      "step": 26603
    },
    {
      "epoch": 0.9107839780896954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1081,
      "step": 26604
    },
    {
      "epoch": 0.9108182129407737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 26605
    },
    {
      "epoch": 0.9108524477918521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1187,
      "step": 26606
    },
    {
      "epoch": 0.9108866826429305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1154,
      "step": 26607
    },
    {
      "epoch": 0.9109209174940089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0925,
      "step": 26608
    },
    {
      "epoch": 0.9109551523450873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1068,
      "step": 26609
    },
    {
      "epoch": 0.9109893871961657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 26610
    },
    {
      "epoch": 0.9110236220472441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0987,
      "step": 26611
    },
    {
      "epoch": 0.9110578568983225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0975,
      "step": 26612
    },
    {
      "epoch": 0.9110920917494009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0874,
      "step": 26613
    },
    {
      "epoch": 0.9111263266004793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0999,
      "step": 26614
    },
    {
      "epoch": 0.9111605614515577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1047,
      "step": 26615
    },
    {
      "epoch": 0.9111947963026361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 26616
    },
    {
      "epoch": 0.9112290311537145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0989,
      "step": 26617
    },
    {
      "epoch": 0.9112632660047929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1115,
      "step": 26618
    },
    {
      "epoch": 0.9112975008558712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0916,
      "step": 26619
    },
    {
      "epoch": 0.9113317357069497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1059,
      "step": 26620
    },
    {
      "epoch": 0.911365970558028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0999,
      "step": 26621
    },
    {
      "epoch": 0.9114002054091065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1011,
      "step": 26622
    },
    {
      "epoch": 0.9114344402601848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0954,
      "step": 26623
    },
    {
      "epoch": 0.9114686751112633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0852,
      "step": 26624
    },
    {
      "epoch": 0.9115029099623416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1077,
      "step": 26625
    },
    {
      "epoch": 0.9115371448134201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0916,
      "step": 26626
    },
    {
      "epoch": 0.9115713796644984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1004,
      "step": 26627
    },
    {
      "epoch": 0.9116056145155769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0993,
      "step": 26628
    },
    {
      "epoch": 0.9116398493666552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0971,
      "step": 26629
    },
    {
      "epoch": 0.9116740842177337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0944,
      "step": 26630
    },
    {
      "epoch": 0.911708319068812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.109,
      "step": 26631
    },
    {
      "epoch": 0.9117425539198905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.094,
      "step": 26632
    },
    {
      "epoch": 0.9117767887709688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0958,
      "step": 26633
    },
    {
      "epoch": 0.9118110236220472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0927,
      "step": 26634
    },
    {
      "epoch": 0.9118452584731257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1118,
      "step": 26635
    },
    {
      "epoch": 0.911879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0998,
      "step": 26636
    },
    {
      "epoch": 0.9119137281752825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1162,
      "step": 26637
    },
    {
      "epoch": 0.9119479630263608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0897,
      "step": 26638
    },
    {
      "epoch": 0.9119821978774393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0931,
      "step": 26639
    },
    {
      "epoch": 0.9120164327285176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0873,
      "step": 26640
    },
    {
      "epoch": 0.9120506675795961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0913,
      "step": 26641
    },
    {
      "epoch": 0.9120849024306744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1133,
      "step": 26642
    },
    {
      "epoch": 0.9121191372817529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0935,
      "step": 26643
    },
    {
      "epoch": 0.9121533721328312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1091,
      "step": 26644
    },
    {
      "epoch": 0.9121876069839097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0962,
      "step": 26645
    },
    {
      "epoch": 0.912221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1047,
      "step": 26646
    },
    {
      "epoch": 0.9122560766860665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1066,
      "step": 26647
    },
    {
      "epoch": 0.9122903115371448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1082,
      "step": 26648
    },
    {
      "epoch": 0.9123245463882232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1029,
      "step": 26649
    },
    {
      "epoch": 0.9123587812393016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1044,
      "step": 26650
    },
    {
      "epoch": 0.91239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1124,
      "step": 26651
    },
    {
      "epoch": 0.9124272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0964,
      "step": 26652
    },
    {
      "epoch": 0.9124614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1042,
      "step": 26653
    },
    {
      "epoch": 0.9124957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 26654
    },
    {
      "epoch": 0.9125299554946936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1121,
      "step": 26655
    },
    {
      "epoch": 0.912564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1123,
      "step": 26656
    },
    {
      "epoch": 0.9125984251968504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1174,
      "step": 26657
    },
    {
      "epoch": 0.9126326600479288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1071,
      "step": 26658
    },
    {
      "epoch": 0.9126668948990072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1031,
      "step": 26659
    },
    {
      "epoch": 0.9127011297500855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0856,
      "step": 26660
    },
    {
      "epoch": 0.912735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1075,
      "step": 26661
    },
    {
      "epoch": 0.9127695994522423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0766,
      "step": 26662
    },
    {
      "epoch": 0.9128038343033208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1076,
      "step": 26663
    },
    {
      "epoch": 0.9128380691543991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1015,
      "step": 26664
    },
    {
      "epoch": 0.9128723040054776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1004,
      "step": 26665
    },
    {
      "epoch": 0.9129065388565559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.089,
      "step": 26666
    },
    {
      "epoch": 0.9129407737076344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0922,
      "step": 26667
    },
    {
      "epoch": 0.9129750085587127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.09,
      "step": 26668
    },
    {
      "epoch": 0.9130092434097912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1165,
      "step": 26669
    },
    {
      "epoch": 0.9130434782608695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0855,
      "step": 26670
    },
    {
      "epoch": 0.913077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1123,
      "step": 26671
    },
    {
      "epoch": 0.9131119479630264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0884,
      "step": 26672
    },
    {
      "epoch": 0.9131461828141048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1066,
      "step": 26673
    },
    {
      "epoch": 0.9131804176651832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1149,
      "step": 26674
    },
    {
      "epoch": 0.9132146525162615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1049,
      "step": 26675
    },
    {
      "epoch": 0.91324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1102,
      "step": 26676
    },
    {
      "epoch": 0.9132831222184183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0933,
      "step": 26677
    },
    {
      "epoch": 0.9133173570694968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0998,
      "step": 26678
    },
    {
      "epoch": 0.9133515919205751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1003,
      "step": 26679
    },
    {
      "epoch": 0.9133858267716536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0965,
      "step": 26680
    },
    {
      "epoch": 0.9134200616227319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1139,
      "step": 26681
    },
    {
      "epoch": 0.9134542964738104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1111,
      "step": 26682
    },
    {
      "epoch": 0.9134885313248887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 26683
    },
    {
      "epoch": 0.9135227661759672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0927,
      "step": 26684
    },
    {
      "epoch": 0.9135570010270455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.089,
      "step": 26685
    },
    {
      "epoch": 0.913591235878124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0958,
      "step": 26686
    },
    {
      "epoch": 0.9136254707292023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1044,
      "step": 26687
    },
    {
      "epoch": 0.9136597055802808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1015,
      "step": 26688
    },
    {
      "epoch": 0.9136939404313591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0851,
      "step": 26689
    },
    {
      "epoch": 0.9137281752824375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1019,
      "step": 26690
    },
    {
      "epoch": 0.9137624101335159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 26691
    },
    {
      "epoch": 0.9137966449845943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1136,
      "step": 26692
    },
    {
      "epoch": 0.9138308798356727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0969,
      "step": 26693
    },
    {
      "epoch": 0.9138651146867511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.101,
      "step": 26694
    },
    {
      "epoch": 0.9138993495378295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1178,
      "step": 26695
    },
    {
      "epoch": 0.9139335843889079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1196,
      "step": 26696
    },
    {
      "epoch": 0.9139678192399863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 26697
    },
    {
      "epoch": 0.9140020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1127,
      "step": 26698
    },
    {
      "epoch": 0.9140362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0926,
      "step": 26699
    },
    {
      "epoch": 0.9140705237932215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1,
      "step": 26700
    },
    {
      "epoch": 0.9141047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0996,
      "step": 26701
    },
    {
      "epoch": 0.9141389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1036,
      "step": 26702
    },
    {
      "epoch": 0.9141732283464566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1041,
      "step": 26703
    },
    {
      "epoch": 0.9142074631975351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1042,
      "step": 26704
    },
    {
      "epoch": 0.9142416980486134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1022,
      "step": 26705
    },
    {
      "epoch": 0.9142759328996919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1023,
      "step": 26706
    },
    {
      "epoch": 0.9143101677507702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1051,
      "step": 26707
    },
    {
      "epoch": 0.9143444026018487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1031,
      "step": 26708
    },
    {
      "epoch": 0.9143786374529271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1102,
      "step": 26709
    },
    {
      "epoch": 0.9144128723040055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0774,
      "step": 26710
    },
    {
      "epoch": 0.9144471071550839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0939,
      "step": 26711
    },
    {
      "epoch": 0.9144813420061623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1052,
      "step": 26712
    },
    {
      "epoch": 0.9145155768572407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1107,
      "step": 26713
    },
    {
      "epoch": 0.9145498117083191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0936,
      "step": 26714
    },
    {
      "epoch": 0.9145840465593975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0944,
      "step": 26715
    },
    {
      "epoch": 0.9146182814104759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0921,
      "step": 26716
    },
    {
      "epoch": 0.9146525162615543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0959,
      "step": 26717
    },
    {
      "epoch": 0.9146867511126326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0939,
      "step": 26718
    },
    {
      "epoch": 0.9147209859637111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1084,
      "step": 26719
    },
    {
      "epoch": 0.9147552208147894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1023,
      "step": 26720
    },
    {
      "epoch": 0.9147894556658679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1022,
      "step": 26721
    },
    {
      "epoch": 0.9148236905169462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1166,
      "step": 26722
    },
    {
      "epoch": 0.9148579253680247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0993,
      "step": 26723
    },
    {
      "epoch": 0.914892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0901,
      "step": 26724
    },
    {
      "epoch": 0.9149263950701815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.09,
      "step": 26725
    },
    {
      "epoch": 0.9149606299212598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1167,
      "step": 26726
    },
    {
      "epoch": 0.9149948647723383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1199,
      "step": 26727
    },
    {
      "epoch": 0.9150290996234166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1067,
      "step": 26728
    },
    {
      "epoch": 0.9150633344744951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1003,
      "step": 26729
    },
    {
      "epoch": 0.9150975693255734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0913,
      "step": 26730
    },
    {
      "epoch": 0.9151318041766519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0923,
      "step": 26731
    },
    {
      "epoch": 0.9151660390277302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1238,
      "step": 26732
    },
    {
      "epoch": 0.9152002738788086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0901,
      "step": 26733
    },
    {
      "epoch": 0.915234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0988,
      "step": 26734
    },
    {
      "epoch": 0.9152687435809654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.096,
      "step": 26735
    },
    {
      "epoch": 0.9153029784320438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0995,
      "step": 26736
    },
    {
      "epoch": 0.9153372132831222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.119,
      "step": 26737
    },
    {
      "epoch": 0.9153714481342006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.101,
      "step": 26738
    },
    {
      "epoch": 0.915405682985279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0984,
      "step": 26739
    },
    {
      "epoch": 0.9154399178363574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0855,
      "step": 26740
    },
    {
      "epoch": 0.9154741526874358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1129,
      "step": 26741
    },
    {
      "epoch": 0.9155083875385142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1018,
      "step": 26742
    },
    {
      "epoch": 0.9155426223895926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.113,
      "step": 26743
    },
    {
      "epoch": 0.915576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1141,
      "step": 26744
    },
    {
      "epoch": 0.9156110920917494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1146,
      "step": 26745
    },
    {
      "epoch": 0.9156453269428279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.096,
      "step": 26746
    },
    {
      "epoch": 0.9156795617939062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0973,
      "step": 26747
    },
    {
      "epoch": 0.9157137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0965,
      "step": 26748
    },
    {
      "epoch": 0.915748031496063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1054,
      "step": 26749
    },
    {
      "epoch": 0.9157822663471414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1084,
      "step": 26750
    },
    {
      "epoch": 0.9158165011982198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1057,
      "step": 26751
    },
    {
      "epoch": 0.9158507360492982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0991,
      "step": 26752
    },
    {
      "epoch": 0.9158849709003766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1156,
      "step": 26753
    },
    {
      "epoch": 0.915919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1028,
      "step": 26754
    },
    {
      "epoch": 0.9159534406025334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1118,
      "step": 26755
    },
    {
      "epoch": 0.9159876754536118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0999,
      "step": 26756
    },
    {
      "epoch": 0.9160219103046902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.108,
      "step": 26757
    },
    {
      "epoch": 0.9160561451557686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1063,
      "step": 26758
    },
    {
      "epoch": 0.916090380006847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0982,
      "step": 26759
    },
    {
      "epoch": 0.9161246148579254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 26760
    },
    {
      "epoch": 0.9161588497090037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 26761
    },
    {
      "epoch": 0.9161930845600822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1227,
      "step": 26762
    },
    {
      "epoch": 0.9162273194111605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.102,
      "step": 26763
    },
    {
      "epoch": 0.916261554262239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1111,
      "step": 26764
    },
    {
      "epoch": 0.9162957891133173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1126,
      "step": 26765
    },
    {
      "epoch": 0.9163300239643958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0993,
      "step": 26766
    },
    {
      "epoch": 0.9163642588154741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0915,
      "step": 26767
    },
    {
      "epoch": 0.9163984936665526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1071,
      "step": 26768
    },
    {
      "epoch": 0.9164327285176309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0945,
      "step": 26769
    },
    {
      "epoch": 0.9164669633687094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1047,
      "step": 26770
    },
    {
      "epoch": 0.9165011982197877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1169,
      "step": 26771
    },
    {
      "epoch": 0.9165354330708662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0953,
      "step": 26772
    },
    {
      "epoch": 0.9165696679219445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0958,
      "step": 26773
    },
    {
      "epoch": 0.916603902773023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1154,
      "step": 26774
    },
    {
      "epoch": 0.9166381376241013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1018,
      "step": 26775
    },
    {
      "epoch": 0.9166723724751797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1176,
      "step": 26776
    },
    {
      "epoch": 0.9167066073262581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1075,
      "step": 26777
    },
    {
      "epoch": 0.9167408421773365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1027,
      "step": 26778
    },
    {
      "epoch": 0.9167750770284149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0915,
      "step": 26779
    },
    {
      "epoch": 0.9168093118794933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1273,
      "step": 26780
    },
    {
      "epoch": 0.9168435467305717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.104,
      "step": 26781
    },
    {
      "epoch": 0.9168777815816501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1093,
      "step": 26782
    },
    {
      "epoch": 0.9169120164327286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0979,
      "step": 26783
    },
    {
      "epoch": 0.9169462512838069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0781,
      "step": 26784
    },
    {
      "epoch": 0.9169804861348854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1104,
      "step": 26785
    },
    {
      "epoch": 0.9170147209859637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1004,
      "step": 26786
    },
    {
      "epoch": 0.9170489558370422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1095,
      "step": 26787
    },
    {
      "epoch": 0.9170831906881205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0992,
      "step": 26788
    },
    {
      "epoch": 0.917117425539199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.112,
      "step": 26789
    },
    {
      "epoch": 0.9171516603902773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0951,
      "step": 26790
    },
    {
      "epoch": 0.9171858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0939,
      "step": 26791
    },
    {
      "epoch": 0.9172201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1012,
      "step": 26792
    },
    {
      "epoch": 0.9172543649435125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0939,
      "step": 26793
    },
    {
      "epoch": 0.9172885997945909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1037,
      "step": 26794
    },
    {
      "epoch": 0.9173228346456693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0891,
      "step": 26795
    },
    {
      "epoch": 0.9173570694967477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0979,
      "step": 26796
    },
    {
      "epoch": 0.9173913043478261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1066,
      "step": 26797
    },
    {
      "epoch": 0.9174255391989045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0916,
      "step": 26798
    },
    {
      "epoch": 0.9174597740499829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1096,
      "step": 26799
    },
    {
      "epoch": 0.9174940089010613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1036,
      "step": 26800
    },
    {
      "epoch": 0.9175282437521397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0997,
      "step": 26801
    },
    {
      "epoch": 0.917562478603218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.097,
      "step": 26802
    },
    {
      "epoch": 0.9175967134542965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0963,
      "step": 26803
    },
    {
      "epoch": 0.9176309483053748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 26804
    },
    {
      "epoch": 0.9176651831564533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0957,
      "step": 26805
    },
    {
      "epoch": 0.9176994180075316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0843,
      "step": 26806
    },
    {
      "epoch": 0.9177336528586101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0966,
      "step": 26807
    },
    {
      "epoch": 0.9177678877096884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0927,
      "step": 26808
    },
    {
      "epoch": 0.9178021225607669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0959,
      "step": 26809
    },
    {
      "epoch": 0.9178363574118452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0973,
      "step": 26810
    },
    {
      "epoch": 0.9178705922629237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1139,
      "step": 26811
    },
    {
      "epoch": 0.917904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.12,
      "step": 26812
    },
    {
      "epoch": 0.9179390619650805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 26813
    },
    {
      "epoch": 0.9179732968161588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1148,
      "step": 26814
    },
    {
      "epoch": 0.9180075316672373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1014,
      "step": 26815
    },
    {
      "epoch": 0.9180417665183156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0982,
      "step": 26816
    },
    {
      "epoch": 0.918076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0987,
      "step": 26817
    },
    {
      "epoch": 0.9181102362204724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1009,
      "step": 26818
    },
    {
      "epoch": 0.9181444710715508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1044,
      "step": 26819
    },
    {
      "epoch": 0.9181787059226293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1054,
      "step": 26820
    },
    {
      "epoch": 0.9182129407737076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0709,
      "step": 26821
    },
    {
      "epoch": 0.9182471756247861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0989,
      "step": 26822
    },
    {
      "epoch": 0.9182814104758644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1031,
      "step": 26823
    },
    {
      "epoch": 0.9183156453269429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0984,
      "step": 26824
    },
    {
      "epoch": 0.9183498801780212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0953,
      "step": 26825
    },
    {
      "epoch": 0.9183841150290997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0975,
      "step": 26826
    },
    {
      "epoch": 0.918418349880178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0976,
      "step": 26827
    },
    {
      "epoch": 0.9184525847312565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1086,
      "step": 26828
    },
    {
      "epoch": 0.9184868195823348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1169,
      "step": 26829
    },
    {
      "epoch": 0.9185210544334133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0979,
      "step": 26830
    },
    {
      "epoch": 0.9185552892844916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0943,
      "step": 26831
    },
    {
      "epoch": 0.91858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0902,
      "step": 26832
    },
    {
      "epoch": 0.9186237589866484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1035,
      "step": 26833
    },
    {
      "epoch": 0.9186579938377268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0985,
      "step": 26834
    },
    {
      "epoch": 0.9186922286888052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0973,
      "step": 26835
    },
    {
      "epoch": 0.9187264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0955,
      "step": 26836
    },
    {
      "epoch": 0.918760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0974,
      "step": 26837
    },
    {
      "epoch": 0.9187949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0931,
      "step": 26838
    },
    {
      "epoch": 0.9188291680931188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0912,
      "step": 26839
    },
    {
      "epoch": 0.9188634029441972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0999,
      "step": 26840
    },
    {
      "epoch": 0.9188976377952756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1025,
      "step": 26841
    },
    {
      "epoch": 0.918931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.122,
      "step": 26842
    },
    {
      "epoch": 0.9189661074974323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1195,
      "step": 26843
    },
    {
      "epoch": 0.9190003423485108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 26844
    },
    {
      "epoch": 0.9190345771995891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1136,
      "step": 26845
    },
    {
      "epoch": 0.9190688120506676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0773,
      "step": 26846
    },
    {
      "epoch": 0.9191030469017459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0867,
      "step": 26847
    },
    {
      "epoch": 0.9191372817528244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0941,
      "step": 26848
    },
    {
      "epoch": 0.9191715166039027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0878,
      "step": 26849
    },
    {
      "epoch": 0.9192057514549812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.081,
      "step": 26850
    },
    {
      "epoch": 0.9192399863060595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1102,
      "step": 26851
    },
    {
      "epoch": 0.919274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0951,
      "step": 26852
    },
    {
      "epoch": 0.9193084560082163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0969,
      "step": 26853
    },
    {
      "epoch": 0.9193426908592948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0927,
      "step": 26854
    },
    {
      "epoch": 0.9193769257103731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1069,
      "step": 26855
    },
    {
      "epoch": 0.9194111605614516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0915,
      "step": 26856
    },
    {
      "epoch": 0.91944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1065,
      "step": 26857
    },
    {
      "epoch": 0.9194796302636083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0972,
      "step": 26858
    },
    {
      "epoch": 0.9195138651146868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1032,
      "step": 26859
    },
    {
      "epoch": 0.9195480999657651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0858,
      "step": 26860
    },
    {
      "epoch": 0.9195823348168436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1138,
      "step": 26861
    },
    {
      "epoch": 0.9196165696679219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0999,
      "step": 26862
    },
    {
      "epoch": 0.9196508045190004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0872,
      "step": 26863
    },
    {
      "epoch": 0.9196850393700787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 26864
    },
    {
      "epoch": 0.9197192742211572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0928,
      "step": 26865
    },
    {
      "epoch": 0.9197535090722355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0909,
      "step": 26866
    },
    {
      "epoch": 0.919787743923314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0997,
      "step": 26867
    },
    {
      "epoch": 0.9198219787743923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0915,
      "step": 26868
    },
    {
      "epoch": 0.9198562136254708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.097,
      "step": 26869
    },
    {
      "epoch": 0.9198904484765491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1113,
      "step": 26870
    },
    {
      "epoch": 0.9199246833276276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0897,
      "step": 26871
    },
    {
      "epoch": 0.9199589181787059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0978,
      "step": 26872
    },
    {
      "epoch": 0.9199931530297843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1089,
      "step": 26873
    },
    {
      "epoch": 0.9200273878808627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.089,
      "step": 26874
    },
    {
      "epoch": 0.9200616227319411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0996,
      "step": 26875
    },
    {
      "epoch": 0.9200958575830195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0994,
      "step": 26876
    },
    {
      "epoch": 0.9201300924340979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1091,
      "step": 26877
    },
    {
      "epoch": 0.9201643272851763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0941,
      "step": 26878
    },
    {
      "epoch": 0.9201985621362547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0903,
      "step": 26879
    },
    {
      "epoch": 0.9202327969873331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0974,
      "step": 26880
    },
    {
      "epoch": 0.9202670318384115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1034,
      "step": 26881
    },
    {
      "epoch": 0.9203012666894899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0926,
      "step": 26882
    },
    {
      "epoch": 0.9203355015405683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0923,
      "step": 26883
    },
    {
      "epoch": 0.9203697363916467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1072,
      "step": 26884
    },
    {
      "epoch": 0.9204039712427251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0974,
      "step": 26885
    },
    {
      "epoch": 0.9204382060938034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.099,
      "step": 26886
    },
    {
      "epoch": 0.9204724409448819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0913,
      "step": 26887
    },
    {
      "epoch": 0.9205066757959602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0945,
      "step": 26888
    },
    {
      "epoch": 0.9205409106470387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 26889
    },
    {
      "epoch": 0.920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.092,
      "step": 26890
    },
    {
      "epoch": 0.9206093803491955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0961,
      "step": 26891
    },
    {
      "epoch": 0.9206436152002738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1,
      "step": 26892
    },
    {
      "epoch": 0.9206778500513523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1025,
      "step": 26893
    },
    {
      "epoch": 0.9207120849024307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1051,
      "step": 26894
    },
    {
      "epoch": 0.9207463197535091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0991,
      "step": 26895
    },
    {
      "epoch": 0.9207805546045875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1079,
      "step": 26896
    },
    {
      "epoch": 0.9208147894556659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1056,
      "step": 26897
    },
    {
      "epoch": 0.9208490243067443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.099,
      "step": 26898
    },
    {
      "epoch": 0.9208832591578227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1001,
      "step": 26899
    },
    {
      "epoch": 0.9209174940089011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0863,
      "step": 26900
    },
    {
      "epoch": 0.9209517288599794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.107,
      "step": 26901
    },
    {
      "epoch": 0.9209859637110579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.106,
      "step": 26902
    },
    {
      "epoch": 0.9210201985621362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.09,
      "step": 26903
    },
    {
      "epoch": 0.9210544334132147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1048,
      "step": 26904
    },
    {
      "epoch": 0.921088668264293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1019,
      "step": 26905
    },
    {
      "epoch": 0.9211229031153715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0984,
      "step": 26906
    },
    {
      "epoch": 0.9211571379664498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1061,
      "step": 26907
    },
    {
      "epoch": 0.9211913728175283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0972,
      "step": 26908
    },
    {
      "epoch": 0.9212256076686066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1096,
      "step": 26909
    },
    {
      "epoch": 0.9212598425196851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0875,
      "step": 26910
    },
    {
      "epoch": 0.9212940773707634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0885,
      "step": 26911
    },
    {
      "epoch": 0.9213283122218419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0908,
      "step": 26912
    },
    {
      "epoch": 0.9213625470729202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1013,
      "step": 26913
    },
    {
      "epoch": 0.9213967819239987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1129,
      "step": 26914
    },
    {
      "epoch": 0.921431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1014,
      "step": 26915
    },
    {
      "epoch": 0.9214652516261554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0971,
      "step": 26916
    },
    {
      "epoch": 0.9214994864772338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1026,
      "step": 26917
    },
    {
      "epoch": 0.9215337213283122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1059,
      "step": 26918
    },
    {
      "epoch": 0.9215679561793906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0917,
      "step": 26919
    },
    {
      "epoch": 0.921602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0902,
      "step": 26920
    },
    {
      "epoch": 0.9216364258815474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0996,
      "step": 26921
    },
    {
      "epoch": 0.9216706607326258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0918,
      "step": 26922
    },
    {
      "epoch": 0.9217048955837042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1029,
      "step": 26923
    },
    {
      "epoch": 0.9217391304347826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.108,
      "step": 26924
    },
    {
      "epoch": 0.921773365285861,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1022,
      "step": 26925
    },
    {
      "epoch": 0.9218076001369394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0964,
      "step": 26926
    },
    {
      "epoch": 0.9218418349880177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1006,
      "step": 26927
    },
    {
      "epoch": 0.9218760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0888,
      "step": 26928
    },
    {
      "epoch": 0.9219103046901745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1092,
      "step": 26929
    },
    {
      "epoch": 0.921944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.095,
      "step": 26930
    },
    {
      "epoch": 0.9219787743923314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0928,
      "step": 26931
    },
    {
      "epoch": 0.9220130092434098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0975,
      "step": 26932
    },
    {
      "epoch": 0.9220472440944882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0926,
      "step": 26933
    },
    {
      "epoch": 0.9220814789455666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1041,
      "step": 26934
    },
    {
      "epoch": 0.922115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1035,
      "step": 26935
    },
    {
      "epoch": 0.9221499486477234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0961,
      "step": 26936
    },
    {
      "epoch": 0.9221841834988018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0989,
      "step": 26937
    },
    {
      "epoch": 0.9222184183498802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0878,
      "step": 26938
    },
    {
      "epoch": 0.9222526532009586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0707,
      "step": 26939
    },
    {
      "epoch": 0.922286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0968,
      "step": 26940
    },
    {
      "epoch": 0.9223211229031154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.107,
      "step": 26941
    },
    {
      "epoch": 0.9223553577541937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0978,
      "step": 26942
    },
    {
      "epoch": 0.9223895926052722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0921,
      "step": 26943
    },
    {
      "epoch": 0.9224238274563505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0946,
      "step": 26944
    },
    {
      "epoch": 0.922458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1067,
      "step": 26945
    },
    {
      "epoch": 0.9224922971585073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0943,
      "step": 26946
    },
    {
      "epoch": 0.9225265320095858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0923,
      "step": 26947
    },
    {
      "epoch": 0.9225607668606641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1065,
      "step": 26948
    },
    {
      "epoch": 0.9225950017117426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0885,
      "step": 26949
    },
    {
      "epoch": 0.9226292365628209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.097,
      "step": 26950
    },
    {
      "epoch": 0.9226634714138994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.128,
      "step": 26951
    },
    {
      "epoch": 0.9226977062649777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0946,
      "step": 26952
    },
    {
      "epoch": 0.9227319411160562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0844,
      "step": 26953
    },
    {
      "epoch": 0.9227661759671345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0922,
      "step": 26954
    },
    {
      "epoch": 0.922800410818213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1005,
      "step": 26955
    },
    {
      "epoch": 0.9228346456692913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0854,
      "step": 26956
    },
    {
      "epoch": 0.9228688805203697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1072,
      "step": 26957
    },
    {
      "epoch": 0.9229031153714481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.097,
      "step": 26958
    },
    {
      "epoch": 0.9229373502225265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0878,
      "step": 26959
    },
    {
      "epoch": 0.9229715850736049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0981,
      "step": 26960
    },
    {
      "epoch": 0.9230058199246833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0896,
      "step": 26961
    },
    {
      "epoch": 0.9230400547757617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0965,
      "step": 26962
    },
    {
      "epoch": 0.9230742896268401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0949,
      "step": 26963
    },
    {
      "epoch": 0.9231085244779185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.092,
      "step": 26964
    },
    {
      "epoch": 0.9231427593289969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0679,
      "step": 26965
    },
    {
      "epoch": 0.9231769941800754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0944,
      "step": 26966
    },
    {
      "epoch": 0.9232112290311537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 26967
    },
    {
      "epoch": 0.9232454638822322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0904,
      "step": 26968
    },
    {
      "epoch": 0.9232796987333105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1031,
      "step": 26969
    },
    {
      "epoch": 0.923313933584389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1031,
      "step": 26970
    },
    {
      "epoch": 0.9233481684354673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0871,
      "step": 26971
    },
    {
      "epoch": 0.9233824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0845,
      "step": 26972
    },
    {
      "epoch": 0.9234166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1192,
      "step": 26973
    },
    {
      "epoch": 0.9234508729887025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0911,
      "step": 26974
    },
    {
      "epoch": 0.9234851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1054,
      "step": 26975
    },
    {
      "epoch": 0.9235193426908593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1017,
      "step": 26976
    },
    {
      "epoch": 0.9235535775419377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0942,
      "step": 26977
    },
    {
      "epoch": 0.9235878123930161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0787,
      "step": 26978
    },
    {
      "epoch": 0.9236220472440945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1007,
      "step": 26979
    },
    {
      "epoch": 0.9236562820951729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0822,
      "step": 26980
    },
    {
      "epoch": 0.9236905169462513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.115,
      "step": 26981
    },
    {
      "epoch": 0.9237247517973297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0997,
      "step": 26982
    },
    {
      "epoch": 0.923758986648408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0861,
      "step": 26983
    },
    {
      "epoch": 0.9237932214994865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0961,
      "step": 26984
    },
    {
      "epoch": 0.9238274563505648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1047,
      "step": 26985
    },
    {
      "epoch": 0.9238616912016433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0897,
      "step": 26986
    },
    {
      "epoch": 0.9238959260527216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0898,
      "step": 26987
    },
    {
      "epoch": 0.9239301609038001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0899,
      "step": 26988
    },
    {
      "epoch": 0.9239643957548784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0954,
      "step": 26989
    },
    {
      "epoch": 0.9239986306059569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1015,
      "step": 26990
    },
    {
      "epoch": 0.9240328654570352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1035,
      "step": 26991
    },
    {
      "epoch": 0.9240671003081137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0957,
      "step": 26992
    },
    {
      "epoch": 0.924101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1026,
      "step": 26993
    },
    {
      "epoch": 0.9241355700102705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.12,
      "step": 26994
    },
    {
      "epoch": 0.9241698048613488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1045,
      "step": 26995
    },
    {
      "epoch": 0.9242040397124273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.099,
      "step": 26996
    },
    {
      "epoch": 0.9242382745635056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0831,
      "step": 26997
    },
    {
      "epoch": 0.924272509414584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1043,
      "step": 26998
    },
    {
      "epoch": 0.9243067442656624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0819,
      "step": 26999
    },
    {
      "epoch": 0.9243409791167408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0908,
      "step": 27000
    },
    {
      "epoch": 0.9243752139678192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0884,
      "step": 27001
    },
    {
      "epoch": 0.9244094488188976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1168,
      "step": 27002
    },
    {
      "epoch": 0.9244436836699761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1033,
      "step": 27003
    },
    {
      "epoch": 0.9244779185210544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1162,
      "step": 27004
    },
    {
      "epoch": 0.9245121533721329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1146,
      "step": 27005
    },
    {
      "epoch": 0.9245463882232112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1024,
      "step": 27006
    },
    {
      "epoch": 0.9245806230742897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1027,
      "step": 27007
    },
    {
      "epoch": 0.924614857925368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0864,
      "step": 27008
    },
    {
      "epoch": 0.9246490927764465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0965,
      "step": 27009
    },
    {
      "epoch": 0.9246833276275248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 27010
    },
    {
      "epoch": 0.9247175624786033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0965,
      "step": 27011
    },
    {
      "epoch": 0.9247517973296816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1039,
      "step": 27012
    },
    {
      "epoch": 0.92478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0779,
      "step": 27013
    },
    {
      "epoch": 0.9248202670318384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0946,
      "step": 27014
    },
    {
      "epoch": 0.9248545018829168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1079,
      "step": 27015
    },
    {
      "epoch": 0.9248887367339952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0975,
      "step": 27016
    },
    {
      "epoch": 0.9249229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1023,
      "step": 27017
    },
    {
      "epoch": 0.924957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0864,
      "step": 27018
    },
    {
      "epoch": 0.9249914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0911,
      "step": 27019
    },
    {
      "epoch": 0.9250256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.085,
      "step": 27020
    },
    {
      "epoch": 0.9250599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0973,
      "step": 27021
    },
    {
      "epoch": 0.9250941458404656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0969,
      "step": 27022
    },
    {
      "epoch": 0.925128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0763,
      "step": 27023
    },
    {
      "epoch": 0.9251626155426224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1006,
      "step": 27024
    },
    {
      "epoch": 0.9251968503937008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 27025
    },
    {
      "epoch": 0.9252310852447792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0868,
      "step": 27026
    },
    {
      "epoch": 0.9252653200958576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0832,
      "step": 27027
    },
    {
      "epoch": 0.9252995549469359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0986,
      "step": 27028
    },
    {
      "epoch": 0.9253337897980144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 27029
    },
    {
      "epoch": 0.9253680246490927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0922,
      "step": 27030
    },
    {
      "epoch": 0.9254022595001712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1122,
      "step": 27031
    },
    {
      "epoch": 0.9254364943512495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1005,
      "step": 27032
    },
    {
      "epoch": 0.925470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0923,
      "step": 27033
    },
    {
      "epoch": 0.9255049640534063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1142,
      "step": 27034
    },
    {
      "epoch": 0.9255391989044848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0868,
      "step": 27035
    },
    {
      "epoch": 0.9255734337555631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1008,
      "step": 27036
    },
    {
      "epoch": 0.9256076686066416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.101,
      "step": 27037
    },
    {
      "epoch": 0.9256419034577199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 27038
    },
    {
      "epoch": 0.9256761383087984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0878,
      "step": 27039
    },
    {
      "epoch": 0.9257103731598768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0919,
      "step": 27040
    },
    {
      "epoch": 0.9257446080109552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0939,
      "step": 27041
    },
    {
      "epoch": 0.9257788428620336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0769,
      "step": 27042
    },
    {
      "epoch": 0.9258130777131119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0836,
      "step": 27043
    },
    {
      "epoch": 0.9258473125641904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1027,
      "step": 27044
    },
    {
      "epoch": 0.9258815474152687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1048,
      "step": 27045
    },
    {
      "epoch": 0.9259157822663472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1129,
      "step": 27046
    },
    {
      "epoch": 0.9259500171174255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1032,
      "step": 27047
    },
    {
      "epoch": 0.925984251968504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0869,
      "step": 27048
    },
    {
      "epoch": 0.9260184868195823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1106,
      "step": 27049
    },
    {
      "epoch": 0.9260527216706608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1071,
      "step": 27050
    },
    {
      "epoch": 0.9260869565217391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0972,
      "step": 27051
    },
    {
      "epoch": 0.9261211913728176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1108,
      "step": 27052
    },
    {
      "epoch": 0.9261554262238959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0911,
      "step": 27053
    },
    {
      "epoch": 0.9261896610749744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0926,
      "step": 27054
    },
    {
      "epoch": 0.9262238959260527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.091,
      "step": 27055
    },
    {
      "epoch": 0.9262581307771312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0823,
      "step": 27056
    },
    {
      "epoch": 0.9262923656282095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0942,
      "step": 27057
    },
    {
      "epoch": 0.9263266004792879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1011,
      "step": 27058
    },
    {
      "epoch": 0.9263608353303663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.112,
      "step": 27059
    },
    {
      "epoch": 0.9263950701814447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1291,
      "step": 27060
    },
    {
      "epoch": 0.9264293050325231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 27061
    },
    {
      "epoch": 0.9264635398836015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1002,
      "step": 27062
    },
    {
      "epoch": 0.9264977747346799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0847,
      "step": 27063
    },
    {
      "epoch": 0.9265320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1093,
      "step": 27064
    },
    {
      "epoch": 0.9265662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0838,
      "step": 27065
    },
    {
      "epoch": 0.9266004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0911,
      "step": 27066
    },
    {
      "epoch": 0.9266347141389935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1034,
      "step": 27067
    },
    {
      "epoch": 0.9266689489900719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1196,
      "step": 27068
    },
    {
      "epoch": 0.9267031838411502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0812,
      "step": 27069
    },
    {
      "epoch": 0.9267374186922287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0845,
      "step": 27070
    },
    {
      "epoch": 0.926771653543307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1059,
      "step": 27071
    },
    {
      "epoch": 0.9268058883943855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1001,
      "step": 27072
    },
    {
      "epoch": 0.9268401232454638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1004,
      "step": 27073
    },
    {
      "epoch": 0.9268743580965423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0972,
      "step": 27074
    },
    {
      "epoch": 0.9269085929476206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1045,
      "step": 27075
    },
    {
      "epoch": 0.9269428277986991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0818,
      "step": 27076
    },
    {
      "epoch": 0.9269770626497775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1092,
      "step": 27077
    },
    {
      "epoch": 0.9270112975008559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1111,
      "step": 27078
    },
    {
      "epoch": 0.9270455323519343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0841,
      "step": 27079
    },
    {
      "epoch": 0.9270797672030127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1046,
      "step": 27080
    },
    {
      "epoch": 0.9271140020540911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0941,
      "step": 27081
    },
    {
      "epoch": 0.9271482369051695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1022,
      "step": 27082
    },
    {
      "epoch": 0.9271824717562479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.106,
      "step": 27083
    },
    {
      "epoch": 0.9272167066073262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1013,
      "step": 27084
    },
    {
      "epoch": 0.9272509414584047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0862,
      "step": 27085
    },
    {
      "epoch": 0.927285176309483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.087,
      "step": 27086
    },
    {
      "epoch": 0.9273194111605615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.094,
      "step": 27087
    },
    {
      "epoch": 0.9273536460116398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0968,
      "step": 27088
    },
    {
      "epoch": 0.9273878808627183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1039,
      "step": 27089
    },
    {
      "epoch": 0.9274221157137966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1023,
      "step": 27090
    },
    {
      "epoch": 0.9274563505648751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 27091
    },
    {
      "epoch": 0.9274905854159534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0926,
      "step": 27092
    },
    {
      "epoch": 0.9275248202670319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1084,
      "step": 27093
    },
    {
      "epoch": 0.9275590551181102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0898,
      "step": 27094
    },
    {
      "epoch": 0.9275932899691887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0911,
      "step": 27095
    },
    {
      "epoch": 0.927627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0988,
      "step": 27096
    },
    {
      "epoch": 0.9276617596713455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0913,
      "step": 27097
    },
    {
      "epoch": 0.9276959945224238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0927,
      "step": 27098
    },
    {
      "epoch": 0.9277302293735022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0994,
      "step": 27099
    },
    {
      "epoch": 0.9277644642245806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0963,
      "step": 27100
    },
    {
      "epoch": 0.927798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1028,
      "step": 27101
    },
    {
      "epoch": 0.9278329339267374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.091,
      "step": 27102
    },
    {
      "epoch": 0.9278671687778158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0936,
      "step": 27103
    },
    {
      "epoch": 0.9279014036288942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1045,
      "step": 27104
    },
    {
      "epoch": 0.9279356384799726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0994,
      "step": 27105
    },
    {
      "epoch": 0.927969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0847,
      "step": 27106
    },
    {
      "epoch": 0.9280041081821294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1089,
      "step": 27107
    },
    {
      "epoch": 0.9280383430332078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 27108
    },
    {
      "epoch": 0.9280725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0891,
      "step": 27109
    },
    {
      "epoch": 0.9281068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0939,
      "step": 27110
    },
    {
      "epoch": 0.928141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 27111
    },
    {
      "epoch": 0.9281752824375213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0965,
      "step": 27112
    },
    {
      "epoch": 0.9282095172885998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0935,
      "step": 27113
    },
    {
      "epoch": 0.9282437521396782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1019,
      "step": 27114
    },
    {
      "epoch": 0.9282779869907566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0703,
      "step": 27115
    },
    {
      "epoch": 0.928312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 27116
    },
    {
      "epoch": 0.9283464566929134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1084,
      "step": 27117
    },
    {
      "epoch": 0.9283806915439918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0965,
      "step": 27118
    },
    {
      "epoch": 0.9284149263950702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0862,
      "step": 27119
    },
    {
      "epoch": 0.9284491612461486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1016,
      "step": 27120
    },
    {
      "epoch": 0.928483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0995,
      "step": 27121
    },
    {
      "epoch": 0.9285176309483054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0882,
      "step": 27122
    },
    {
      "epoch": 0.9285518657993838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0829,
      "step": 27123
    },
    {
      "epoch": 0.9285861006504622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0949,
      "step": 27124
    },
    {
      "epoch": 0.9286203355015406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0908,
      "step": 27125
    },
    {
      "epoch": 0.928654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0854,
      "step": 27126
    },
    {
      "epoch": 0.9286888052036973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.082,
      "step": 27127
    },
    {
      "epoch": 0.9287230400547758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1023,
      "step": 27128
    },
    {
      "epoch": 0.9287572749058541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0956,
      "step": 27129
    },
    {
      "epoch": 0.9287915097569326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.113,
      "step": 27130
    },
    {
      "epoch": 0.9288257446080109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1026,
      "step": 27131
    },
    {
      "epoch": 0.9288599794590894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0915,
      "step": 27132
    },
    {
      "epoch": 0.9288942143101677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0873,
      "step": 27133
    },
    {
      "epoch": 0.9289284491612462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0934,
      "step": 27134
    },
    {
      "epoch": 0.9289626840123245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1028,
      "step": 27135
    },
    {
      "epoch": 0.928996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0808,
      "step": 27136
    },
    {
      "epoch": 0.9290311537144813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 27137
    },
    {
      "epoch": 0.9290653885655598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0963,
      "step": 27138
    },
    {
      "epoch": 0.9290996234166381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0917,
      "step": 27139
    },
    {
      "epoch": 0.9291338582677166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 27140
    },
    {
      "epoch": 0.9291680931187949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0833,
      "step": 27141
    },
    {
      "epoch": 0.9292023279698733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0916,
      "step": 27142
    },
    {
      "epoch": 0.9292365628209517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0977,
      "step": 27143
    },
    {
      "epoch": 0.9292707976720301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1088,
      "step": 27144
    },
    {
      "epoch": 0.9293050325231085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.071,
      "step": 27145
    },
    {
      "epoch": 0.9293392673741869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0969,
      "step": 27146
    },
    {
      "epoch": 0.9293735022252653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.095,
      "step": 27147
    },
    {
      "epoch": 0.9294077370763437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0835,
      "step": 27148
    },
    {
      "epoch": 0.9294419719274221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0933,
      "step": 27149
    },
    {
      "epoch": 0.9294762067785005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1085,
      "step": 27150
    },
    {
      "epoch": 0.929510441629579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0971,
      "step": 27151
    },
    {
      "epoch": 0.9295446764806573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1006,
      "step": 27152
    },
    {
      "epoch": 0.9295789113317358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1142,
      "step": 27153
    },
    {
      "epoch": 0.9296131461828141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0886,
      "step": 27154
    },
    {
      "epoch": 0.9296473810338926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1064,
      "step": 27155
    },
    {
      "epoch": 0.9296816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0952,
      "step": 27156
    },
    {
      "epoch": 0.9297158507360493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0839,
      "step": 27157
    },
    {
      "epoch": 0.9297500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1092,
      "step": 27158
    },
    {
      "epoch": 0.9297843204382061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.109,
      "step": 27159
    },
    {
      "epoch": 0.9298185552892845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0922,
      "step": 27160
    },
    {
      "epoch": 0.9298527901403629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0894,
      "step": 27161
    },
    {
      "epoch": 0.9298870249914413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0891,
      "step": 27162
    },
    {
      "epoch": 0.9299212598425197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 27163
    },
    {
      "epoch": 0.9299554946935981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0885,
      "step": 27164
    },
    {
      "epoch": 0.9299897295446765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0942,
      "step": 27165
    },
    {
      "epoch": 0.9300239643957549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.083,
      "step": 27166
    },
    {
      "epoch": 0.9300581992468333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0979,
      "step": 27167
    },
    {
      "epoch": 0.9300924340979116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.099,
      "step": 27168
    },
    {
      "epoch": 0.9301266689489901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 27169
    },
    {
      "epoch": 0.9301609038000684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1086,
      "step": 27170
    },
    {
      "epoch": 0.9301951386511469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1037,
      "step": 27171
    },
    {
      "epoch": 0.9302293735022252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0779,
      "step": 27172
    },
    {
      "epoch": 0.9302636083533037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1002,
      "step": 27173
    },
    {
      "epoch": 0.930297843204382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0927,
      "step": 27174
    },
    {
      "epoch": 0.9303320780554605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0922,
      "step": 27175
    },
    {
      "epoch": 0.9303663129065388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0838,
      "step": 27176
    },
    {
      "epoch": 0.9304005477576173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0955,
      "step": 27177
    },
    {
      "epoch": 0.9304347826086956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0965,
      "step": 27178
    },
    {
      "epoch": 0.9304690174597741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0959,
      "step": 27179
    },
    {
      "epoch": 0.9305032523108524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0808,
      "step": 27180
    },
    {
      "epoch": 0.9305374871619309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0816,
      "step": 27181
    },
    {
      "epoch": 0.9305717220130092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1018,
      "step": 27182
    },
    {
      "epoch": 0.9306059568640876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0819,
      "step": 27183
    },
    {
      "epoch": 0.930640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 27184
    },
    {
      "epoch": 0.9306744265662444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1063,
      "step": 27185
    },
    {
      "epoch": 0.9307086614173228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.102,
      "step": 27186
    },
    {
      "epoch": 0.9307428962684012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0821,
      "step": 27187
    },
    {
      "epoch": 0.9307771311194797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0896,
      "step": 27188
    },
    {
      "epoch": 0.930811365970558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0976,
      "step": 27189
    },
    {
      "epoch": 0.9308456008216365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0868,
      "step": 27190
    },
    {
      "epoch": 0.9308798356727148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0954,
      "step": 27191
    },
    {
      "epoch": 0.9309140705237933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.086,
      "step": 27192
    },
    {
      "epoch": 0.9309483053748716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0967,
      "step": 27193
    },
    {
      "epoch": 0.9309825402259501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1034,
      "step": 27194
    },
    {
      "epoch": 0.9310167750770284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1003,
      "step": 27195
    },
    {
      "epoch": 0.9310510099281069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 27196
    },
    {
      "epoch": 0.9310852447791852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1086,
      "step": 27197
    },
    {
      "epoch": 0.9311194796302636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1138,
      "step": 27198
    },
    {
      "epoch": 0.931153714481342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1101,
      "step": 27199
    },
    {
      "epoch": 0.9311879493324204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1056,
      "step": 27200
    },
    {
      "epoch": 0.9312221841834988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.083,
      "step": 27201
    },
    {
      "epoch": 0.9312564190345772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0899,
      "step": 27202
    },
    {
      "epoch": 0.9312906538856556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 27203
    },
    {
      "epoch": 0.931324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0801,
      "step": 27204
    },
    {
      "epoch": 0.9313591235878124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0847,
      "step": 27205
    },
    {
      "epoch": 0.9313933584388908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0861,
      "step": 27206
    },
    {
      "epoch": 0.9314275932899692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0966,
      "step": 27207
    },
    {
      "epoch": 0.9314618281410476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0978,
      "step": 27208
    },
    {
      "epoch": 0.931496062992126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0972,
      "step": 27209
    },
    {
      "epoch": 0.9315302978432044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0927,
      "step": 27210
    },
    {
      "epoch": 0.9315645326942827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0875,
      "step": 27211
    },
    {
      "epoch": 0.9315987675453612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0795,
      "step": 27212
    },
    {
      "epoch": 0.9316330023964395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0841,
      "step": 27213
    },
    {
      "epoch": 0.931667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0973,
      "step": 27214
    },
    {
      "epoch": 0.9317014720985963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0923,
      "step": 27215
    },
    {
      "epoch": 0.9317357069496748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1024,
      "step": 27216
    },
    {
      "epoch": 0.9317699418007531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0957,
      "step": 27217
    },
    {
      "epoch": 0.9318041766518316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0884,
      "step": 27218
    },
    {
      "epoch": 0.9318384115029099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0985,
      "step": 27219
    },
    {
      "epoch": 0.9318726463539884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1053,
      "step": 27220
    },
    {
      "epoch": 0.9319068812050667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1098,
      "step": 27221
    },
    {
      "epoch": 0.9319411160561452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0927,
      "step": 27222
    },
    {
      "epoch": 0.9319753509072235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0835,
      "step": 27223
    },
    {
      "epoch": 0.932009585758302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0924,
      "step": 27224
    },
    {
      "epoch": 0.9320438206093804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0993,
      "step": 27225
    },
    {
      "epoch": 0.9320780554604587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0853,
      "step": 27226
    },
    {
      "epoch": 0.9321122903115372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 27227
    },
    {
      "epoch": 0.9321465251626155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0924,
      "step": 27228
    },
    {
      "epoch": 0.932180760013694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0806,
      "step": 27229
    },
    {
      "epoch": 0.9322149948647723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0946,
      "step": 27230
    },
    {
      "epoch": 0.9322492297158508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0858,
      "step": 27231
    },
    {
      "epoch": 0.9322834645669291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1026,
      "step": 27232
    },
    {
      "epoch": 0.9323176994180076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0942,
      "step": 27233
    },
    {
      "epoch": 0.9323519342690859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.096,
      "step": 27234
    },
    {
      "epoch": 0.9323861691201644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.099,
      "step": 27235
    },
    {
      "epoch": 0.9324204039712427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1026,
      "step": 27236
    },
    {
      "epoch": 0.9324546388223212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0903,
      "step": 27237
    },
    {
      "epoch": 0.9324888736733995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.077,
      "step": 27238
    },
    {
      "epoch": 0.932523108524478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0926,
      "step": 27239
    },
    {
      "epoch": 0.9325573433755563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1028,
      "step": 27240
    },
    {
      "epoch": 0.9325915782266347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0855,
      "step": 27241
    },
    {
      "epoch": 0.9326258130777131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1068,
      "step": 27242
    },
    {
      "epoch": 0.9326600479287915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0966,
      "step": 27243
    },
    {
      "epoch": 0.9326942827798699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0787,
      "step": 27244
    },
    {
      "epoch": 0.9327285176309483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1051,
      "step": 27245
    },
    {
      "epoch": 0.9327627524820267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0814,
      "step": 27246
    },
    {
      "epoch": 0.9327969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.08,
      "step": 27247
    },
    {
      "epoch": 0.9328312221841835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0924,
      "step": 27248
    },
    {
      "epoch": 0.9328654570352619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0939,
      "step": 27249
    },
    {
      "epoch": 0.9328996918863403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0941,
      "step": 27250
    },
    {
      "epoch": 0.9329339267374187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0932,
      "step": 27251
    },
    {
      "epoch": 0.932968161588497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0885,
      "step": 27252
    },
    {
      "epoch": 0.9330023964395755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0994,
      "step": 27253
    },
    {
      "epoch": 0.9330366312906538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1085,
      "step": 27254
    },
    {
      "epoch": 0.9330708661417323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0976,
      "step": 27255
    },
    {
      "epoch": 0.9331051009928106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0954,
      "step": 27256
    },
    {
      "epoch": 0.9331393358438891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1005,
      "step": 27257
    },
    {
      "epoch": 0.9331735706949674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.085,
      "step": 27258
    },
    {
      "epoch": 0.9332078055460459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0975,
      "step": 27259
    },
    {
      "epoch": 0.9332420403971242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0871,
      "step": 27260
    },
    {
      "epoch": 0.9332762752482027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0837,
      "step": 27261
    },
    {
      "epoch": 0.9333105100992811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0924,
      "step": 27262
    },
    {
      "epoch": 0.9333447449503595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0981,
      "step": 27263
    },
    {
      "epoch": 0.9333789798014379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0917,
      "step": 27264
    },
    {
      "epoch": 0.9334132146525163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0845,
      "step": 27265
    },
    {
      "epoch": 0.9334474495035947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1043,
      "step": 27266
    },
    {
      "epoch": 0.933481684354673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1027,
      "step": 27267
    },
    {
      "epoch": 0.9335159192057515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1042,
      "step": 27268
    },
    {
      "epoch": 0.9335501540568298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1139,
      "step": 27269
    },
    {
      "epoch": 0.9335843889079083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0958,
      "step": 27270
    },
    {
      "epoch": 0.9336186237589866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.076,
      "step": 27271
    },
    {
      "epoch": 0.9336528586100651,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.086,
      "step": 27272
    },
    {
      "epoch": 0.9336870934611434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0907,
      "step": 27273
    },
    {
      "epoch": 0.9337213283122219,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0914,
      "step": 27274
    },
    {
      "epoch": 0.9337555631633002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0875,
      "step": 27275
    },
    {
      "epoch": 0.9337897980143787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.086,
      "step": 27276
    },
    {
      "epoch": 0.933824032865457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0896,
      "step": 27277
    },
    {
      "epoch": 0.9338582677165355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 27278
    },
    {
      "epoch": 0.9338925025676138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 27279
    },
    {
      "epoch": 0.9339267374186923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0841,
      "step": 27280
    },
    {
      "epoch": 0.9339609722697706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0895,
      "step": 27281
    },
    {
      "epoch": 0.933995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.087,
      "step": 27282
    },
    {
      "epoch": 0.9340294419719274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.095,
      "step": 27283
    },
    {
      "epoch": 0.9340636768230058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1085,
      "step": 27284
    },
    {
      "epoch": 0.9340979116740842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0881,
      "step": 27285
    },
    {
      "epoch": 0.9341321465251626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0945,
      "step": 27286
    },
    {
      "epoch": 0.934166381376241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1096,
      "step": 27287
    },
    {
      "epoch": 0.9342006162273194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0901,
      "step": 27288
    },
    {
      "epoch": 0.9342348510783978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0953,
      "step": 27289
    },
    {
      "epoch": 0.9342690859294762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0831,
      "step": 27290
    },
    {
      "epoch": 0.9343033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0982,
      "step": 27291
    },
    {
      "epoch": 0.934337555631633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0953,
      "step": 27292
    },
    {
      "epoch": 0.9343717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0918,
      "step": 27293
    },
    {
      "epoch": 0.9344060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0899,
      "step": 27294
    },
    {
      "epoch": 0.9344402601848681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0788,
      "step": 27295
    },
    {
      "epoch": 0.9344744950359466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0975,
      "step": 27296
    },
    {
      "epoch": 0.9345087298870249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 27297
    },
    {
      "epoch": 0.9345429647381034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0862,
      "step": 27298
    },
    {
      "epoch": 0.9345771995891818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.099,
      "step": 27299
    },
    {
      "epoch": 0.9346114344402602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0963,
      "step": 27300
    },
    {
      "epoch": 0.9346456692913386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1138,
      "step": 27301
    },
    {
      "epoch": 0.934679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0983,
      "step": 27302
    },
    {
      "epoch": 0.9347141389934954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0807,
      "step": 27303
    },
    {
      "epoch": 0.9347483738445738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 27304
    },
    {
      "epoch": 0.9347826086956522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0975,
      "step": 27305
    },
    {
      "epoch": 0.9348168435467306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 27306
    },
    {
      "epoch": 0.934851078397809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0947,
      "step": 27307
    },
    {
      "epoch": 0.9348853132488874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0809,
      "step": 27308
    },
    {
      "epoch": 0.9349195480999658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1021,
      "step": 27309
    },
    {
      "epoch": 0.9349537829510441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0659,
      "step": 27310
    },
    {
      "epoch": 0.9349880178021226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1142,
      "step": 27311
    },
    {
      "epoch": 0.9350222526532009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 27312
    },
    {
      "epoch": 0.9350564875042794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0853,
      "step": 27313
    },
    {
      "epoch": 0.9350907223553577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.091,
      "step": 27314
    },
    {
      "epoch": 0.9351249572064362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0871,
      "step": 27315
    },
    {
      "epoch": 0.9351591920575145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1046,
      "step": 27316
    },
    {
      "epoch": 0.935193426908593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0825,
      "step": 27317
    },
    {
      "epoch": 0.9352276617596713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0866,
      "step": 27318
    },
    {
      "epoch": 0.9352618966107498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0975,
      "step": 27319
    },
    {
      "epoch": 0.9352961314618281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0858,
      "step": 27320
    },
    {
      "epoch": 0.9353303663129066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0971,
      "step": 27321
    },
    {
      "epoch": 0.9353646011639849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1042,
      "step": 27322
    },
    {
      "epoch": 0.9353988360150634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0898,
      "step": 27323
    },
    {
      "epoch": 0.9354330708661417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1094,
      "step": 27324
    },
    {
      "epoch": 0.9354673057172201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0823,
      "step": 27325
    },
    {
      "epoch": 0.9355015405682985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0872,
      "step": 27326
    },
    {
      "epoch": 0.9355357754193769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0791,
      "step": 27327
    },
    {
      "epoch": 0.9355700102704553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0845,
      "step": 27328
    },
    {
      "epoch": 0.9356042451215337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.103,
      "step": 27329
    },
    {
      "epoch": 0.9356384799726121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0822,
      "step": 27330
    },
    {
      "epoch": 0.9356727148236905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0957,
      "step": 27331
    },
    {
      "epoch": 0.9357069496747689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0864,
      "step": 27332
    },
    {
      "epoch": 0.9357411845258473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0901,
      "step": 27333
    },
    {
      "epoch": 0.9357754193769258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0925,
      "step": 27334
    },
    {
      "epoch": 0.9358096542280041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0921,
      "step": 27335
    },
    {
      "epoch": 0.9358438890790826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0827,
      "step": 27336
    },
    {
      "epoch": 0.9358781239301609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0802,
      "step": 27337
    },
    {
      "epoch": 0.9359123587812394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0786,
      "step": 27338
    },
    {
      "epoch": 0.9359465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1013,
      "step": 27339
    },
    {
      "epoch": 0.9359808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0866,
      "step": 27340
    },
    {
      "epoch": 0.9360150633344745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0929,
      "step": 27341
    },
    {
      "epoch": 0.9360492981855529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0934,
      "step": 27342
    },
    {
      "epoch": 0.9360835330366313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.088,
      "step": 27343
    },
    {
      "epoch": 0.9361177678877097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0966,
      "step": 27344
    },
    {
      "epoch": 0.9361520027387881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0983,
      "step": 27345
    },
    {
      "epoch": 0.9361862375898665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0911,
      "step": 27346
    },
    {
      "epoch": 0.9362204724409449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1018,
      "step": 27347
    },
    {
      "epoch": 0.9362547072920233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0858,
      "step": 27348
    },
    {
      "epoch": 0.9362889421431017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0945,
      "step": 27349
    },
    {
      "epoch": 0.9363231769941801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0947,
      "step": 27350
    },
    {
      "epoch": 0.9363574118452584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0945,
      "step": 27351
    },
    {
      "epoch": 0.9363916466963369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0885,
      "step": 27352
    },
    {
      "epoch": 0.9364258815474152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1032,
      "step": 27353
    },
    {
      "epoch": 0.9364601163984937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0893,
      "step": 27354
    },
    {
      "epoch": 0.936494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.082,
      "step": 27355
    },
    {
      "epoch": 0.9365285861006505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1027,
      "step": 27356
    },
    {
      "epoch": 0.9365628209517288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1169,
      "step": 27357
    },
    {
      "epoch": 0.9365970558028073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1024,
      "step": 27358
    },
    {
      "epoch": 0.9366312906538856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1,
      "step": 27359
    },
    {
      "epoch": 0.9366655255049641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0911,
      "step": 27360
    },
    {
      "epoch": 0.9366997603560424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0897,
      "step": 27361
    },
    {
      "epoch": 0.9367339952071209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0941,
      "step": 27362
    },
    {
      "epoch": 0.9367682300581992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0923,
      "step": 27363
    },
    {
      "epoch": 0.9368024649092777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0901,
      "step": 27364
    },
    {
      "epoch": 0.936836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0998,
      "step": 27365
    },
    {
      "epoch": 0.9368709346114344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0909,
      "step": 27366
    },
    {
      "epoch": 0.9369051694625128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1031,
      "step": 27367
    },
    {
      "epoch": 0.9369394043135912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1128,
      "step": 27368
    },
    {
      "epoch": 0.9369736391646696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.091,
      "step": 27369
    },
    {
      "epoch": 0.937007874015748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0893,
      "step": 27370
    },
    {
      "epoch": 0.9370421088668265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0825,
      "step": 27371
    },
    {
      "epoch": 0.9370763437179048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0903,
      "step": 27372
    },
    {
      "epoch": 0.9371105785689833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0851,
      "step": 27373
    },
    {
      "epoch": 0.9371448134200616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0922,
      "step": 27374
    },
    {
      "epoch": 0.9371790482711401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0792,
      "step": 27375
    },
    {
      "epoch": 0.9372132831222184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0995,
      "step": 27376
    },
    {
      "epoch": 0.9372475179732969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1117,
      "step": 27377
    },
    {
      "epoch": 0.9372817528243752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0868,
      "step": 27378
    },
    {
      "epoch": 0.9373159876754537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0828,
      "step": 27379
    },
    {
      "epoch": 0.937350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0955,
      "step": 27380
    },
    {
      "epoch": 0.9373844573776104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0869,
      "step": 27381
    },
    {
      "epoch": 0.9374186922286888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1013,
      "step": 27382
    },
    {
      "epoch": 0.9374529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1057,
      "step": 27383
    },
    {
      "epoch": 0.9374871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.073,
      "step": 27384
    },
    {
      "epoch": 0.937521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.095,
      "step": 27385
    },
    {
      "epoch": 0.9375556316330024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0829,
      "step": 27386
    },
    {
      "epoch": 0.9375898664840808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0972,
      "step": 27387
    },
    {
      "epoch": 0.9376241013351592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0989,
      "step": 27388
    },
    {
      "epoch": 0.9376583361862376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1038,
      "step": 27389
    },
    {
      "epoch": 0.937692571037316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0823,
      "step": 27390
    },
    {
      "epoch": 0.9377268058883944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1118,
      "step": 27391
    },
    {
      "epoch": 0.9377610407394728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0869,
      "step": 27392
    },
    {
      "epoch": 0.9377952755905512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 27393
    },
    {
      "epoch": 0.9378295104416295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0931,
      "step": 27394
    },
    {
      "epoch": 0.937863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0954,
      "step": 27395
    },
    {
      "epoch": 0.9378979801437863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0858,
      "step": 27396
    },
    {
      "epoch": 0.9379322149948648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0866,
      "step": 27397
    },
    {
      "epoch": 0.9379664498459431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0859,
      "step": 27398
    },
    {
      "epoch": 0.9380006846970216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0758,
      "step": 27399
    },
    {
      "epoch": 0.9380349195480999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0977,
      "step": 27400
    },
    {
      "epoch": 0.9380691543991784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0944,
      "step": 27401
    },
    {
      "epoch": 0.9381033892502567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0816,
      "step": 27402
    },
    {
      "epoch": 0.9381376241013352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1036,
      "step": 27403
    },
    {
      "epoch": 0.9381718589524135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0958,
      "step": 27404
    },
    {
      "epoch": 0.938206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1065,
      "step": 27405
    },
    {
      "epoch": 0.9382403286545703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0923,
      "step": 27406
    },
    {
      "epoch": 0.9382745635056488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1091,
      "step": 27407
    },
    {
      "epoch": 0.9383087983567272,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0771,
      "step": 27408
    },
    {
      "epoch": 0.9383430332078055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0849,
      "step": 27409
    },
    {
      "epoch": 0.938377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 27410
    },
    {
      "epoch": 0.9384115029099623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0837,
      "step": 27411
    },
    {
      "epoch": 0.9384457377610408,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0812,
      "step": 27412
    },
    {
      "epoch": 0.9384799726121191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 27413
    },
    {
      "epoch": 0.9385142074631976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0923,
      "step": 27414
    },
    {
      "epoch": 0.9385484423142759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0865,
      "step": 27415
    },
    {
      "epoch": 0.9385826771653544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1078,
      "step": 27416
    },
    {
      "epoch": 0.9386169120164327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0794,
      "step": 27417
    },
    {
      "epoch": 0.9386511468675112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0988,
      "step": 27418
    },
    {
      "epoch": 0.9386853817185895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0847,
      "step": 27419
    },
    {
      "epoch": 0.938719616569668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0967,
      "step": 27420
    },
    {
      "epoch": 0.9387538514207463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1001,
      "step": 27421
    },
    {
      "epoch": 0.9387880862718248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0992,
      "step": 27422
    },
    {
      "epoch": 0.9388223211229031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.094,
      "step": 27423
    },
    {
      "epoch": 0.9388565559739815,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.101,
      "step": 27424
    },
    {
      "epoch": 0.9388907908250599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.081,
      "step": 27425
    },
    {
      "epoch": 0.9389250256761383,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0853,
      "step": 27426
    },
    {
      "epoch": 0.9389592605272167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0913,
      "step": 27427
    },
    {
      "epoch": 0.9389934953782951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0967,
      "step": 27428
    },
    {
      "epoch": 0.9390277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1092,
      "step": 27429
    },
    {
      "epoch": 0.9390619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1128,
      "step": 27430
    },
    {
      "epoch": 0.9390961999315303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0951,
      "step": 27431
    },
    {
      "epoch": 0.9391304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0991,
      "step": 27432
    },
    {
      "epoch": 0.939164669633687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.103,
      "step": 27433
    },
    {
      "epoch": 0.9391989044847655,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0751,
      "step": 27434
    },
    {
      "epoch": 0.9392331393358438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0987,
      "step": 27435
    },
    {
      "epoch": 0.9392673741869223,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0975,
      "step": 27436
    },
    {
      "epoch": 0.9393016090380006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0955,
      "step": 27437
    },
    {
      "epoch": 0.9393358438890791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0855,
      "step": 27438
    },
    {
      "epoch": 0.9393700787401574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.076,
      "step": 27439
    },
    {
      "epoch": 0.9394043135912359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1094,
      "step": 27440
    },
    {
      "epoch": 0.9394385484423142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0862,
      "step": 27441
    },
    {
      "epoch": 0.9394727832933927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 27442
    },
    {
      "epoch": 0.939507018144471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.08,
      "step": 27443
    },
    {
      "epoch": 0.9395412529955495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 27444
    },
    {
      "epoch": 0.9395754878466279,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0898,
      "step": 27445
    },
    {
      "epoch": 0.9396097226977063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0969,
      "step": 27446
    },
    {
      "epoch": 0.9396439575487847,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1011,
      "step": 27447
    },
    {
      "epoch": 0.939678192399863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0867,
      "step": 27448
    },
    {
      "epoch": 0.9397124272509415,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0821,
      "step": 27449
    },
    {
      "epoch": 0.9397466621020198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 27450
    },
    {
      "epoch": 0.9397808969530983,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.089,
      "step": 27451
    },
    {
      "epoch": 0.9398151318041766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1036,
      "step": 27452
    },
    {
      "epoch": 0.9398493666552551,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1029,
      "step": 27453
    },
    {
      "epoch": 0.9398836015063334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0819,
      "step": 27454
    },
    {
      "epoch": 0.9399178363574119,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0908,
      "step": 27455
    },
    {
      "epoch": 0.9399520712084902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0988,
      "step": 27456
    },
    {
      "epoch": 0.9399863060595687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1064,
      "step": 27457
    },
    {
      "epoch": 0.940020540910647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0956,
      "step": 27458
    },
    {
      "epoch": 0.9400547757617255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0753,
      "step": 27459
    },
    {
      "epoch": 0.9400890106128038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0882,
      "step": 27460
    },
    {
      "epoch": 0.9401232454638823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1002,
      "step": 27461
    },
    {
      "epoch": 0.9401574803149606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 27462
    },
    {
      "epoch": 0.940191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0809,
      "step": 27463
    },
    {
      "epoch": 0.9402259500171174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0898,
      "step": 27464
    },
    {
      "epoch": 0.9402601848681958,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0956,
      "step": 27465
    },
    {
      "epoch": 0.9402944197192742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0896,
      "step": 27466
    },
    {
      "epoch": 0.9403286545703526,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0896,
      "step": 27467
    },
    {
      "epoch": 0.940362889421431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0939,
      "step": 27468
    },
    {
      "epoch": 0.9403971242725094,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0924,
      "step": 27469
    },
    {
      "epoch": 0.9404313591235878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.099,
      "step": 27470
    },
    {
      "epoch": 0.9404655939746662,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.083,
      "step": 27471
    },
    {
      "epoch": 0.9404998288257446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0911,
      "step": 27472
    },
    {
      "epoch": 0.940534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0861,
      "step": 27473
    },
    {
      "epoch": 0.9405682985279014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1002,
      "step": 27474
    },
    {
      "epoch": 0.9406025333789798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0946,
      "step": 27475
    },
    {
      "epoch": 0.9406367682300582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1016,
      "step": 27476
    },
    {
      "epoch": 0.9406710030811366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 27477
    },
    {
      "epoch": 0.9407052379322149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0899,
      "step": 27478
    },
    {
      "epoch": 0.9407394727832934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1045,
      "step": 27479
    },
    {
      "epoch": 0.9407737076343717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0853,
      "step": 27480
    },
    {
      "epoch": 0.9408079424854502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1013,
      "step": 27481
    },
    {
      "epoch": 0.9408421773365286,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0921,
      "step": 27482
    },
    {
      "epoch": 0.940876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0913,
      "step": 27483
    },
    {
      "epoch": 0.9409106470386854,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0791,
      "step": 27484
    },
    {
      "epoch": 0.9409448818897638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0713,
      "step": 27485
    },
    {
      "epoch": 0.9409791167408422,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0805,
      "step": 27486
    },
    {
      "epoch": 0.9410133515919206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0959,
      "step": 27487
    },
    {
      "epoch": 0.941047586442999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.099,
      "step": 27488
    },
    {
      "epoch": 0.9410818212940774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0952,
      "step": 27489
    },
    {
      "epoch": 0.9411160561451558,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0844,
      "step": 27490
    },
    {
      "epoch": 0.9411502909962342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0812,
      "step": 27491
    },
    {
      "epoch": 0.9411845258473126,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0888,
      "step": 27492
    },
    {
      "epoch": 0.9412187606983909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0771,
      "step": 27493
    },
    {
      "epoch": 0.9412529955494694,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0911,
      "step": 27494
    },
    {
      "epoch": 0.9412872304005477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1013,
      "step": 27495
    },
    {
      "epoch": 0.9413214652516262,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0996,
      "step": 27496
    },
    {
      "epoch": 0.9413557001027045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0883,
      "step": 27497
    },
    {
      "epoch": 0.941389934953783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0628,
      "step": 27498
    },
    {
      "epoch": 0.9414241698048613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0971,
      "step": 27499
    },
    {
      "epoch": 0.9414584046559398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1106,
      "step": 27500
    },
    {
      "epoch": 0.9414926395070181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0757,
      "step": 27501
    },
    {
      "epoch": 0.9415268743580966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0652,
      "step": 27502
    },
    {
      "epoch": 0.9415611092091749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0908,
      "step": 27503
    },
    {
      "epoch": 0.9415953440602534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0921,
      "step": 27504
    },
    {
      "epoch": 0.9416295789113317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0977,
      "step": 27505
    },
    {
      "epoch": 0.9416638137624102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0846,
      "step": 27506
    },
    {
      "epoch": 0.9416980486134885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0958,
      "step": 27507
    },
    {
      "epoch": 0.9417322834645669,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0957,
      "step": 27508
    },
    {
      "epoch": 0.9417665183156453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0894,
      "step": 27509
    },
    {
      "epoch": 0.9418007531667237,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 27510
    },
    {
      "epoch": 0.9418349880178021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0903,
      "step": 27511
    },
    {
      "epoch": 0.9418692228688805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0931,
      "step": 27512
    },
    {
      "epoch": 0.9419034577199589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 27513
    },
    {
      "epoch": 0.9419376925710373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1054,
      "step": 27514
    },
    {
      "epoch": 0.9419719274221157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0953,
      "step": 27515
    },
    {
      "epoch": 0.9420061622731941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0851,
      "step": 27516
    },
    {
      "epoch": 0.9420403971242725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0807,
      "step": 27517
    },
    {
      "epoch": 0.9420746319753509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0906,
      "step": 27518
    },
    {
      "epoch": 0.9421088668264294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0952,
      "step": 27519
    },
    {
      "epoch": 0.9421431016775077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0956,
      "step": 27520
    },
    {
      "epoch": 0.9421773365285862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0902,
      "step": 27521
    },
    {
      "epoch": 0.9422115713796645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0868,
      "step": 27522
    },
    {
      "epoch": 0.9422458062307429,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0941,
      "step": 27523
    },
    {
      "epoch": 0.9422800410818213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0752,
      "step": 27524
    },
    {
      "epoch": 0.9423142759328997,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0802,
      "step": 27525
    },
    {
      "epoch": 0.9423485107839781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0977,
      "step": 27526
    },
    {
      "epoch": 0.9423827456350565,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0896,
      "step": 27527
    },
    {
      "epoch": 0.9424169804861349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.082,
      "step": 27528
    },
    {
      "epoch": 0.9424512153372133,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0939,
      "step": 27529
    },
    {
      "epoch": 0.9424854501882917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0881,
      "step": 27530
    },
    {
      "epoch": 0.9425196850393701,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1057,
      "step": 27531
    },
    {
      "epoch": 0.9425539198904485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.078,
      "step": 27532
    },
    {
      "epoch": 0.9425881547415269,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0904,
      "step": 27533
    },
    {
      "epoch": 0.9426223895926052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.084,
      "step": 27534
    },
    {
      "epoch": 0.9426566244436837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.105,
      "step": 27535
    },
    {
      "epoch": 0.942690859294762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0922,
      "step": 27536
    },
    {
      "epoch": 0.9427250941458405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.085,
      "step": 27537
    },
    {
      "epoch": 0.9427593289969188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1004,
      "step": 27538
    },
    {
      "epoch": 0.9427935638479973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0903,
      "step": 27539
    },
    {
      "epoch": 0.9428277986990756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0895,
      "step": 27540
    },
    {
      "epoch": 0.9428620335501541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0874,
      "step": 27541
    },
    {
      "epoch": 0.9428962684012324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1069,
      "step": 27542
    },
    {
      "epoch": 0.9429305032523109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0841,
      "step": 27543
    },
    {
      "epoch": 0.9429647381033892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0845,
      "step": 27544
    },
    {
      "epoch": 0.9429989729544677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0908,
      "step": 27545
    },
    {
      "epoch": 0.943033207805546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1071,
      "step": 27546
    },
    {
      "epoch": 0.9430674426566245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0819,
      "step": 27547
    },
    {
      "epoch": 0.9431016775077028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0937,
      "step": 27548
    },
    {
      "epoch": 0.9431359123587812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0673,
      "step": 27549
    },
    {
      "epoch": 0.9431701472098596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0946,
      "step": 27550
    },
    {
      "epoch": 0.943204382060938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0954,
      "step": 27551
    },
    {
      "epoch": 0.9432386169120164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0904,
      "step": 27552
    },
    {
      "epoch": 0.9432728517630948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0944,
      "step": 27553
    },
    {
      "epoch": 0.9433070866141732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1002,
      "step": 27554
    },
    {
      "epoch": 0.9433413214652516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0987,
      "step": 27555
    },
    {
      "epoch": 0.9433755563163301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1129,
      "step": 27556
    },
    {
      "epoch": 0.9434097911674084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0886,
      "step": 27557
    },
    {
      "epoch": 0.9434440260184869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0932,
      "step": 27558
    },
    {
      "epoch": 0.9434782608695652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0836,
      "step": 27559
    },
    {
      "epoch": 0.9435124957206437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 27560
    },
    {
      "epoch": 0.943546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0878,
      "step": 27561
    },
    {
      "epoch": 0.9435809654228005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0821,
      "step": 27562
    },
    {
      "epoch": 0.9436152002738788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.096,
      "step": 27563
    },
    {
      "epoch": 0.9436494351249572,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0897,
      "step": 27564
    },
    {
      "epoch": 0.9436836699760356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0952,
      "step": 27565
    },
    {
      "epoch": 0.943717904827114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0995,
      "step": 27566
    },
    {
      "epoch": 0.9437521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1049,
      "step": 27567
    },
    {
      "epoch": 0.9437863745292708,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.091,
      "step": 27568
    },
    {
      "epoch": 0.9438206093803492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0852,
      "step": 27569
    },
    {
      "epoch": 0.9438548442314276,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0829,
      "step": 27570
    },
    {
      "epoch": 0.943889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 27571
    },
    {
      "epoch": 0.9439233139335844,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.089,
      "step": 27572
    },
    {
      "epoch": 0.9439575487846628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1127,
      "step": 27573
    },
    {
      "epoch": 0.9439917836357412,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0997,
      "step": 27574
    },
    {
      "epoch": 0.9440260184868196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0951,
      "step": 27575
    },
    {
      "epoch": 0.944060253337898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0864,
      "step": 27576
    },
    {
      "epoch": 0.9440944881889763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0873,
      "step": 27577
    },
    {
      "epoch": 0.9441287230400548,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0769,
      "step": 27578
    },
    {
      "epoch": 0.9441629578911331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0942,
      "step": 27579
    },
    {
      "epoch": 0.9441971927422116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.091,
      "step": 27580
    },
    {
      "epoch": 0.9442314275932899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0816,
      "step": 27581
    },
    {
      "epoch": 0.9442656624443684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0978,
      "step": 27582
    },
    {
      "epoch": 0.9442998972954467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0869,
      "step": 27583
    },
    {
      "epoch": 0.9443341321465252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0976,
      "step": 27584
    },
    {
      "epoch": 0.9443683669976035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0801,
      "step": 27585
    },
    {
      "epoch": 0.944402601848682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0773,
      "step": 27586
    },
    {
      "epoch": 0.9444368366997603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0869,
      "step": 27587
    },
    {
      "epoch": 0.9444710715508388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0779,
      "step": 27588
    },
    {
      "epoch": 0.9445053064019171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.084,
      "step": 27589
    },
    {
      "epoch": 0.9445395412529956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.091,
      "step": 27590
    },
    {
      "epoch": 0.9445737761040739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1015,
      "step": 27591
    },
    {
      "epoch": 0.9446080109551523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0856,
      "step": 27592
    },
    {
      "epoch": 0.9446422458062308,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1012,
      "step": 27593
    },
    {
      "epoch": 0.9446764806573091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0888,
      "step": 27594
    },
    {
      "epoch": 0.9447107155083876,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0898,
      "step": 27595
    },
    {
      "epoch": 0.9447449503594659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.089,
      "step": 27596
    },
    {
      "epoch": 0.9447791852105444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1027,
      "step": 27597
    },
    {
      "epoch": 0.9448134200616227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 27598
    },
    {
      "epoch": 0.9448476549127012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0913,
      "step": 27599
    },
    {
      "epoch": 0.9448818897637795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0873,
      "step": 27600
    },
    {
      "epoch": 0.944916124614858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0871,
      "step": 27601
    },
    {
      "epoch": 0.9449503594659363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0826,
      "step": 27602
    },
    {
      "epoch": 0.9449845943170148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1067,
      "step": 27603
    },
    {
      "epoch": 0.9450188291680931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.076,
      "step": 27604
    },
    {
      "epoch": 0.9450530640191716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0909,
      "step": 27605
    },
    {
      "epoch": 0.9450872988702499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0934,
      "step": 27606
    },
    {
      "epoch": 0.9451215337213283,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.084,
      "step": 27607
    },
    {
      "epoch": 0.9451557685724067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1048,
      "step": 27608
    },
    {
      "epoch": 0.9451900034234851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0822,
      "step": 27609
    },
    {
      "epoch": 0.9452242382745635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.097,
      "step": 27610
    },
    {
      "epoch": 0.9452584731256419,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0995,
      "step": 27611
    },
    {
      "epoch": 0.9452927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0763,
      "step": 27612
    },
    {
      "epoch": 0.9453269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0751,
      "step": 27613
    },
    {
      "epoch": 0.9453611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.08,
      "step": 27614
    },
    {
      "epoch": 0.9453954125299555,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0962,
      "step": 27615
    },
    {
      "epoch": 0.9454296473810339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 27616
    },
    {
      "epoch": 0.9454638822321123,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.088,
      "step": 27617
    },
    {
      "epoch": 0.9454981170831906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0894,
      "step": 27618
    },
    {
      "epoch": 0.9455323519342691,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1009,
      "step": 27619
    },
    {
      "epoch": 0.9455665867853474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0812,
      "step": 27620
    },
    {
      "epoch": 0.9456008216364259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0933,
      "step": 27621
    },
    {
      "epoch": 0.9456350564875042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0934,
      "step": 27622
    },
    {
      "epoch": 0.9456692913385827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0969,
      "step": 27623
    },
    {
      "epoch": 0.945703526189661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.094,
      "step": 27624
    },
    {
      "epoch": 0.9457377610407395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0885,
      "step": 27625
    },
    {
      "epoch": 0.9457719958918178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0942,
      "step": 27626
    },
    {
      "epoch": 0.9458062307428963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0851,
      "step": 27627
    },
    {
      "epoch": 0.9458404655939746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0934,
      "step": 27628
    },
    {
      "epoch": 0.9458747004450531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.097,
      "step": 27629
    },
    {
      "epoch": 0.9459089352961315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0902,
      "step": 27630
    },
    {
      "epoch": 0.9459431701472099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1078,
      "step": 27631
    },
    {
      "epoch": 0.9459774049982883,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 27632
    },
    {
      "epoch": 0.9460116398493666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0817,
      "step": 27633
    },
    {
      "epoch": 0.9460458747004451,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0861,
      "step": 27634
    },
    {
      "epoch": 0.9460801095515234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0783,
      "step": 27635
    },
    {
      "epoch": 0.9461143444026019,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0874,
      "step": 27636
    },
    {
      "epoch": 0.9461485792536802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 27637
    },
    {
      "epoch": 0.9461828141047587,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.081,
      "step": 27638
    },
    {
      "epoch": 0.946217048955837,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0882,
      "step": 27639
    },
    {
      "epoch": 0.9462512838069155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.087,
      "step": 27640
    },
    {
      "epoch": 0.9462855186579938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0948,
      "step": 27641
    },
    {
      "epoch": 0.9463197535090723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0774,
      "step": 27642
    },
    {
      "epoch": 0.9463539883601506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1,
      "step": 27643
    },
    {
      "epoch": 0.9463882232112291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0683,
      "step": 27644
    },
    {
      "epoch": 0.9464224580623074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1069,
      "step": 27645
    },
    {
      "epoch": 0.9464566929133859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0909,
      "step": 27646
    },
    {
      "epoch": 0.9464909277644642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0879,
      "step": 27647
    },
    {
      "epoch": 0.9465251626155426,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0932,
      "step": 27648
    },
    {
      "epoch": 0.946559397466621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0888,
      "step": 27649
    },
    {
      "epoch": 0.9465936323176994,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 27650
    },
    {
      "epoch": 0.9466278671687778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0868,
      "step": 27651
    },
    {
      "epoch": 0.9466621020198562,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.084,
      "step": 27652
    },
    {
      "epoch": 0.9466963368709346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1086,
      "step": 27653
    },
    {
      "epoch": 0.946730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1002,
      "step": 27654
    },
    {
      "epoch": 0.9467648065730914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0801,
      "step": 27655
    },
    {
      "epoch": 0.9467990414241698,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0817,
      "step": 27656
    },
    {
      "epoch": 0.9468332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0774,
      "step": 27657
    },
    {
      "epoch": 0.9468675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0934,
      "step": 27658
    },
    {
      "epoch": 0.946901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0849,
      "step": 27659
    },
    {
      "epoch": 0.9469359808284834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0861,
      "step": 27660
    },
    {
      "epoch": 0.9469702156795617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0893,
      "step": 27661
    },
    {
      "epoch": 0.9470044505306402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0928,
      "step": 27662
    },
    {
      "epoch": 0.9470386853817185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0857,
      "step": 27663
    },
    {
      "epoch": 0.947072920232797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0863,
      "step": 27664
    },
    {
      "epoch": 0.9471071550838754,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0866,
      "step": 27665
    },
    {
      "epoch": 0.9471413899349538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0939,
      "step": 27666
    },
    {
      "epoch": 0.9471756247860322,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0869,
      "step": 27667
    },
    {
      "epoch": 0.9472098596371106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0792,
      "step": 27668
    },
    {
      "epoch": 0.947244094488189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0775,
      "step": 27669
    },
    {
      "epoch": 0.9472783293392674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0902,
      "step": 27670
    },
    {
      "epoch": 0.9473125641903458,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0871,
      "step": 27671
    },
    {
      "epoch": 0.9473467990414242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0815,
      "step": 27672
    },
    {
      "epoch": 0.9473810338925026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0897,
      "step": 27673
    },
    {
      "epoch": 0.947415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0969,
      "step": 27674
    },
    {
      "epoch": 0.9474495035946594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.098,
      "step": 27675
    },
    {
      "epoch": 0.9474837384457377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0919,
      "step": 27676
    },
    {
      "epoch": 0.9475179732968162,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0778,
      "step": 27677
    },
    {
      "epoch": 0.9475522081478945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0851,
      "step": 27678
    },
    {
      "epoch": 0.947586442998973,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0962,
      "step": 27679
    },
    {
      "epoch": 0.9476206778500513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0993,
      "step": 27680
    },
    {
      "epoch": 0.9476549127011298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0811,
      "step": 27681
    },
    {
      "epoch": 0.9476891475522081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0962,
      "step": 27682
    },
    {
      "epoch": 0.9477233824032866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0763,
      "step": 27683
    },
    {
      "epoch": 0.9477576172543649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0903,
      "step": 27684
    },
    {
      "epoch": 0.9477918521054434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0796,
      "step": 27685
    },
    {
      "epoch": 0.9478260869565217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0758,
      "step": 27686
    },
    {
      "epoch": 0.9478603218076002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0808,
      "step": 27687
    },
    {
      "epoch": 0.9478945566586785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0885,
      "step": 27688
    },
    {
      "epoch": 0.947928791509757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0872,
      "step": 27689
    },
    {
      "epoch": 0.9479630263608353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0829,
      "step": 27690
    },
    {
      "epoch": 0.9479972612119137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0894,
      "step": 27691
    },
    {
      "epoch": 0.9480314960629921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1039,
      "step": 27692
    },
    {
      "epoch": 0.9480657309140705,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0762,
      "step": 27693
    },
    {
      "epoch": 0.9480999657651489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.076,
      "step": 27694
    },
    {
      "epoch": 0.9481342006162273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0968,
      "step": 27695
    },
    {
      "epoch": 0.9481684354673057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 27696
    },
    {
      "epoch": 0.9482026703183841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0764,
      "step": 27697
    },
    {
      "epoch": 0.9482369051694625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0878,
      "step": 27698
    },
    {
      "epoch": 0.9482711400205409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0874,
      "step": 27699
    },
    {
      "epoch": 0.9483053748716193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0848,
      "step": 27700
    },
    {
      "epoch": 0.9483396097226977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.092,
      "step": 27701
    },
    {
      "epoch": 0.9483738445737762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0762,
      "step": 27702
    },
    {
      "epoch": 0.9484080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1096,
      "step": 27703
    },
    {
      "epoch": 0.948442314275933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1012,
      "step": 27704
    },
    {
      "epoch": 0.9484765491270113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0931,
      "step": 27705
    },
    {
      "epoch": 0.9485107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0973,
      "step": 27706
    },
    {
      "epoch": 0.9485450188291681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1055,
      "step": 27707
    },
    {
      "epoch": 0.9485792536802465,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0847,
      "step": 27708
    },
    {
      "epoch": 0.9486134885313249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0951,
      "step": 27709
    },
    {
      "epoch": 0.9486477233824033,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0995,
      "step": 27710
    },
    {
      "epoch": 0.9486819582334817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0916,
      "step": 27711
    },
    {
      "epoch": 0.9487161930845601,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1016,
      "step": 27712
    },
    {
      "epoch": 0.9487504279356385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0711,
      "step": 27713
    },
    {
      "epoch": 0.9487846627867169,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0812,
      "step": 27714
    },
    {
      "epoch": 0.9488188976377953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0811,
      "step": 27715
    },
    {
      "epoch": 0.9488531324888737,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0935,
      "step": 27716
    },
    {
      "epoch": 0.948887367339952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 27717
    },
    {
      "epoch": 0.9489216021910305,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.068,
      "step": 27718
    },
    {
      "epoch": 0.9489558370421088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0868,
      "step": 27719
    },
    {
      "epoch": 0.9489900718931873,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 27720
    },
    {
      "epoch": 0.9490243067442656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.083,
      "step": 27721
    },
    {
      "epoch": 0.9490585415953441,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0779,
      "step": 27722
    },
    {
      "epoch": 0.9490927764464224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.09,
      "step": 27723
    },
    {
      "epoch": 0.9491270112975009,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.087,
      "step": 27724
    },
    {
      "epoch": 0.9491612461485792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0764,
      "step": 27725
    },
    {
      "epoch": 0.9491954809996577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0907,
      "step": 27726
    },
    {
      "epoch": 0.949229715850736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 27727
    },
    {
      "epoch": 0.9492639507018145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0791,
      "step": 27728
    },
    {
      "epoch": 0.9492981855528928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.09,
      "step": 27729
    },
    {
      "epoch": 0.9493324204039713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0944,
      "step": 27730
    },
    {
      "epoch": 0.9493666552550496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1024,
      "step": 27731
    },
    {
      "epoch": 0.949400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.088,
      "step": 27732
    },
    {
      "epoch": 0.9494351249572064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0852,
      "step": 27733
    },
    {
      "epoch": 0.9494693598082848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0932,
      "step": 27734
    },
    {
      "epoch": 0.9495035946593632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0796,
      "step": 27735
    },
    {
      "epoch": 0.9495378295104416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0716,
      "step": 27736
    },
    {
      "epoch": 0.94957206436152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.071,
      "step": 27737
    },
    {
      "epoch": 0.9496062992125984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0927,
      "step": 27738
    },
    {
      "epoch": 0.9496405340636769,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.078,
      "step": 27739
    },
    {
      "epoch": 0.9496747689147552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0931,
      "step": 27740
    },
    {
      "epoch": 0.9497090037658337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 27741
    },
    {
      "epoch": 0.949743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0765,
      "step": 27742
    },
    {
      "epoch": 0.9497774734679905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1003,
      "step": 27743
    },
    {
      "epoch": 0.9498117083190688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0802,
      "step": 27744
    },
    {
      "epoch": 0.9498459431701473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 27745
    },
    {
      "epoch": 0.9498801780212256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0637,
      "step": 27746
    },
    {
      "epoch": 0.949914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0854,
      "step": 27747
    },
    {
      "epoch": 0.9499486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0827,
      "step": 27748
    },
    {
      "epoch": 0.9499828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0822,
      "step": 27749
    },
    {
      "epoch": 0.9500171174255392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0928,
      "step": 27750
    },
    {
      "epoch": 0.9500513522766176,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0862,
      "step": 27751
    },
    {
      "epoch": 0.950085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0827,
      "step": 27752
    },
    {
      "epoch": 0.9501198219787744,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0892,
      "step": 27753
    },
    {
      "epoch": 0.9501540568298528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0757,
      "step": 27754
    },
    {
      "epoch": 0.9501882916809312,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0802,
      "step": 27755
    },
    {
      "epoch": 0.9502225265320096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0906,
      "step": 27756
    },
    {
      "epoch": 0.950256761383088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0901,
      "step": 27757
    },
    {
      "epoch": 0.9502909962341664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 27758
    },
    {
      "epoch": 0.9503252310852448,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0807,
      "step": 27759
    },
    {
      "epoch": 0.9503594659363231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.082,
      "step": 27760
    },
    {
      "epoch": 0.9503937007874016,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0805,
      "step": 27761
    },
    {
      "epoch": 0.9504279356384799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0886,
      "step": 27762
    },
    {
      "epoch": 0.9504621704895584,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0837,
      "step": 27763
    },
    {
      "epoch": 0.9504964053406367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0909,
      "step": 27764
    },
    {
      "epoch": 0.9505306401917152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.081,
      "step": 27765
    },
    {
      "epoch": 0.9505648750427935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0953,
      "step": 27766
    },
    {
      "epoch": 0.950599109893872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0874,
      "step": 27767
    },
    {
      "epoch": 0.9506333447449503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0893,
      "step": 27768
    },
    {
      "epoch": 0.9506675795960288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0968,
      "step": 27769
    },
    {
      "epoch": 0.9507018144471071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1003,
      "step": 27770
    },
    {
      "epoch": 0.9507360492981856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0816,
      "step": 27771
    },
    {
      "epoch": 0.9507702841492639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0816,
      "step": 27772
    },
    {
      "epoch": 0.9508045190003424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0919,
      "step": 27773
    },
    {
      "epoch": 0.9508387538514207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0906,
      "step": 27774
    },
    {
      "epoch": 0.9508729887024991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.086,
      "step": 27775
    },
    {
      "epoch": 0.9509072235535776,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0955,
      "step": 27776
    },
    {
      "epoch": 0.9509414584046559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0867,
      "step": 27777
    },
    {
      "epoch": 0.9509756932557344,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 27778
    },
    {
      "epoch": 0.9510099281068127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 27779
    },
    {
      "epoch": 0.9510441629578912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0952,
      "step": 27780
    },
    {
      "epoch": 0.9510783978089695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1031,
      "step": 27781
    },
    {
      "epoch": 0.951112632660048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0659,
      "step": 27782
    },
    {
      "epoch": 0.9511468675111263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0742,
      "step": 27783
    },
    {
      "epoch": 0.9511811023622048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 27784
    },
    {
      "epoch": 0.9512153372132831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0781,
      "step": 27785
    },
    {
      "epoch": 0.9512495720643616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0914,
      "step": 27786
    },
    {
      "epoch": 0.9512838069154399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1009,
      "step": 27787
    },
    {
      "epoch": 0.9513180417665184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0761,
      "step": 27788
    },
    {
      "epoch": 0.9513522766175967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0883,
      "step": 27789
    },
    {
      "epoch": 0.9513865114686751,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0876,
      "step": 27790
    },
    {
      "epoch": 0.9514207463197535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0831,
      "step": 27791
    },
    {
      "epoch": 0.9514549811708319,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0866,
      "step": 27792
    },
    {
      "epoch": 0.9514892160219103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1067,
      "step": 27793
    },
    {
      "epoch": 0.9515234508729887,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 27794
    },
    {
      "epoch": 0.9515576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0952,
      "step": 27795
    },
    {
      "epoch": 0.9515919205751455,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0777,
      "step": 27796
    },
    {
      "epoch": 0.9516261554262239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.083,
      "step": 27797
    },
    {
      "epoch": 0.9516603902773023,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.082,
      "step": 27798
    },
    {
      "epoch": 0.9516946251283807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0852,
      "step": 27799
    },
    {
      "epoch": 0.9517288599794591,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0788,
      "step": 27800
    },
    {
      "epoch": 0.9517630948305374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1138,
      "step": 27801
    },
    {
      "epoch": 0.9517973296816159,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0958,
      "step": 27802
    },
    {
      "epoch": 0.9518315645326942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.088,
      "step": 27803
    },
    {
      "epoch": 0.9518657993837727,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0888,
      "step": 27804
    },
    {
      "epoch": 0.951900034234851,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0807,
      "step": 27805
    },
    {
      "epoch": 0.9519342690859295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0862,
      "step": 27806
    },
    {
      "epoch": 0.9519685039370078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0712,
      "step": 27807
    },
    {
      "epoch": 0.9520027387880863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0824,
      "step": 27808
    },
    {
      "epoch": 0.9520369736391646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0889,
      "step": 27809
    },
    {
      "epoch": 0.9520712084902431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.082,
      "step": 27810
    },
    {
      "epoch": 0.9521054433413214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0776,
      "step": 27811
    },
    {
      "epoch": 0.9521396781923999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0858,
      "step": 27812
    },
    {
      "epoch": 0.9521739130434783,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0822,
      "step": 27813
    },
    {
      "epoch": 0.9522081478945567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0749,
      "step": 27814
    },
    {
      "epoch": 0.9522423827456351,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0832,
      "step": 27815
    },
    {
      "epoch": 0.9522766175967134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1039,
      "step": 27816
    },
    {
      "epoch": 0.9523108524477919,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0857,
      "step": 27817
    },
    {
      "epoch": 0.9523450872988702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0947,
      "step": 27818
    },
    {
      "epoch": 0.9523793221499487,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0889,
      "step": 27819
    },
    {
      "epoch": 0.952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0672,
      "step": 27820
    },
    {
      "epoch": 0.9524477918521055,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0982,
      "step": 27821
    },
    {
      "epoch": 0.9524820267031838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0895,
      "step": 27822
    },
    {
      "epoch": 0.9525162615542623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0853,
      "step": 27823
    },
    {
      "epoch": 0.9525504964053406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0896,
      "step": 27824
    },
    {
      "epoch": 0.9525847312564191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1043,
      "step": 27825
    },
    {
      "epoch": 0.9526189661074974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0852,
      "step": 27826
    },
    {
      "epoch": 0.9526532009585759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0723,
      "step": 27827
    },
    {
      "epoch": 0.9526874358096542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0884,
      "step": 27828
    },
    {
      "epoch": 0.9527216706607327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0812,
      "step": 27829
    },
    {
      "epoch": 0.952755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0818,
      "step": 27830
    },
    {
      "epoch": 0.9527901403628894,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.083,
      "step": 27831
    },
    {
      "epoch": 0.9528243752139678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0765,
      "step": 27832
    },
    {
      "epoch": 0.9528586100650462,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 27833
    },
    {
      "epoch": 0.9528928449161246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 27834
    },
    {
      "epoch": 0.952927079767203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.068,
      "step": 27835
    },
    {
      "epoch": 0.9529613146182814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.085,
      "step": 27836
    },
    {
      "epoch": 0.9529955494693598,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 27837
    },
    {
      "epoch": 0.9530297843204382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0998,
      "step": 27838
    },
    {
      "epoch": 0.9530640191715166,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.088,
      "step": 27839
    },
    {
      "epoch": 0.953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0891,
      "step": 27840
    },
    {
      "epoch": 0.9531324888736734,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0828,
      "step": 27841
    },
    {
      "epoch": 0.9531667237247518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1047,
      "step": 27842
    },
    {
      "epoch": 0.9532009585758302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0901,
      "step": 27843
    },
    {
      "epoch": 0.9532351934269085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0913,
      "step": 27844
    },
    {
      "epoch": 0.953269428277987,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0819,
      "step": 27845
    },
    {
      "epoch": 0.9533036631290653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0787,
      "step": 27846
    },
    {
      "epoch": 0.9533378979801438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0896,
      "step": 27847
    },
    {
      "epoch": 0.9533721328312221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0992,
      "step": 27848
    },
    {
      "epoch": 0.9534063676823006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0672,
      "step": 27849
    },
    {
      "epoch": 0.953440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0763,
      "step": 27850
    },
    {
      "epoch": 0.9534748373844574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0827,
      "step": 27851
    },
    {
      "epoch": 0.9535090722355358,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0761,
      "step": 27852
    },
    {
      "epoch": 0.9535433070866142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0674,
      "step": 27853
    },
    {
      "epoch": 0.9535775419376926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0649,
      "step": 27854
    },
    {
      "epoch": 0.953611776788771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0973,
      "step": 27855
    },
    {
      "epoch": 0.9536460116398494,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0772,
      "step": 27856
    },
    {
      "epoch": 0.9536802464909278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0822,
      "step": 27857
    },
    {
      "epoch": 0.9537144813420062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.099,
      "step": 27858
    },
    {
      "epoch": 0.9537487161930845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0907,
      "step": 27859
    },
    {
      "epoch": 0.953782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0839,
      "step": 27860
    },
    {
      "epoch": 0.9538171858952413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0886,
      "step": 27861
    },
    {
      "epoch": 0.9538514207463198,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0873,
      "step": 27862
    },
    {
      "epoch": 0.9538856555973981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0869,
      "step": 27863
    },
    {
      "epoch": 0.9539198904484766,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0866,
      "step": 27864
    },
    {
      "epoch": 0.9539541252995549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0919,
      "step": 27865
    },
    {
      "epoch": 0.9539883601506334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0806,
      "step": 27866
    },
    {
      "epoch": 0.9540225950017117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0901,
      "step": 27867
    },
    {
      "epoch": 0.9540568298527902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0851,
      "step": 27868
    },
    {
      "epoch": 0.9540910647038685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0919,
      "step": 27869
    },
    {
      "epoch": 0.954125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0925,
      "step": 27870
    },
    {
      "epoch": 0.9541595344060253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0811,
      "step": 27871
    },
    {
      "epoch": 0.9541937692571038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0714,
      "step": 27872
    },
    {
      "epoch": 0.9542280041081821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0826,
      "step": 27873
    },
    {
      "epoch": 0.9542622389592605,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0841,
      "step": 27874
    },
    {
      "epoch": 0.9542964738103389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0871,
      "step": 27875
    },
    {
      "epoch": 0.9543307086614173,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0849,
      "step": 27876
    },
    {
      "epoch": 0.9543649435124957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.098,
      "step": 27877
    },
    {
      "epoch": 0.9543991783635741,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.088,
      "step": 27878
    },
    {
      "epoch": 0.9544334132146525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0713,
      "step": 27879
    },
    {
      "epoch": 0.9544676480657309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.076,
      "step": 27880
    },
    {
      "epoch": 0.9545018829168093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0807,
      "step": 27881
    },
    {
      "epoch": 0.9545361177678877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0752,
      "step": 27882
    },
    {
      "epoch": 0.9545703526189661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0939,
      "step": 27883
    },
    {
      "epoch": 0.9546045874700445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0947,
      "step": 27884
    },
    {
      "epoch": 0.9546388223211228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0948,
      "step": 27885
    },
    {
      "epoch": 0.9546730571722013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0957,
      "step": 27886
    },
    {
      "epoch": 0.9547072920232798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 27887
    },
    {
      "epoch": 0.9547415268743581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.082,
      "step": 27888
    },
    {
      "epoch": 0.9547757617254365,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0825,
      "step": 27889
    },
    {
      "epoch": 0.9548099965765149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.078,
      "step": 27890
    },
    {
      "epoch": 0.9548442314275933,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.081,
      "step": 27891
    },
    {
      "epoch": 0.9548784662786717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0919,
      "step": 27892
    },
    {
      "epoch": 0.9549127011297501,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0706,
      "step": 27893
    },
    {
      "epoch": 0.9549469359808285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0717,
      "step": 27894
    },
    {
      "epoch": 0.9549811708319069,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0766,
      "step": 27895
    },
    {
      "epoch": 0.9550154056829853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0846,
      "step": 27896
    },
    {
      "epoch": 0.9550496405340637,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0895,
      "step": 27897
    },
    {
      "epoch": 0.9550838753851421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.083,
      "step": 27898
    },
    {
      "epoch": 0.9551181102362205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0825,
      "step": 27899
    },
    {
      "epoch": 0.9551523450872988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0857,
      "step": 27900
    },
    {
      "epoch": 0.9551865799383773,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0814,
      "step": 27901
    },
    {
      "epoch": 0.9552208147894556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0762,
      "step": 27902
    },
    {
      "epoch": 0.9552550496405341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 27903
    },
    {
      "epoch": 0.9552892844916124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0753,
      "step": 27904
    },
    {
      "epoch": 0.9553235193426909,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0664,
      "step": 27905
    },
    {
      "epoch": 0.9553577541937692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 27906
    },
    {
      "epoch": 0.9553919890448477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0898,
      "step": 27907
    },
    {
      "epoch": 0.955426223895926,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0778,
      "step": 27908
    },
    {
      "epoch": 0.9554604587470045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0604,
      "step": 27909
    },
    {
      "epoch": 0.9554946935980828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 27910
    },
    {
      "epoch": 0.9555289284491613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0831,
      "step": 27911
    },
    {
      "epoch": 0.9555631633002396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0966,
      "step": 27912
    },
    {
      "epoch": 0.9555973981513181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0808,
      "step": 27913
    },
    {
      "epoch": 0.9556316330023964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.083,
      "step": 27914
    },
    {
      "epoch": 0.9556658678534748,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0697,
      "step": 27915
    },
    {
      "epoch": 0.9557001027045532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0712,
      "step": 27916
    },
    {
      "epoch": 0.9557343375556316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0831,
      "step": 27917
    },
    {
      "epoch": 0.95576857240671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.063,
      "step": 27918
    },
    {
      "epoch": 0.9558028072577884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0844,
      "step": 27919
    },
    {
      "epoch": 0.9558370421088668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0769,
      "step": 27920
    },
    {
      "epoch": 0.9558712769599452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0767,
      "step": 27921
    },
    {
      "epoch": 0.9559055118110236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0879,
      "step": 27922
    },
    {
      "epoch": 0.955939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.101,
      "step": 27923
    },
    {
      "epoch": 0.9559739815131805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0922,
      "step": 27924
    },
    {
      "epoch": 0.9560082163642588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0907,
      "step": 27925
    },
    {
      "epoch": 0.9560424512153373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.101,
      "step": 27926
    },
    {
      "epoch": 0.9560766860664156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0851,
      "step": 27927
    },
    {
      "epoch": 0.9561109209174941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0736,
      "step": 27928
    },
    {
      "epoch": 0.9561451557685724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0996,
      "step": 27929
    },
    {
      "epoch": 0.9561793906196508,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0869,
      "step": 27930
    },
    {
      "epoch": 0.9562136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0795,
      "step": 27931
    },
    {
      "epoch": 0.9562478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0832,
      "step": 27932
    },
    {
      "epoch": 0.956282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0933,
      "step": 27933
    },
    {
      "epoch": 0.9563163300239644,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 27934
    },
    {
      "epoch": 0.9563505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0916,
      "step": 27935
    },
    {
      "epoch": 0.9563847997261212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0826,
      "step": 27936
    },
    {
      "epoch": 0.9564190345771996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0859,
      "step": 27937
    },
    {
      "epoch": 0.956453269428278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0842,
      "step": 27938
    },
    {
      "epoch": 0.9564875042793564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0982,
      "step": 27939
    },
    {
      "epoch": 0.9565217391304348,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0875,
      "step": 27940
    },
    {
      "epoch": 0.9565559739815132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0795,
      "step": 27941
    },
    {
      "epoch": 0.9565902088325916,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0932,
      "step": 27942
    },
    {
      "epoch": 0.95662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0618,
      "step": 27943
    },
    {
      "epoch": 0.9566586785347484,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0798,
      "step": 27944
    },
    {
      "epoch": 0.9566929133858267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0893,
      "step": 27945
    },
    {
      "epoch": 0.9567271482369052,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0743,
      "step": 27946
    },
    {
      "epoch": 0.9567613830879835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0739,
      "step": 27947
    },
    {
      "epoch": 0.956795617939062,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1,
      "step": 27948
    },
    {
      "epoch": 0.9568298527901403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0992,
      "step": 27949
    },
    {
      "epoch": 0.9568640876412188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0831,
      "step": 27950
    },
    {
      "epoch": 0.9568983224922971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0963,
      "step": 27951
    },
    {
      "epoch": 0.9569325573433756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0985,
      "step": 27952
    },
    {
      "epoch": 0.9569667921944539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0875,
      "step": 27953
    },
    {
      "epoch": 0.9570010270455324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0757,
      "step": 27954
    },
    {
      "epoch": 0.9570352618966107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0882,
      "step": 27955
    },
    {
      "epoch": 0.9570694967476892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0904,
      "step": 27956
    },
    {
      "epoch": 0.9571037315987675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0729,
      "step": 27957
    },
    {
      "epoch": 0.957137966449846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0886,
      "step": 27958
    },
    {
      "epoch": 0.9571722013009243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0775,
      "step": 27959
    },
    {
      "epoch": 0.9572064361520027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0756,
      "step": 27960
    },
    {
      "epoch": 0.9572406710030812,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0692,
      "step": 27961
    },
    {
      "epoch": 0.9572749058541595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1015,
      "step": 27962
    },
    {
      "epoch": 0.957309140705238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0794,
      "step": 27963
    },
    {
      "epoch": 0.9573433755563163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1032,
      "step": 27964
    },
    {
      "epoch": 0.9573776104073948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0831,
      "step": 27965
    },
    {
      "epoch": 0.9574118452584731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0885,
      "step": 27966
    },
    {
      "epoch": 0.9574460801095516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0878,
      "step": 27967
    },
    {
      "epoch": 0.9574803149606299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0836,
      "step": 27968
    },
    {
      "epoch": 0.9575145498117084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0783,
      "step": 27969
    },
    {
      "epoch": 0.9575487846627867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0773,
      "step": 27970
    },
    {
      "epoch": 0.9575830195138652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1053,
      "step": 27971
    },
    {
      "epoch": 0.9576172543649435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0828,
      "step": 27972
    },
    {
      "epoch": 0.957651489216022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0777,
      "step": 27973
    },
    {
      "epoch": 0.9576857240671003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0871,
      "step": 27974
    },
    {
      "epoch": 0.9577199589181787,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0739,
      "step": 27975
    },
    {
      "epoch": 0.9577541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0753,
      "step": 27976
    },
    {
      "epoch": 0.9577884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0728,
      "step": 27977
    },
    {
      "epoch": 0.9578226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0778,
      "step": 27978
    },
    {
      "epoch": 0.9578568983224923,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0771,
      "step": 27979
    },
    {
      "epoch": 0.9578911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 27980
    },
    {
      "epoch": 0.9579253680246491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0952,
      "step": 27981
    },
    {
      "epoch": 0.9579596028757275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0828,
      "step": 27982
    },
    {
      "epoch": 0.9579938377268059,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0903,
      "step": 27983
    },
    {
      "epoch": 0.9580280725778842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0762,
      "step": 27984
    },
    {
      "epoch": 0.9580623074289627,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0926,
      "step": 27985
    },
    {
      "epoch": 0.958096542280041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0874,
      "step": 27986
    },
    {
      "epoch": 0.9581307771311195,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.085,
      "step": 27987
    },
    {
      "epoch": 0.9581650119821978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0965,
      "step": 27988
    },
    {
      "epoch": 0.9581992468332763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 27989
    },
    {
      "epoch": 0.9582334816843546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0726,
      "step": 27990
    },
    {
      "epoch": 0.9582677165354331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0791,
      "step": 27991
    },
    {
      "epoch": 0.9583019513865114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0784,
      "step": 27992
    },
    {
      "epoch": 0.9583361862375899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0879,
      "step": 27993
    },
    {
      "epoch": 0.9583704210886682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0835,
      "step": 27994
    },
    {
      "epoch": 0.9584046559397467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0881,
      "step": 27995
    },
    {
      "epoch": 0.9584388907908251,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0763,
      "step": 27996
    },
    {
      "epoch": 0.9584731256419035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0757,
      "step": 27997
    },
    {
      "epoch": 0.9585073604929819,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 27998
    },
    {
      "epoch": 0.9585415953440602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0849,
      "step": 27999
    },
    {
      "epoch": 0.9585758301951387,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0992,
      "step": 28000
    },
    {
      "epoch": 0.958610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0908,
      "step": 28001
    },
    {
      "epoch": 0.9586442998972955,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0678,
      "step": 28002
    },
    {
      "epoch": 0.9586785347483738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0919,
      "step": 28003
    },
    {
      "epoch": 0.9587127695994523,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0729,
      "step": 28004
    },
    {
      "epoch": 0.9587470044505306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0739,
      "step": 28005
    },
    {
      "epoch": 0.9587812393016091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.073,
      "step": 28006
    },
    {
      "epoch": 0.9588154741526874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 28007
    },
    {
      "epoch": 0.9588497090037659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0777,
      "step": 28008
    },
    {
      "epoch": 0.9588839438548442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0841,
      "step": 28009
    },
    {
      "epoch": 0.9589181787059227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0899,
      "step": 28010
    },
    {
      "epoch": 0.958952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0988,
      "step": 28011
    },
    {
      "epoch": 0.9589866484080795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0841,
      "step": 28012
    },
    {
      "epoch": 0.9590208832591578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1048,
      "step": 28013
    },
    {
      "epoch": 0.9590551181102362,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0946,
      "step": 28014
    },
    {
      "epoch": 0.9590893529613146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0861,
      "step": 28015
    },
    {
      "epoch": 0.959123587812393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0787,
      "step": 28016
    },
    {
      "epoch": 0.9591578226634714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 28017
    },
    {
      "epoch": 0.9591920575145498,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0895,
      "step": 28018
    },
    {
      "epoch": 0.9592262923656282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0654,
      "step": 28019
    },
    {
      "epoch": 0.9592605272167066,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0934,
      "step": 28020
    },
    {
      "epoch": 0.959294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0735,
      "step": 28021
    },
    {
      "epoch": 0.9593289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0708,
      "step": 28022
    },
    {
      "epoch": 0.9593632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0812,
      "step": 28023
    },
    {
      "epoch": 0.9593974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0744,
      "step": 28024
    },
    {
      "epoch": 0.9594317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0958,
      "step": 28025
    },
    {
      "epoch": 0.959465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0745,
      "step": 28026
    },
    {
      "epoch": 0.9595001711742553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0894,
      "step": 28027
    },
    {
      "epoch": 0.9595344060253338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0888,
      "step": 28028
    },
    {
      "epoch": 0.9595686408764121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0815,
      "step": 28029
    },
    {
      "epoch": 0.9596028757274906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0726,
      "step": 28030
    },
    {
      "epoch": 0.9596371105785689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1054,
      "step": 28031
    },
    {
      "epoch": 0.9596713454296474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.084,
      "step": 28032
    },
    {
      "epoch": 0.9597055802807258,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0934,
      "step": 28033
    },
    {
      "epoch": 0.9597398151318042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0755,
      "step": 28034
    },
    {
      "epoch": 0.9597740499828826,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0911,
      "step": 28035
    },
    {
      "epoch": 0.959808284833961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 28036
    },
    {
      "epoch": 0.9598425196850394,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0771,
      "step": 28037
    },
    {
      "epoch": 0.9598767545361178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0833,
      "step": 28038
    },
    {
      "epoch": 0.9599109893871962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0665,
      "step": 28039
    },
    {
      "epoch": 0.9599452242382746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0821,
      "step": 28040
    },
    {
      "epoch": 0.959979459089353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0822,
      "step": 28041
    },
    {
      "epoch": 0.9600136939404313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0771,
      "step": 28042
    },
    {
      "epoch": 0.9600479287915098,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0915,
      "step": 28043
    },
    {
      "epoch": 0.9600821636425881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.092,
      "step": 28044
    },
    {
      "epoch": 0.9601163984936666,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0848,
      "step": 28045
    },
    {
      "epoch": 0.9601506333447449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0754,
      "step": 28046
    },
    {
      "epoch": 0.9601848681958234,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0888,
      "step": 28047
    },
    {
      "epoch": 0.9602191030469017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0955,
      "step": 28048
    },
    {
      "epoch": 0.9602533378979802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0774,
      "step": 28049
    },
    {
      "epoch": 0.9602875727490585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 28050
    },
    {
      "epoch": 0.960321807600137,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0612,
      "step": 28051
    },
    {
      "epoch": 0.9603560424512153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.077,
      "step": 28052
    },
    {
      "epoch": 0.9603902773022938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.085,
      "step": 28053
    },
    {
      "epoch": 0.9604245121533721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0788,
      "step": 28054
    },
    {
      "epoch": 0.9604587470044506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0824,
      "step": 28055
    },
    {
      "epoch": 0.9604929818555289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0956,
      "step": 28056
    },
    {
      "epoch": 0.9605272167066073,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0818,
      "step": 28057
    },
    {
      "epoch": 0.9605614515576857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0796,
      "step": 28058
    },
    {
      "epoch": 0.9605956864087641,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0843,
      "step": 28059
    },
    {
      "epoch": 0.9606299212598425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 28060
    },
    {
      "epoch": 0.9606641561109209,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0816,
      "step": 28061
    },
    {
      "epoch": 0.9606983909619993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0995,
      "step": 28062
    },
    {
      "epoch": 0.9607326258130777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0871,
      "step": 28063
    },
    {
      "epoch": 0.9607668606641561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1026,
      "step": 28064
    },
    {
      "epoch": 0.9608010955152345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0792,
      "step": 28065
    },
    {
      "epoch": 0.9608353303663129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0882,
      "step": 28066
    },
    {
      "epoch": 0.9608695652173913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0801,
      "step": 28067
    },
    {
      "epoch": 0.9609038000684696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0794,
      "step": 28068
    },
    {
      "epoch": 0.9609380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0775,
      "step": 28069
    },
    {
      "epoch": 0.9609722697706266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0872,
      "step": 28070
    },
    {
      "epoch": 0.9610065046217049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0932,
      "step": 28071
    },
    {
      "epoch": 0.9610407394727833,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0702,
      "step": 28072
    },
    {
      "epoch": 0.9610749743238617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0826,
      "step": 28073
    },
    {
      "epoch": 0.9611092091749401,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0694,
      "step": 28074
    },
    {
      "epoch": 0.9611434440260185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0621,
      "step": 28075
    },
    {
      "epoch": 0.9611776788770969,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0992,
      "step": 28076
    },
    {
      "epoch": 0.9612119137281753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0831,
      "step": 28077
    },
    {
      "epoch": 0.9612461485792537,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.088,
      "step": 28078
    },
    {
      "epoch": 0.9612803834303321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0855,
      "step": 28079
    },
    {
      "epoch": 0.9613146182814105,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 28080
    },
    {
      "epoch": 0.9613488531324889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0775,
      "step": 28081
    },
    {
      "epoch": 0.9613830879835673,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0868,
      "step": 28082
    },
    {
      "epoch": 0.9614173228346456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0842,
      "step": 28083
    },
    {
      "epoch": 0.9614515576857241,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0687,
      "step": 28084
    },
    {
      "epoch": 0.9614857925368024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0815,
      "step": 28085
    },
    {
      "epoch": 0.9615200273878809,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0858,
      "step": 28086
    },
    {
      "epoch": 0.9615542622389592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0819,
      "step": 28087
    },
    {
      "epoch": 0.9615884970900377,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0824,
      "step": 28088
    },
    {
      "epoch": 0.961622731941116,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0723,
      "step": 28089
    },
    {
      "epoch": 0.9616569667921945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0756,
      "step": 28090
    },
    {
      "epoch": 0.9616912016432728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0706,
      "step": 28091
    },
    {
      "epoch": 0.9617254364943513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0853,
      "step": 28092
    },
    {
      "epoch": 0.9617596713454296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0786,
      "step": 28093
    },
    {
      "epoch": 0.9617939061965081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.081,
      "step": 28094
    },
    {
      "epoch": 0.9618281410475864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0721,
      "step": 28095
    },
    {
      "epoch": 0.9618623758986649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0954,
      "step": 28096
    },
    {
      "epoch": 0.9618966107497432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0823,
      "step": 28097
    },
    {
      "epoch": 0.9619308456008216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.091,
      "step": 28098
    },
    {
      "epoch": 0.9619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0882,
      "step": 28099
    },
    {
      "epoch": 0.9619993153029784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0749,
      "step": 28100
    },
    {
      "epoch": 0.9620335501540568,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0771,
      "step": 28101
    },
    {
      "epoch": 0.9620677850051352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0768,
      "step": 28102
    },
    {
      "epoch": 0.9621020198562136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0836,
      "step": 28103
    },
    {
      "epoch": 0.962136254707292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0806,
      "step": 28104
    },
    {
      "epoch": 0.9621704895583704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0816,
      "step": 28105
    },
    {
      "epoch": 0.9622047244094488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1002,
      "step": 28106
    },
    {
      "epoch": 0.9622389592605273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0653,
      "step": 28107
    },
    {
      "epoch": 0.9622731941116056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.076,
      "step": 28108
    },
    {
      "epoch": 0.9623074289626841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0783,
      "step": 28109
    },
    {
      "epoch": 0.9623416638137624,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0808,
      "step": 28110
    },
    {
      "epoch": 0.9623758986648409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0739,
      "step": 28111
    },
    {
      "epoch": 0.9624101335159192,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0958,
      "step": 28112
    },
    {
      "epoch": 0.9624443683669976,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0837,
      "step": 28113
    },
    {
      "epoch": 0.962478603218076,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.092,
      "step": 28114
    },
    {
      "epoch": 0.9625128380691544,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0773,
      "step": 28115
    },
    {
      "epoch": 0.9625470729202328,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0873,
      "step": 28116
    },
    {
      "epoch": 0.9625813077713112,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0811,
      "step": 28117
    },
    {
      "epoch": 0.9626155426223896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0753,
      "step": 28118
    },
    {
      "epoch": 0.962649777473468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0833,
      "step": 28119
    },
    {
      "epoch": 0.9626840123245464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0771,
      "step": 28120
    },
    {
      "epoch": 0.9627182471756248,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0745,
      "step": 28121
    },
    {
      "epoch": 0.9627524820267032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0832,
      "step": 28122
    },
    {
      "epoch": 0.9627867168777816,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0723,
      "step": 28123
    },
    {
      "epoch": 0.96282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0915,
      "step": 28124
    },
    {
      "epoch": 0.9628551865799384,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0795,
      "step": 28125
    },
    {
      "epoch": 0.9628894214310167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1051,
      "step": 28126
    },
    {
      "epoch": 0.9629236562820952,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0753,
      "step": 28127
    },
    {
      "epoch": 0.9629578911331735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0802,
      "step": 28128
    },
    {
      "epoch": 0.962992125984252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0767,
      "step": 28129
    },
    {
      "epoch": 0.9630263608353303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0934,
      "step": 28130
    },
    {
      "epoch": 0.9630605956864088,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0724,
      "step": 28131
    },
    {
      "epoch": 0.9630948305374871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0767,
      "step": 28132
    },
    {
      "epoch": 0.9631290653885656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0683,
      "step": 28133
    },
    {
      "epoch": 0.9631633002396439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0749,
      "step": 28134
    },
    {
      "epoch": 0.9631975350907224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0717,
      "step": 28135
    },
    {
      "epoch": 0.9632317699418007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.104,
      "step": 28136
    },
    {
      "epoch": 0.9632660047928792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0709,
      "step": 28137
    },
    {
      "epoch": 0.9633002396439575,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0754,
      "step": 28138
    },
    {
      "epoch": 0.963334474495036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.084,
      "step": 28139
    },
    {
      "epoch": 0.9633687093461143,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 28140
    },
    {
      "epoch": 0.9634029441971927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0738,
      "step": 28141
    },
    {
      "epoch": 0.9634371790482711,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0883,
      "step": 28142
    },
    {
      "epoch": 0.9634714138993495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0666,
      "step": 28143
    },
    {
      "epoch": 0.963505648750428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0687,
      "step": 28144
    },
    {
      "epoch": 0.9635398836015063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0756,
      "step": 28145
    },
    {
      "epoch": 0.9635741184525848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0859,
      "step": 28146
    },
    {
      "epoch": 0.9636083533036631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0711,
      "step": 28147
    },
    {
      "epoch": 0.9636425881547416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0806,
      "step": 28148
    },
    {
      "epoch": 0.9636768230058199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0987,
      "step": 28149
    },
    {
      "epoch": 0.9637110578568984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0919,
      "step": 28150
    },
    {
      "epoch": 0.9637452927079767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0703,
      "step": 28151
    },
    {
      "epoch": 0.9637795275590552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0923,
      "step": 28152
    },
    {
      "epoch": 0.9638137624101335,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0822,
      "step": 28153
    },
    {
      "epoch": 0.963847997261212,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0876,
      "step": 28154
    },
    {
      "epoch": 0.9638822321122903,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0854,
      "step": 28155
    },
    {
      "epoch": 0.9639164669633687,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0709,
      "step": 28156
    },
    {
      "epoch": 0.9639507018144471,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.092,
      "step": 28157
    },
    {
      "epoch": 0.9639849366655255,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0863,
      "step": 28158
    },
    {
      "epoch": 0.9640191715166039,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0881,
      "step": 28159
    },
    {
      "epoch": 0.9640534063676823,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0815,
      "step": 28160
    },
    {
      "epoch": 0.9640876412187607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0814,
      "step": 28161
    },
    {
      "epoch": 0.9641218760698391,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0855,
      "step": 28162
    },
    {
      "epoch": 0.9641561109209175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0892,
      "step": 28163
    },
    {
      "epoch": 0.9641903457719959,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0806,
      "step": 28164
    },
    {
      "epoch": 0.9642245806230743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0945,
      "step": 28165
    },
    {
      "epoch": 0.9642588154741527,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0722,
      "step": 28166
    },
    {
      "epoch": 0.964293050325231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0867,
      "step": 28167
    },
    {
      "epoch": 0.9643272851763095,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0852,
      "step": 28168
    },
    {
      "epoch": 0.9643615200273878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0703,
      "step": 28169
    },
    {
      "epoch": 0.9643957548784663,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0817,
      "step": 28170
    },
    {
      "epoch": 0.9644299897295446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0765,
      "step": 28171
    },
    {
      "epoch": 0.9644642245806231,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0789,
      "step": 28172
    },
    {
      "epoch": 0.9644984594317014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0901,
      "step": 28173
    },
    {
      "epoch": 0.9645326942827799,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0824,
      "step": 28174
    },
    {
      "epoch": 0.9645669291338582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0851,
      "step": 28175
    },
    {
      "epoch": 0.9646011639849367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.07,
      "step": 28176
    },
    {
      "epoch": 0.964635398836015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.079,
      "step": 28177
    },
    {
      "epoch": 0.9646696336870935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0778,
      "step": 28178
    },
    {
      "epoch": 0.9647038685381718,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0732,
      "step": 28179
    },
    {
      "epoch": 0.9647381033892503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0828,
      "step": 28180
    },
    {
      "epoch": 0.9647723382403287,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0769,
      "step": 28181
    },
    {
      "epoch": 0.964806573091407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0815,
      "step": 28182
    },
    {
      "epoch": 0.9648408079424855,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0768,
      "step": 28183
    },
    {
      "epoch": 0.9648750427935638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.097,
      "step": 28184
    },
    {
      "epoch": 0.9649092776446423,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0989,
      "step": 28185
    },
    {
      "epoch": 0.9649435124957206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0821,
      "step": 28186
    },
    {
      "epoch": 0.9649777473467991,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0851,
      "step": 28187
    },
    {
      "epoch": 0.9650119821978774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0808,
      "step": 28188
    },
    {
      "epoch": 0.9650462170489559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0773,
      "step": 28189
    },
    {
      "epoch": 0.9650804519000342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1036,
      "step": 28190
    },
    {
      "epoch": 0.9651146867511127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 28191
    },
    {
      "epoch": 0.965148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0854,
      "step": 28192
    },
    {
      "epoch": 0.9651831564532695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0718,
      "step": 28193
    },
    {
      "epoch": 0.9652173913043478,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0783,
      "step": 28194
    },
    {
      "epoch": 0.9652516261554263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 28195
    },
    {
      "epoch": 0.9652858610065046,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0756,
      "step": 28196
    },
    {
      "epoch": 0.965320095857583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0846,
      "step": 28197
    },
    {
      "epoch": 0.9653543307086614,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 28198
    },
    {
      "epoch": 0.9653885655597398,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0841,
      "step": 28199
    },
    {
      "epoch": 0.9654228004108182,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 28200
    },
    {
      "epoch": 0.9654570352618966,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0802,
      "step": 28201
    },
    {
      "epoch": 0.965491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0839,
      "step": 28202
    },
    {
      "epoch": 0.9655255049640534,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0628,
      "step": 28203
    },
    {
      "epoch": 0.9655597398151318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0704,
      "step": 28204
    },
    {
      "epoch": 0.9655939746662102,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0791,
      "step": 28205
    },
    {
      "epoch": 0.9656282095172886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1001,
      "step": 28206
    },
    {
      "epoch": 0.965662444368367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0852,
      "step": 28207
    },
    {
      "epoch": 0.9656966792194454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0862,
      "step": 28208
    },
    {
      "epoch": 0.9657309140705238,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.084,
      "step": 28209
    },
    {
      "epoch": 0.9657651489216021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 28210
    },
    {
      "epoch": 0.9657993837726806,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0815,
      "step": 28211
    },
    {
      "epoch": 0.9658336186237589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0811,
      "step": 28212
    },
    {
      "epoch": 0.9658678534748374,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 28213
    },
    {
      "epoch": 0.9659020883259157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0738,
      "step": 28214
    },
    {
      "epoch": 0.9659363231769942,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0898,
      "step": 28215
    },
    {
      "epoch": 0.9659705580280725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0892,
      "step": 28216
    },
    {
      "epoch": 0.966004792879151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.067,
      "step": 28217
    },
    {
      "epoch": 0.9660390277302294,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 28218
    },
    {
      "epoch": 0.9660732625813078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0631,
      "step": 28219
    },
    {
      "epoch": 0.9661074974323862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0766,
      "step": 28220
    },
    {
      "epoch": 0.9661417322834646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 28221
    },
    {
      "epoch": 0.966175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0835,
      "step": 28222
    },
    {
      "epoch": 0.9662102019856214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0699,
      "step": 28223
    },
    {
      "epoch": 0.9662444368366998,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0841,
      "step": 28224
    },
    {
      "epoch": 0.9662786716877781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.075,
      "step": 28225
    },
    {
      "epoch": 0.9663129065388566,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0853,
      "step": 28226
    },
    {
      "epoch": 0.9663471413899349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0861,
      "step": 28227
    },
    {
      "epoch": 0.9663813762410134,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0909,
      "step": 28228
    },
    {
      "epoch": 0.9664156110920917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0892,
      "step": 28229
    },
    {
      "epoch": 0.9664498459431702,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0833,
      "step": 28230
    },
    {
      "epoch": 0.9664840807942485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 28231
    },
    {
      "epoch": 0.966518315645327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0864,
      "step": 28232
    },
    {
      "epoch": 0.9665525504964053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0838,
      "step": 28233
    },
    {
      "epoch": 0.9665867853474838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0794,
      "step": 28234
    },
    {
      "epoch": 0.9666210201985621,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0822,
      "step": 28235
    },
    {
      "epoch": 0.9666552550496406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0768,
      "step": 28236
    },
    {
      "epoch": 0.9666894899007189,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0848,
      "step": 28237
    },
    {
      "epoch": 0.9667237247517974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0831,
      "step": 28238
    },
    {
      "epoch": 0.9667579596028757,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0934,
      "step": 28239
    },
    {
      "epoch": 0.9667921944539541,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0721,
      "step": 28240
    },
    {
      "epoch": 0.9668264293050325,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1007,
      "step": 28241
    },
    {
      "epoch": 0.9668606641561109,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0866,
      "step": 28242
    },
    {
      "epoch": 0.9668948990071893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0832,
      "step": 28243
    },
    {
      "epoch": 0.9669291338582677,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0861,
      "step": 28244
    },
    {
      "epoch": 0.9669633687093461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 28245
    },
    {
      "epoch": 0.9669976035604245,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0796,
      "step": 28246
    },
    {
      "epoch": 0.9670318384115029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0756,
      "step": 28247
    },
    {
      "epoch": 0.9670660732625813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0808,
      "step": 28248
    },
    {
      "epoch": 0.9671003081136597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0847,
      "step": 28249
    },
    {
      "epoch": 0.9671345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0699,
      "step": 28250
    },
    {
      "epoch": 0.9671687778158164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0744,
      "step": 28251
    },
    {
      "epoch": 0.9672030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0818,
      "step": 28252
    },
    {
      "epoch": 0.9672372475179732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0596,
      "step": 28253
    },
    {
      "epoch": 0.9672714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0721,
      "step": 28254
    },
    {
      "epoch": 0.9673057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0794,
      "step": 28255
    },
    {
      "epoch": 0.9673399520712085,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0745,
      "step": 28256
    },
    {
      "epoch": 0.9673741869222869,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0919,
      "step": 28257
    },
    {
      "epoch": 0.9674084217733653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0753,
      "step": 28258
    },
    {
      "epoch": 0.9674426566244437,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 28259
    },
    {
      "epoch": 0.9674768914755221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0933,
      "step": 28260
    },
    {
      "epoch": 0.9675111263266005,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0691,
      "step": 28261
    },
    {
      "epoch": 0.9675453611776789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0659,
      "step": 28262
    },
    {
      "epoch": 0.9675795960287573,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0819,
      "step": 28263
    },
    {
      "epoch": 0.9676138308798357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0836,
      "step": 28264
    },
    {
      "epoch": 0.9676480657309141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 28265
    },
    {
      "epoch": 0.9676823005819924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0753,
      "step": 28266
    },
    {
      "epoch": 0.9677165354330709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0912,
      "step": 28267
    },
    {
      "epoch": 0.9677507702841492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0788,
      "step": 28268
    },
    {
      "epoch": 0.9677850051352277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0755,
      "step": 28269
    },
    {
      "epoch": 0.967819239986306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0972,
      "step": 28270
    },
    {
      "epoch": 0.9678534748373845,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0815,
      "step": 28271
    },
    {
      "epoch": 0.9678877096884628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0877,
      "step": 28272
    },
    {
      "epoch": 0.9679219445395413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0722,
      "step": 28273
    },
    {
      "epoch": 0.9679561793906196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0744,
      "step": 28274
    },
    {
      "epoch": 0.9679904142416981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0959,
      "step": 28275
    },
    {
      "epoch": 0.9680246490927764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0859,
      "step": 28276
    },
    {
      "epoch": 0.9680588839438549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.071,
      "step": 28277
    },
    {
      "epoch": 0.9680931187949332,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0741,
      "step": 28278
    },
    {
      "epoch": 0.9681273536460117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0714,
      "step": 28279
    },
    {
      "epoch": 0.96816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0842,
      "step": 28280
    },
    {
      "epoch": 0.9681958233481684,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0751,
      "step": 28281
    },
    {
      "epoch": 0.9682300581992468,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0732,
      "step": 28282
    },
    {
      "epoch": 0.9682642930503252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0811,
      "step": 28283
    },
    {
      "epoch": 0.9682985279014036,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0901,
      "step": 28284
    },
    {
      "epoch": 0.968332762752482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0737,
      "step": 28285
    },
    {
      "epoch": 0.9683669976035604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0802,
      "step": 28286
    },
    {
      "epoch": 0.9684012324546388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0881,
      "step": 28287
    },
    {
      "epoch": 0.9684354673057172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0911,
      "step": 28288
    },
    {
      "epoch": 0.9684697021567956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0847,
      "step": 28289
    },
    {
      "epoch": 0.968503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0843,
      "step": 28290
    },
    {
      "epoch": 0.9685381718589524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0711,
      "step": 28291
    },
    {
      "epoch": 0.9685724067100309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0968,
      "step": 28292
    },
    {
      "epoch": 0.9686066415611092,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0909,
      "step": 28293
    },
    {
      "epoch": 0.9686408764121877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0815,
      "step": 28294
    },
    {
      "epoch": 0.968675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0979,
      "step": 28295
    },
    {
      "epoch": 0.9687093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0933,
      "step": 28296
    },
    {
      "epoch": 0.9687435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0748,
      "step": 28297
    },
    {
      "epoch": 0.9687778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0891,
      "step": 28298
    },
    {
      "epoch": 0.9688120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0733,
      "step": 28299
    },
    {
      "epoch": 0.968846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0812,
      "step": 28300
    },
    {
      "epoch": 0.9688805203697364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0895,
      "step": 28301
    },
    {
      "epoch": 0.9689147552208148,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0774,
      "step": 28302
    },
    {
      "epoch": 0.9689489900718932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0662,
      "step": 28303
    },
    {
      "epoch": 0.9689832249229716,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0739,
      "step": 28304
    },
    {
      "epoch": 0.96901745977405,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.077,
      "step": 28305
    },
    {
      "epoch": 0.9690516946251284,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0878,
      "step": 28306
    },
    {
      "epoch": 0.9690859294762068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0923,
      "step": 28307
    },
    {
      "epoch": 0.9691201643272852,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.076,
      "step": 28308
    },
    {
      "epoch": 0.9691543991783635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0569,
      "step": 28309
    },
    {
      "epoch": 0.969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 28310
    },
    {
      "epoch": 0.9692228688805203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0803,
      "step": 28311
    },
    {
      "epoch": 0.9692571037315988,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0778,
      "step": 28312
    },
    {
      "epoch": 0.9692913385826771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0594,
      "step": 28313
    },
    {
      "epoch": 0.9693255734337556,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0685,
      "step": 28314
    },
    {
      "epoch": 0.9693598082848339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0852,
      "step": 28315
    },
    {
      "epoch": 0.9693940431359124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0816,
      "step": 28316
    },
    {
      "epoch": 0.9694282779869907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0712,
      "step": 28317
    },
    {
      "epoch": 0.9694625128380692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0796,
      "step": 28318
    },
    {
      "epoch": 0.9694967476891475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0821,
      "step": 28319
    },
    {
      "epoch": 0.969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0856,
      "step": 28320
    },
    {
      "epoch": 0.9695652173913043,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0954,
      "step": 28321
    },
    {
      "epoch": 0.9695994522423828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0962,
      "step": 28322
    },
    {
      "epoch": 0.9696336870934611,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0843,
      "step": 28323
    },
    {
      "epoch": 0.9696679219445395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0542,
      "step": 28324
    },
    {
      "epoch": 0.9697021567956179,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0747,
      "step": 28325
    },
    {
      "epoch": 0.9697363916466963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0959,
      "step": 28326
    },
    {
      "epoch": 0.9697706264977747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.086,
      "step": 28327
    },
    {
      "epoch": 0.9698048613488531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0783,
      "step": 28328
    },
    {
      "epoch": 0.9698390961999316,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0956,
      "step": 28329
    },
    {
      "epoch": 0.9698733310510099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.089,
      "step": 28330
    },
    {
      "epoch": 0.9699075659020884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0703,
      "step": 28331
    },
    {
      "epoch": 0.9699418007531667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0806,
      "step": 28332
    },
    {
      "epoch": 0.9699760356042452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0756,
      "step": 28333
    },
    {
      "epoch": 0.9700102704553235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0721,
      "step": 28334
    },
    {
      "epoch": 0.970044505306402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0625,
      "step": 28335
    },
    {
      "epoch": 0.9700787401574803,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0818,
      "step": 28336
    },
    {
      "epoch": 0.9701129750085588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 28337
    },
    {
      "epoch": 0.9701472098596371,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0843,
      "step": 28338
    },
    {
      "epoch": 0.9701814447107155,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0709,
      "step": 28339
    },
    {
      "epoch": 0.9702156795617939,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0885,
      "step": 28340
    },
    {
      "epoch": 0.9702499144128723,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0852,
      "step": 28341
    },
    {
      "epoch": 0.9702841492639507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.073,
      "step": 28342
    },
    {
      "epoch": 0.9703183841150291,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0813,
      "step": 28343
    },
    {
      "epoch": 0.9703526189661075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0908,
      "step": 28344
    },
    {
      "epoch": 0.9703868538171859,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1003,
      "step": 28345
    },
    {
      "epoch": 0.9704210886682643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0789,
      "step": 28346
    },
    {
      "epoch": 0.9704553235193427,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0822,
      "step": 28347
    },
    {
      "epoch": 0.9704895583704211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0824,
      "step": 28348
    },
    {
      "epoch": 0.9705237932214995,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0734,
      "step": 28349
    },
    {
      "epoch": 0.9705580280725778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0839,
      "step": 28350
    },
    {
      "epoch": 0.9705922629236563,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0791,
      "step": 28351
    },
    {
      "epoch": 0.9706264977747346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 28352
    },
    {
      "epoch": 0.9706607326258131,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0729,
      "step": 28353
    },
    {
      "epoch": 0.9706949674768914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0851,
      "step": 28354
    },
    {
      "epoch": 0.9707292023279699,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0815,
      "step": 28355
    },
    {
      "epoch": 0.9707634371790482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0537,
      "step": 28356
    },
    {
      "epoch": 0.9707976720301267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0704,
      "step": 28357
    },
    {
      "epoch": 0.970831906881205,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0781,
      "step": 28358
    },
    {
      "epoch": 0.9708661417322835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0854,
      "step": 28359
    },
    {
      "epoch": 0.9709003765833618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0765,
      "step": 28360
    },
    {
      "epoch": 0.9709346114344403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0732,
      "step": 28361
    },
    {
      "epoch": 0.9709688462855186,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1005,
      "step": 28362
    },
    {
      "epoch": 0.9710030811365971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0734,
      "step": 28363
    },
    {
      "epoch": 0.9710373159876755,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0708,
      "step": 28364
    },
    {
      "epoch": 0.9710715508387538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0725,
      "step": 28365
    },
    {
      "epoch": 0.9711057856898323,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0953,
      "step": 28366
    },
    {
      "epoch": 0.9711400205409106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0807,
      "step": 28367
    },
    {
      "epoch": 0.9711742553919891,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0929,
      "step": 28368
    },
    {
      "epoch": 0.9712084902430674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.082,
      "step": 28369
    },
    {
      "epoch": 0.9712427250941459,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0834,
      "step": 28370
    },
    {
      "epoch": 0.9712769599452242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0765,
      "step": 28371
    },
    {
      "epoch": 0.9713111947963027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0775,
      "step": 28372
    },
    {
      "epoch": 0.971345429647381,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0773,
      "step": 28373
    },
    {
      "epoch": 0.9713796644984595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0814,
      "step": 28374
    },
    {
      "epoch": 0.9714138993495378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0886,
      "step": 28375
    },
    {
      "epoch": 0.9714481342006163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0697,
      "step": 28376
    },
    {
      "epoch": 0.9714823690516946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0728,
      "step": 28377
    },
    {
      "epoch": 0.9715166039027731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0787,
      "step": 28378
    },
    {
      "epoch": 0.9715508387538514,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0945,
      "step": 28379
    },
    {
      "epoch": 0.9715850736049298,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0936,
      "step": 28380
    },
    {
      "epoch": 0.9716193084560082,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0843,
      "step": 28381
    },
    {
      "epoch": 0.9716535433070866,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0955,
      "step": 28382
    },
    {
      "epoch": 0.971687778158165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0876,
      "step": 28383
    },
    {
      "epoch": 0.9717220130092434,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.094,
      "step": 28384
    },
    {
      "epoch": 0.9717562478603218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0725,
      "step": 28385
    },
    {
      "epoch": 0.9717904827114002,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0801,
      "step": 28386
    },
    {
      "epoch": 0.9718247175624786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0748,
      "step": 28387
    },
    {
      "epoch": 0.971858952413557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 28388
    },
    {
      "epoch": 0.9718931872646354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0777,
      "step": 28389
    },
    {
      "epoch": 0.9719274221157138,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0691,
      "step": 28390
    },
    {
      "epoch": 0.9719616569667922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0674,
      "step": 28391
    },
    {
      "epoch": 0.9719958918178706,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 28392
    },
    {
      "epoch": 0.972030126668949,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0863,
      "step": 28393
    },
    {
      "epoch": 0.9720643615200274,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0736,
      "step": 28394
    },
    {
      "epoch": 0.9720985963711057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0827,
      "step": 28395
    },
    {
      "epoch": 0.9721328312221842,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0677,
      "step": 28396
    },
    {
      "epoch": 0.9721670660732625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0818,
      "step": 28397
    },
    {
      "epoch": 0.972201300924341,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0731,
      "step": 28398
    },
    {
      "epoch": 0.9722355357754193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0774,
      "step": 28399
    },
    {
      "epoch": 0.9722697706264978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.083,
      "step": 28400
    },
    {
      "epoch": 0.9723040054775762,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0739,
      "step": 28401
    },
    {
      "epoch": 0.9723382403286546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0706,
      "step": 28402
    },
    {
      "epoch": 0.972372475179733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0735,
      "step": 28403
    },
    {
      "epoch": 0.9724067100308114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0773,
      "step": 28404
    },
    {
      "epoch": 0.9724409448818898,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 28405
    },
    {
      "epoch": 0.9724751797329682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0745,
      "step": 28406
    },
    {
      "epoch": 0.9725094145840466,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0916,
      "step": 28407
    },
    {
      "epoch": 0.972543649435125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0695,
      "step": 28408
    },
    {
      "epoch": 0.9725778842862034,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0861,
      "step": 28409
    },
    {
      "epoch": 0.9726121191372817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 28410
    },
    {
      "epoch": 0.9726463539883602,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.067,
      "step": 28411
    },
    {
      "epoch": 0.9726805888394385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1039,
      "step": 28412
    },
    {
      "epoch": 0.972714823690517,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0858,
      "step": 28413
    },
    {
      "epoch": 0.9727490585415953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0601,
      "step": 28414
    },
    {
      "epoch": 0.9727832933926738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0751,
      "step": 28415
    },
    {
      "epoch": 0.9728175282437521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0888,
      "step": 28416
    },
    {
      "epoch": 0.9728517630948306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 28417
    },
    {
      "epoch": 0.9728859979459089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0841,
      "step": 28418
    },
    {
      "epoch": 0.9729202327969874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0744,
      "step": 28419
    },
    {
      "epoch": 0.9729544676480657,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0852,
      "step": 28420
    },
    {
      "epoch": 0.9729887024991442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0755,
      "step": 28421
    },
    {
      "epoch": 0.9730229373502225,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0833,
      "step": 28422
    },
    {
      "epoch": 0.973057172201301,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0879,
      "step": 28423
    },
    {
      "epoch": 0.9730914070523793,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0816,
      "step": 28424
    },
    {
      "epoch": 0.9731256419034577,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0626,
      "step": 28425
    },
    {
      "epoch": 0.9731598767545361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0738,
      "step": 28426
    },
    {
      "epoch": 0.9731941116056145,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0783,
      "step": 28427
    },
    {
      "epoch": 0.9732283464566929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0764,
      "step": 28428
    },
    {
      "epoch": 0.9732625813077713,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0763,
      "step": 28429
    },
    {
      "epoch": 0.9732968161588497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 28430
    },
    {
      "epoch": 0.9733310510099281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0601,
      "step": 28431
    },
    {
      "epoch": 0.9733652858610065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0498,
      "step": 28432
    },
    {
      "epoch": 0.9733995207120849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0621,
      "step": 28433
    },
    {
      "epoch": 0.9734337555631632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0647,
      "step": 28434
    },
    {
      "epoch": 0.9734679904142417,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.065,
      "step": 28435
    },
    {
      "epoch": 0.97350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0762,
      "step": 28436
    },
    {
      "epoch": 0.9735364601163985,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0729,
      "step": 28437
    },
    {
      "epoch": 0.973570694967477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0817,
      "step": 28438
    },
    {
      "epoch": 0.9736049298185553,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0819,
      "step": 28439
    },
    {
      "epoch": 0.9736391646696337,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.085,
      "step": 28440
    },
    {
      "epoch": 0.9736733995207121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0823,
      "step": 28441
    },
    {
      "epoch": 0.9737076343717905,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0661,
      "step": 28442
    },
    {
      "epoch": 0.9737418692228689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0715,
      "step": 28443
    },
    {
      "epoch": 0.9737761040739473,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0737,
      "step": 28444
    },
    {
      "epoch": 0.9738103389250257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0818,
      "step": 28445
    },
    {
      "epoch": 0.9738445737761041,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0795,
      "step": 28446
    },
    {
      "epoch": 0.9738788086271825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0878,
      "step": 28447
    },
    {
      "epoch": 0.9739130434782609,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.079,
      "step": 28448
    },
    {
      "epoch": 0.9739472783293392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0924,
      "step": 28449
    },
    {
      "epoch": 0.9739815131804177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0845,
      "step": 28450
    },
    {
      "epoch": 0.974015748031496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0849,
      "step": 28451
    },
    {
      "epoch": 0.9740499828825745,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.086,
      "step": 28452
    },
    {
      "epoch": 0.9740842177336528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0691,
      "step": 28453
    },
    {
      "epoch": 0.9741184525847313,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0851,
      "step": 28454
    },
    {
      "epoch": 0.9741526874358096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0623,
      "step": 28455
    },
    {
      "epoch": 0.9741869222868881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0883,
      "step": 28456
    },
    {
      "epoch": 0.9742211571379664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0772,
      "step": 28457
    },
    {
      "epoch": 0.9742553919890449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0608,
      "step": 28458
    },
    {
      "epoch": 0.9742896268401232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.071,
      "step": 28459
    },
    {
      "epoch": 0.9743238616912017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0812,
      "step": 28460
    },
    {
      "epoch": 0.97435809654228,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0833,
      "step": 28461
    },
    {
      "epoch": 0.9743923313933585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0771,
      "step": 28462
    },
    {
      "epoch": 0.9744265662444368,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0839,
      "step": 28463
    },
    {
      "epoch": 0.9744608010955152,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0767,
      "step": 28464
    },
    {
      "epoch": 0.9744950359465936,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0688,
      "step": 28465
    },
    {
      "epoch": 0.974529270797672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0606,
      "step": 28466
    },
    {
      "epoch": 0.9745635056487504,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0829,
      "step": 28467
    },
    {
      "epoch": 0.9745977404998288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0851,
      "step": 28468
    },
    {
      "epoch": 0.9746319753509072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0856,
      "step": 28469
    },
    {
      "epoch": 0.9746662102019856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 28470
    },
    {
      "epoch": 0.974700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 28471
    },
    {
      "epoch": 0.9747346799041424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0712,
      "step": 28472
    },
    {
      "epoch": 0.9747689147552208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0828,
      "step": 28473
    },
    {
      "epoch": 0.9748031496062992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0758,
      "step": 28474
    },
    {
      "epoch": 0.9748373844573777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0883,
      "step": 28475
    },
    {
      "epoch": 0.974871619308456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0764,
      "step": 28476
    },
    {
      "epoch": 0.9749058541595345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0913,
      "step": 28477
    },
    {
      "epoch": 0.9749400890106128,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0735,
      "step": 28478
    },
    {
      "epoch": 0.9749743238616912,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0603,
      "step": 28479
    },
    {
      "epoch": 0.9750085587127696,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0918,
      "step": 28480
    },
    {
      "epoch": 0.975042793563848,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0678,
      "step": 28481
    },
    {
      "epoch": 0.9750770284149264,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0818,
      "step": 28482
    },
    {
      "epoch": 0.9751112632660048,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0706,
      "step": 28483
    },
    {
      "epoch": 0.9751454981170832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0811,
      "step": 28484
    },
    {
      "epoch": 0.9751797329681616,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0935,
      "step": 28485
    },
    {
      "epoch": 0.97521396781924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 28486
    },
    {
      "epoch": 0.9752482026703184,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0829,
      "step": 28487
    },
    {
      "epoch": 0.9752824375213968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0624,
      "step": 28488
    },
    {
      "epoch": 0.9753166723724752,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.081,
      "step": 28489
    },
    {
      "epoch": 0.9753509072235536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0862,
      "step": 28490
    },
    {
      "epoch": 0.975385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0725,
      "step": 28491
    },
    {
      "epoch": 0.9754193769257103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0626,
      "step": 28492
    },
    {
      "epoch": 0.9754536117767888,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0717,
      "step": 28493
    },
    {
      "epoch": 0.9754878466278671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0885,
      "step": 28494
    },
    {
      "epoch": 0.9755220814789456,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0849,
      "step": 28495
    },
    {
      "epoch": 0.9755563163300239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0659,
      "step": 28496
    },
    {
      "epoch": 0.9755905511811024,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0788,
      "step": 28497
    },
    {
      "epoch": 0.9756247860321807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 28498
    },
    {
      "epoch": 0.9756590208832592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0775,
      "step": 28499
    },
    {
      "epoch": 0.9756932557343375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0583,
      "step": 28500
    },
    {
      "epoch": 0.975727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0674,
      "step": 28501
    },
    {
      "epoch": 0.9757617254364943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0758,
      "step": 28502
    },
    {
      "epoch": 0.9757959602875728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0848,
      "step": 28503
    },
    {
      "epoch": 0.9758301951386511,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0758,
      "step": 28504
    },
    {
      "epoch": 0.9758644299897296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0948,
      "step": 28505
    },
    {
      "epoch": 0.9758986648408079,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0758,
      "step": 28506
    },
    {
      "epoch": 0.9759328996918863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0666,
      "step": 28507
    },
    {
      "epoch": 0.9759671345429647,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0525,
      "step": 28508
    },
    {
      "epoch": 0.9760013693940431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0708,
      "step": 28509
    },
    {
      "epoch": 0.9760356042451215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0831,
      "step": 28510
    },
    {
      "epoch": 0.9760698390961999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0841,
      "step": 28511
    },
    {
      "epoch": 0.9761040739472784,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.051,
      "step": 28512
    },
    {
      "epoch": 0.9761383087983567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0653,
      "step": 28513
    },
    {
      "epoch": 0.9761725436494352,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0559,
      "step": 28514
    },
    {
      "epoch": 0.9762067785005135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0966,
      "step": 28515
    },
    {
      "epoch": 0.976241013351592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0713,
      "step": 28516
    },
    {
      "epoch": 0.9762752482026703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 28517
    },
    {
      "epoch": 0.9763094830537488,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0835,
      "step": 28518
    },
    {
      "epoch": 0.9763437179048271,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0662,
      "step": 28519
    },
    {
      "epoch": 0.9763779527559056,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0688,
      "step": 28520
    },
    {
      "epoch": 0.9764121876069839,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0667,
      "step": 28521
    },
    {
      "epoch": 0.9764464224580623,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0696,
      "step": 28522
    },
    {
      "epoch": 0.9764806573091407,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0835,
      "step": 28523
    },
    {
      "epoch": 0.9765148921602191,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0832,
      "step": 28524
    },
    {
      "epoch": 0.9765491270112975,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0624,
      "step": 28525
    },
    {
      "epoch": 0.9765833618623759,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0739,
      "step": 28526
    },
    {
      "epoch": 0.9766175967134543,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0635,
      "step": 28527
    },
    {
      "epoch": 0.9766518315645327,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0645,
      "step": 28528
    },
    {
      "epoch": 0.9766860664156111,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.067,
      "step": 28529
    },
    {
      "epoch": 0.9767203012666895,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0662,
      "step": 28530
    },
    {
      "epoch": 0.9767545361177679,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0716,
      "step": 28531
    },
    {
      "epoch": 0.9767887709688463,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0798,
      "step": 28532
    },
    {
      "epoch": 0.9768230058199247,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0938,
      "step": 28533
    },
    {
      "epoch": 0.9768572406710031,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0663,
      "step": 28534
    },
    {
      "epoch": 0.9768914755220814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0796,
      "step": 28535
    },
    {
      "epoch": 0.9769257103731599,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0709,
      "step": 28536
    },
    {
      "epoch": 0.9769599452242382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0794,
      "step": 28537
    },
    {
      "epoch": 0.9769941800753167,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.077,
      "step": 28538
    },
    {
      "epoch": 0.977028414926395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0668,
      "step": 28539
    },
    {
      "epoch": 0.9770626497774735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0763,
      "step": 28540
    },
    {
      "epoch": 0.9770968846285518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0872,
      "step": 28541
    },
    {
      "epoch": 0.9771311194796303,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.08,
      "step": 28542
    },
    {
      "epoch": 0.9771653543307086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0696,
      "step": 28543
    },
    {
      "epoch": 0.9771995891817871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0628,
      "step": 28544
    },
    {
      "epoch": 0.9772338240328654,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0743,
      "step": 28545
    },
    {
      "epoch": 0.9772680588839439,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0711,
      "step": 28546
    },
    {
      "epoch": 0.9773022937350222,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0681,
      "step": 28547
    },
    {
      "epoch": 0.9773365285861007,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0866,
      "step": 28548
    },
    {
      "epoch": 0.9773707634371791,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0671,
      "step": 28549
    },
    {
      "epoch": 0.9774049982882574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0623,
      "step": 28550
    },
    {
      "epoch": 0.9774392331393359,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0732,
      "step": 28551
    },
    {
      "epoch": 0.9774734679904142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0786,
      "step": 28552
    },
    {
      "epoch": 0.9775077028414927,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0691,
      "step": 28553
    },
    {
      "epoch": 0.977541937692571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.082,
      "step": 28554
    },
    {
      "epoch": 0.9775761725436495,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0876,
      "step": 28555
    },
    {
      "epoch": 0.9776104073947278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0866,
      "step": 28556
    },
    {
      "epoch": 0.9776446422458063,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0903,
      "step": 28557
    },
    {
      "epoch": 0.9776788770968846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0697,
      "step": 28558
    },
    {
      "epoch": 0.9777131119479631,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0856,
      "step": 28559
    },
    {
      "epoch": 0.9777473467990414,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0864,
      "step": 28560
    },
    {
      "epoch": 0.9777815816501199,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0889,
      "step": 28561
    },
    {
      "epoch": 0.9778158165011982,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0697,
      "step": 28562
    },
    {
      "epoch": 0.9778500513522767,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0853,
      "step": 28563
    },
    {
      "epoch": 0.977884286203355,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0714,
      "step": 28564
    },
    {
      "epoch": 0.9779185210544334,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0716,
      "step": 28565
    },
    {
      "epoch": 0.9779527559055118,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.048,
      "step": 28566
    },
    {
      "epoch": 0.9779869907565902,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0548,
      "step": 28567
    },
    {
      "epoch": 0.9780212256076686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0717,
      "step": 28568
    },
    {
      "epoch": 0.978055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0735,
      "step": 28569
    },
    {
      "epoch": 0.9780896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0895,
      "step": 28570
    },
    {
      "epoch": 0.9781239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0652,
      "step": 28571
    },
    {
      "epoch": 0.9781581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0674,
      "step": 28572
    },
    {
      "epoch": 0.9781923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0714,
      "step": 28573
    },
    {
      "epoch": 0.978226634714139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0798,
      "step": 28574
    },
    {
      "epoch": 0.9782608695652174,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0697,
      "step": 28575
    },
    {
      "epoch": 0.9782951044162957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0747,
      "step": 28576
    },
    {
      "epoch": 0.9783293392673742,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0737,
      "step": 28577
    },
    {
      "epoch": 0.9783635741184525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 28578
    },
    {
      "epoch": 0.978397808969531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0739,
      "step": 28579
    },
    {
      "epoch": 0.9784320438206093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0733,
      "step": 28580
    },
    {
      "epoch": 0.9784662786716878,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0876,
      "step": 28581
    },
    {
      "epoch": 0.9785005135227661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0766,
      "step": 28582
    },
    {
      "epoch": 0.9785347483738446,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0929,
      "step": 28583
    },
    {
      "epoch": 0.9785689832249229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 28584
    },
    {
      "epoch": 0.9786032180760014,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0774,
      "step": 28585
    },
    {
      "epoch": 0.9786374529270798,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0823,
      "step": 28586
    },
    {
      "epoch": 0.9786716877781582,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0681,
      "step": 28587
    },
    {
      "epoch": 0.9787059226292366,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0815,
      "step": 28588
    },
    {
      "epoch": 0.978740157480315,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0844,
      "step": 28589
    },
    {
      "epoch": 0.9787743923313934,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0742,
      "step": 28590
    },
    {
      "epoch": 0.9788086271824717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0778,
      "step": 28591
    },
    {
      "epoch": 0.9788428620335502,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0698,
      "step": 28592
    },
    {
      "epoch": 0.9788770968846285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0777,
      "step": 28593
    },
    {
      "epoch": 0.978911331735707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0813,
      "step": 28594
    },
    {
      "epoch": 0.9789455665867853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0714,
      "step": 28595
    },
    {
      "epoch": 0.9789798014378638,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0849,
      "step": 28596
    },
    {
      "epoch": 0.9790140362889421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0791,
      "step": 28597
    },
    {
      "epoch": 0.9790482711400206,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0626,
      "step": 28598
    },
    {
      "epoch": 0.9790825059910989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0734,
      "step": 28599
    },
    {
      "epoch": 0.9791167408421774,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0935,
      "step": 28600
    },
    {
      "epoch": 0.9791509756932557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0643,
      "step": 28601
    },
    {
      "epoch": 0.9791852105443342,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0724,
      "step": 28602
    },
    {
      "epoch": 0.9792194453954125,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.081,
      "step": 28603
    },
    {
      "epoch": 0.979253680246491,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0805,
      "step": 28604
    },
    {
      "epoch": 0.9792879150975693,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 28605
    },
    {
      "epoch": 0.9793221499486477,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0856,
      "step": 28606
    },
    {
      "epoch": 0.9793563847997261,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 28607
    },
    {
      "epoch": 0.9793906196508045,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0628,
      "step": 28608
    },
    {
      "epoch": 0.9794248545018829,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0669,
      "step": 28609
    },
    {
      "epoch": 0.9794590893529613,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0705,
      "step": 28610
    },
    {
      "epoch": 0.9794933242040397,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 28611
    },
    {
      "epoch": 0.9795275590551181,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0829,
      "step": 28612
    },
    {
      "epoch": 0.9795617939061965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 28613
    },
    {
      "epoch": 0.9795960287572749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0844,
      "step": 28614
    },
    {
      "epoch": 0.9796302636083533,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0732,
      "step": 28615
    },
    {
      "epoch": 0.9796644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0766,
      "step": 28616
    },
    {
      "epoch": 0.97969873331051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0664,
      "step": 28617
    },
    {
      "epoch": 0.9797329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0663,
      "step": 28618
    },
    {
      "epoch": 0.9797672030126668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0809,
      "step": 28619
    },
    {
      "epoch": 0.9798014378637453,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 28620
    },
    {
      "epoch": 0.9798356727148236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0757,
      "step": 28621
    },
    {
      "epoch": 0.9798699075659021,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0821,
      "step": 28622
    },
    {
      "epoch": 0.9799041424169805,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0698,
      "step": 28623
    },
    {
      "epoch": 0.9799383772680589,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0713,
      "step": 28624
    },
    {
      "epoch": 0.9799726121191373,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0562,
      "step": 28625
    },
    {
      "epoch": 0.9800068469702157,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0752,
      "step": 28626
    },
    {
      "epoch": 0.9800410818212941,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0539,
      "step": 28627
    },
    {
      "epoch": 0.9800753166723725,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 28628
    },
    {
      "epoch": 0.9801095515234509,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0687,
      "step": 28629
    },
    {
      "epoch": 0.9801437863745293,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0667,
      "step": 28630
    },
    {
      "epoch": 0.9801780212256077,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0732,
      "step": 28631
    },
    {
      "epoch": 0.980212256076686,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0699,
      "step": 28632
    },
    {
      "epoch": 0.9802464909277645,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0879,
      "step": 28633
    },
    {
      "epoch": 0.9802807257788428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0739,
      "step": 28634
    },
    {
      "epoch": 0.9803149606299213,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0636,
      "step": 28635
    },
    {
      "epoch": 0.9803491954809996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0763,
      "step": 28636
    },
    {
      "epoch": 0.9803834303320781,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0909,
      "step": 28637
    },
    {
      "epoch": 0.9804176651831564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0736,
      "step": 28638
    },
    {
      "epoch": 0.9804519000342349,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0699,
      "step": 28639
    },
    {
      "epoch": 0.9804861348853132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0679,
      "step": 28640
    },
    {
      "epoch": 0.9805203697363917,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0715,
      "step": 28641
    },
    {
      "epoch": 0.98055460458747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0819,
      "step": 28642
    },
    {
      "epoch": 0.9805888394385485,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0786,
      "step": 28643
    },
    {
      "epoch": 0.9806230742896268,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0719,
      "step": 28644
    },
    {
      "epoch": 0.9806573091407053,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0891,
      "step": 28645
    },
    {
      "epoch": 0.9806915439917836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.068,
      "step": 28646
    },
    {
      "epoch": 0.980725778842862,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0674,
      "step": 28647
    },
    {
      "epoch": 0.9807600136939404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0725,
      "step": 28648
    },
    {
      "epoch": 0.9807942485450188,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0882,
      "step": 28649
    },
    {
      "epoch": 0.9808284833960972,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0881,
      "step": 28650
    },
    {
      "epoch": 0.9808627182471756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 28651
    },
    {
      "epoch": 0.980896953098254,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 28652
    },
    {
      "epoch": 0.9809311879493324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 28653
    },
    {
      "epoch": 0.9809654228004108,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0713,
      "step": 28654
    },
    {
      "epoch": 0.9809996576514892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0899,
      "step": 28655
    },
    {
      "epoch": 0.9810338925025676,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0726,
      "step": 28656
    },
    {
      "epoch": 0.981068127353646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.077,
      "step": 28657
    },
    {
      "epoch": 0.9811023622047244,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0839,
      "step": 28658
    },
    {
      "epoch": 0.9811365970558028,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0631,
      "step": 28659
    },
    {
      "epoch": 0.9811708319068813,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0882,
      "step": 28660
    },
    {
      "epoch": 0.9812050667579596,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0764,
      "step": 28661
    },
    {
      "epoch": 0.981239301609038,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0834,
      "step": 28662
    },
    {
      "epoch": 0.9812735364601164,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.069,
      "step": 28663
    },
    {
      "epoch": 0.9813077713111948,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.069,
      "step": 28664
    },
    {
      "epoch": 0.9813420061622732,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0713,
      "step": 28665
    },
    {
      "epoch": 0.9813762410133516,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0668,
      "step": 28666
    },
    {
      "epoch": 0.98141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0677,
      "step": 28667
    },
    {
      "epoch": 0.9814447107155084,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0806,
      "step": 28668
    },
    {
      "epoch": 0.9814789455665868,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0683,
      "step": 28669
    },
    {
      "epoch": 0.9815131804176652,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 28670
    },
    {
      "epoch": 0.9815474152687436,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0761,
      "step": 28671
    },
    {
      "epoch": 0.981581650119822,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0735,
      "step": 28672
    },
    {
      "epoch": 0.9816158849709004,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.08,
      "step": 28673
    },
    {
      "epoch": 0.9816501198219788,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 28674
    },
    {
      "epoch": 0.9816843546730571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 28675
    },
    {
      "epoch": 0.9817185895241356,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.082,
      "step": 28676
    },
    {
      "epoch": 0.9817528243752139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0708,
      "step": 28677
    },
    {
      "epoch": 0.9817870592262924,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0774,
      "step": 28678
    },
    {
      "epoch": 0.9818212940773707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0694,
      "step": 28679
    },
    {
      "epoch": 0.9818555289284492,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0775,
      "step": 28680
    },
    {
      "epoch": 0.9818897637795275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0729,
      "step": 28681
    },
    {
      "epoch": 0.981923998630606,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0765,
      "step": 28682
    },
    {
      "epoch": 0.9819582334816843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 28683
    },
    {
      "epoch": 0.9819924683327628,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0768,
      "step": 28684
    },
    {
      "epoch": 0.9820267031838411,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 28685
    },
    {
      "epoch": 0.9820609380349196,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0992,
      "step": 28686
    },
    {
      "epoch": 0.9820951728859979,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0751,
      "step": 28687
    },
    {
      "epoch": 0.9821294077370764,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0695,
      "step": 28688
    },
    {
      "epoch": 0.9821636425881547,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0729,
      "step": 28689
    },
    {
      "epoch": 0.9821978774392331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0692,
      "step": 28690
    },
    {
      "epoch": 0.9822321122903115,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0861,
      "step": 28691
    },
    {
      "epoch": 0.9822663471413899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0681,
      "step": 28692
    },
    {
      "epoch": 0.9823005819924683,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0688,
      "step": 28693
    },
    {
      "epoch": 0.9823348168435467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0801,
      "step": 28694
    },
    {
      "epoch": 0.9823690516946252,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0881,
      "step": 28695
    },
    {
      "epoch": 0.9824032865457035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.07,
      "step": 28696
    },
    {
      "epoch": 0.982437521396782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0743,
      "step": 28697
    },
    {
      "epoch": 0.9824717562478603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0792,
      "step": 28698
    },
    {
      "epoch": 0.9825059910989388,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0636,
      "step": 28699
    },
    {
      "epoch": 0.9825402259500171,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0788,
      "step": 28700
    },
    {
      "epoch": 0.9825744608010956,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 28701
    },
    {
      "epoch": 0.9826086956521739,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 28702
    },
    {
      "epoch": 0.9826429305032524,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0638,
      "step": 28703
    },
    {
      "epoch": 0.9826771653543307,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0724,
      "step": 28704
    },
    {
      "epoch": 0.9827114002054091,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0856,
      "step": 28705
    },
    {
      "epoch": 0.9827456350564875,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0767,
      "step": 28706
    },
    {
      "epoch": 0.9827798699075659,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0709,
      "step": 28707
    },
    {
      "epoch": 0.9828141047586443,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0704,
      "step": 28708
    },
    {
      "epoch": 0.9828483396097227,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0783,
      "step": 28709
    },
    {
      "epoch": 0.9828825744608011,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 28710
    },
    {
      "epoch": 0.9829168093118795,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0736,
      "step": 28711
    },
    {
      "epoch": 0.9829510441629579,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0727,
      "step": 28712
    },
    {
      "epoch": 0.9829852790140363,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 28713
    },
    {
      "epoch": 0.9830195138651147,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0651,
      "step": 28714
    },
    {
      "epoch": 0.9830537487161931,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0751,
      "step": 28715
    },
    {
      "epoch": 0.9830879835672715,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0897,
      "step": 28716
    },
    {
      "epoch": 0.9831222184183499,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 28717
    },
    {
      "epoch": 0.9831564532694282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0933,
      "step": 28718
    },
    {
      "epoch": 0.9831906881205067,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0622,
      "step": 28719
    },
    {
      "epoch": 0.983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0915,
      "step": 28720
    },
    {
      "epoch": 0.9832591578226635,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0663,
      "step": 28721
    },
    {
      "epoch": 0.9832933926737418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0775,
      "step": 28722
    },
    {
      "epoch": 0.9833276275248203,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0735,
      "step": 28723
    },
    {
      "epoch": 0.9833618623758986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0777,
      "step": 28724
    },
    {
      "epoch": 0.9833960972269771,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0751,
      "step": 28725
    },
    {
      "epoch": 0.9834303320780554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0728,
      "step": 28726
    },
    {
      "epoch": 0.9834645669291339,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0768,
      "step": 28727
    },
    {
      "epoch": 0.9834988017802122,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0875,
      "step": 28728
    },
    {
      "epoch": 0.9835330366312907,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0598,
      "step": 28729
    },
    {
      "epoch": 0.983567271482369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0739,
      "step": 28730
    },
    {
      "epoch": 0.9836015063334475,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 28731
    },
    {
      "epoch": 0.9836357411845259,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0744,
      "step": 28732
    },
    {
      "epoch": 0.9836699760356042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0723,
      "step": 28733
    },
    {
      "epoch": 0.9837042108866827,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 28734
    },
    {
      "epoch": 0.983738445737761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0708,
      "step": 28735
    },
    {
      "epoch": 0.9837726805888395,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0736,
      "step": 28736
    },
    {
      "epoch": 0.9838069154399178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.084,
      "step": 28737
    },
    {
      "epoch": 0.9838411502909963,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0608,
      "step": 28738
    },
    {
      "epoch": 0.9838753851420746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0703,
      "step": 28739
    },
    {
      "epoch": 0.9839096199931531,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 28740
    },
    {
      "epoch": 0.9839438548442314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0725,
      "step": 28741
    },
    {
      "epoch": 0.9839780896953099,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0668,
      "step": 28742
    },
    {
      "epoch": 0.9840123245463882,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0747,
      "step": 28743
    },
    {
      "epoch": 0.9840465593974667,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 28744
    },
    {
      "epoch": 0.984080794248545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0565,
      "step": 28745
    },
    {
      "epoch": 0.9841150290996235,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0786,
      "step": 28746
    },
    {
      "epoch": 0.9841492639507018,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 28747
    },
    {
      "epoch": 0.9841834988017802,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0675,
      "step": 28748
    },
    {
      "epoch": 0.9842177336528586,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0699,
      "step": 28749
    },
    {
      "epoch": 0.984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.077,
      "step": 28750
    },
    {
      "epoch": 0.9842862033550154,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.075,
      "step": 28751
    },
    {
      "epoch": 0.9843204382060938,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0727,
      "step": 28752
    },
    {
      "epoch": 0.9843546730571722,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0772,
      "step": 28753
    },
    {
      "epoch": 0.9843889079082506,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0681,
      "step": 28754
    },
    {
      "epoch": 0.984423142759329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0663,
      "step": 28755
    },
    {
      "epoch": 0.9844573776104074,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0681,
      "step": 28756
    },
    {
      "epoch": 0.9844916124614858,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0728,
      "step": 28757
    },
    {
      "epoch": 0.9845258473125642,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0659,
      "step": 28758
    },
    {
      "epoch": 0.9845600821636425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0727,
      "step": 28759
    },
    {
      "epoch": 0.984594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0725,
      "step": 28760
    },
    {
      "epoch": 0.9846285518657993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0883,
      "step": 28761
    },
    {
      "epoch": 0.9846627867168778,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0724,
      "step": 28762
    },
    {
      "epoch": 0.9846970215679561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0686,
      "step": 28763
    },
    {
      "epoch": 0.9847312564190346,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0677,
      "step": 28764
    },
    {
      "epoch": 0.9847654912701129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0653,
      "step": 28765
    },
    {
      "epoch": 0.9847997261211914,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0824,
      "step": 28766
    },
    {
      "epoch": 0.9848339609722697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0613,
      "step": 28767
    },
    {
      "epoch": 0.9848681958233482,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0633,
      "step": 28768
    },
    {
      "epoch": 0.9849024306744266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0705,
      "step": 28769
    },
    {
      "epoch": 0.984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0441,
      "step": 28770
    },
    {
      "epoch": 0.9849709003765834,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0782,
      "step": 28771
    },
    {
      "epoch": 0.9850051352276618,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0742,
      "step": 28772
    },
    {
      "epoch": 0.9850393700787402,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0669,
      "step": 28773
    },
    {
      "epoch": 0.9850736049298185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0766,
      "step": 28774
    },
    {
      "epoch": 0.985107839780897,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0762,
      "step": 28775
    },
    {
      "epoch": 0.9851420746319753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0634,
      "step": 28776
    },
    {
      "epoch": 0.9851763094830538,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0555,
      "step": 28777
    },
    {
      "epoch": 0.9852105443341321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0644,
      "step": 28778
    },
    {
      "epoch": 0.9852447791852106,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.082,
      "step": 28779
    },
    {
      "epoch": 0.9852790140362889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 28780
    },
    {
      "epoch": 0.9853132488873674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0809,
      "step": 28781
    },
    {
      "epoch": 0.9853474837384457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0814,
      "step": 28782
    },
    {
      "epoch": 0.9853817185895242,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0693,
      "step": 28783
    },
    {
      "epoch": 0.9854159534406025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0609,
      "step": 28784
    },
    {
      "epoch": 0.985450188291681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0713,
      "step": 28785
    },
    {
      "epoch": 0.9854844231427593,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0929,
      "step": 28786
    },
    {
      "epoch": 0.9855186579938378,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0754,
      "step": 28787
    },
    {
      "epoch": 0.9855528928449161,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0772,
      "step": 28788
    },
    {
      "epoch": 0.9855871276959945,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0712,
      "step": 28789
    },
    {
      "epoch": 0.9856213625470729,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0697,
      "step": 28790
    },
    {
      "epoch": 0.9856555973981513,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0879,
      "step": 28791
    },
    {
      "epoch": 0.9856898322492297,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 28792
    },
    {
      "epoch": 0.9857240671003081,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0675,
      "step": 28793
    },
    {
      "epoch": 0.9857583019513865,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.073,
      "step": 28794
    },
    {
      "epoch": 0.9857925368024649,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0884,
      "step": 28795
    },
    {
      "epoch": 0.9858267716535433,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0737,
      "step": 28796
    },
    {
      "epoch": 0.9858610065046217,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0678,
      "step": 28797
    },
    {
      "epoch": 0.9858952413557001,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0653,
      "step": 28798
    },
    {
      "epoch": 0.9859294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0765,
      "step": 28799
    },
    {
      "epoch": 0.9859637110578569,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0892,
      "step": 28800
    },
    {
      "epoch": 0.9859979459089353,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0702,
      "step": 28801
    },
    {
      "epoch": 0.9860321807600136,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.091,
      "step": 28802
    },
    {
      "epoch": 0.9860664156110921,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0697,
      "step": 28803
    },
    {
      "epoch": 0.9861006504621704,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0625,
      "step": 28804
    },
    {
      "epoch": 0.9861348853132489,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.06,
      "step": 28805
    },
    {
      "epoch": 0.9861691201643273,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0691,
      "step": 28806
    },
    {
      "epoch": 0.9862033550154057,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0598,
      "step": 28807
    },
    {
      "epoch": 0.9862375898664841,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0661,
      "step": 28808
    },
    {
      "epoch": 0.9862718247175625,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0823,
      "step": 28809
    },
    {
      "epoch": 0.9863060595686409,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0597,
      "step": 28810
    },
    {
      "epoch": 0.9863402944197193,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0846,
      "step": 28811
    },
    {
      "epoch": 0.9863745292707977,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0684,
      "step": 28812
    },
    {
      "epoch": 0.9864087641218761,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0665,
      "step": 28813
    },
    {
      "epoch": 0.9864429989729545,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.063,
      "step": 28814
    },
    {
      "epoch": 0.9864772338240329,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0778,
      "step": 28815
    },
    {
      "epoch": 0.9865114686751113,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0761,
      "step": 28816
    },
    {
      "epoch": 0.9865457035261896,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.093,
      "step": 28817
    },
    {
      "epoch": 0.9865799383772681,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0765,
      "step": 28818
    },
    {
      "epoch": 0.9866141732283464,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0737,
      "step": 28819
    },
    {
      "epoch": 0.9866484080794249,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0739,
      "step": 28820
    },
    {
      "epoch": 0.9866826429305032,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0646,
      "step": 28821
    },
    {
      "epoch": 0.9867168777815817,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0804,
      "step": 28822
    },
    {
      "epoch": 0.98675111263266,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0627,
      "step": 28823
    },
    {
      "epoch": 0.9867853474837385,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0712,
      "step": 28824
    },
    {
      "epoch": 0.9868195823348168,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 28825
    },
    {
      "epoch": 0.9868538171858953,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0777,
      "step": 28826
    },
    {
      "epoch": 0.9868880520369736,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0764,
      "step": 28827
    },
    {
      "epoch": 0.9869222868880521,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0788,
      "step": 28828
    },
    {
      "epoch": 0.9869565217391304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0649,
      "step": 28829
    },
    {
      "epoch": 0.9869907565902089,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0676,
      "step": 28830
    },
    {
      "epoch": 0.9870249914412872,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0743,
      "step": 28831
    },
    {
      "epoch": 0.9870592262923656,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0905,
      "step": 28832
    },
    {
      "epoch": 0.987093461143444,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0738,
      "step": 28833
    },
    {
      "epoch": 0.9871276959945224,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0703,
      "step": 28834
    },
    {
      "epoch": 0.9871619308456008,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0698,
      "step": 28835
    },
    {
      "epoch": 0.9871961656966792,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0845,
      "step": 28836
    },
    {
      "epoch": 0.9872304005477576,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.062,
      "step": 28837
    },
    {
      "epoch": 0.987264635398836,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0919,
      "step": 28838
    },
    {
      "epoch": 0.9872988702499144,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 28839
    },
    {
      "epoch": 0.9873331051009928,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0549,
      "step": 28840
    },
    {
      "epoch": 0.9873673399520712,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0807,
      "step": 28841
    },
    {
      "epoch": 0.9874015748031496,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0769,
      "step": 28842
    },
    {
      "epoch": 0.9874358096542281,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0924,
      "step": 28843
    },
    {
      "epoch": 0.9874700445053064,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0823,
      "step": 28844
    },
    {
      "epoch": 0.9875042793563849,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0502,
      "step": 28845
    },
    {
      "epoch": 0.9875385142074632,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0727,
      "step": 28846
    },
    {
      "epoch": 0.9875727490585416,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0645,
      "step": 28847
    },
    {
      "epoch": 0.98760698390962,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 28848
    },
    {
      "epoch": 0.9876412187606984,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0927,
      "step": 28849
    },
    {
      "epoch": 0.9876754536117768,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0944,
      "step": 28850
    },
    {
      "epoch": 0.9877096884628552,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0701,
      "step": 28851
    },
    {
      "epoch": 0.9877439233139336,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0604,
      "step": 28852
    },
    {
      "epoch": 0.987778158165012,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0647,
      "step": 28853
    },
    {
      "epoch": 0.9878123930160904,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0747,
      "step": 28854
    },
    {
      "epoch": 0.9878466278671688,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0648,
      "step": 28855
    },
    {
      "epoch": 0.9878808627182472,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0735,
      "step": 28856
    },
    {
      "epoch": 0.9879150975693256,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0499,
      "step": 28857
    },
    {
      "epoch": 0.987949332420404,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0722,
      "step": 28858
    },
    {
      "epoch": 0.9879835672714824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.056,
      "step": 28859
    },
    {
      "epoch": 0.9880178021225607,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0725,
      "step": 28860
    },
    {
      "epoch": 0.9880520369736392,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0776,
      "step": 28861
    },
    {
      "epoch": 0.9880862718247175,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0608,
      "step": 28862
    },
    {
      "epoch": 0.988120506675796,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0622,
      "step": 28863
    },
    {
      "epoch": 0.9881547415268743,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0803,
      "step": 28864
    },
    {
      "epoch": 0.9881889763779528,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0857,
      "step": 28865
    },
    {
      "epoch": 0.9882232112290311,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 28866
    },
    {
      "epoch": 0.9882574460801096,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0779,
      "step": 28867
    },
    {
      "epoch": 0.9882916809311879,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0794,
      "step": 28868
    },
    {
      "epoch": 0.9883259157822664,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0654,
      "step": 28869
    },
    {
      "epoch": 0.9883601506333447,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0726,
      "step": 28870
    },
    {
      "epoch": 0.9883943854844232,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0799,
      "step": 28871
    },
    {
      "epoch": 0.9884286203355015,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0738,
      "step": 28872
    },
    {
      "epoch": 0.98846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0695,
      "step": 28873
    },
    {
      "epoch": 0.9884970900376583,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0766,
      "step": 28874
    },
    {
      "epoch": 0.9885313248887367,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0683,
      "step": 28875
    },
    {
      "epoch": 0.9885655597398151,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0712,
      "step": 28876
    },
    {
      "epoch": 0.9885997945908935,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.08,
      "step": 28877
    },
    {
      "epoch": 0.9886340294419719,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0579,
      "step": 28878
    },
    {
      "epoch": 0.9886682642930503,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0587,
      "step": 28879
    },
    {
      "epoch": 0.9887024991441288,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0792,
      "step": 28880
    },
    {
      "epoch": 0.9887367339952071,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0668,
      "step": 28881
    },
    {
      "epoch": 0.9887709688462856,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0681,
      "step": 28882
    },
    {
      "epoch": 0.9888052036973639,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0751,
      "step": 28883
    },
    {
      "epoch": 0.9888394385484424,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0829,
      "step": 28884
    },
    {
      "epoch": 0.9888736733995207,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0634,
      "step": 28885
    },
    {
      "epoch": 0.9889079082505992,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0726,
      "step": 28886
    },
    {
      "epoch": 0.9889421431016775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0756,
      "step": 28887
    },
    {
      "epoch": 0.988976377952756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0767,
      "step": 28888
    },
    {
      "epoch": 0.9890106128038343,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0642,
      "step": 28889
    },
    {
      "epoch": 0.9890448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0649,
      "step": 28890
    },
    {
      "epoch": 0.9890790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.074,
      "step": 28891
    },
    {
      "epoch": 0.9891133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0753,
      "step": 28892
    },
    {
      "epoch": 0.9891475522081479,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0713,
      "step": 28893
    },
    {
      "epoch": 0.9891817870592263,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0732,
      "step": 28894
    },
    {
      "epoch": 0.9892160219103047,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0843,
      "step": 28895
    },
    {
      "epoch": 0.9892502567613831,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0789,
      "step": 28896
    },
    {
      "epoch": 0.9892844916124615,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0807,
      "step": 28897
    },
    {
      "epoch": 0.9893187264635399,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0717,
      "step": 28898
    },
    {
      "epoch": 0.9893529613146183,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0649,
      "step": 28899
    },
    {
      "epoch": 0.9893871961656967,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0874,
      "step": 28900
    },
    {
      "epoch": 0.989421431016775,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0703,
      "step": 28901
    },
    {
      "epoch": 0.9894556658678535,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0854,
      "step": 28902
    },
    {
      "epoch": 0.9894899007189318,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0719,
      "step": 28903
    },
    {
      "epoch": 0.9895241355700103,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0519,
      "step": 28904
    },
    {
      "epoch": 0.9895583704210886,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0776,
      "step": 28905
    },
    {
      "epoch": 0.9895926052721671,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.06,
      "step": 28906
    },
    {
      "epoch": 0.9896268401232454,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0775,
      "step": 28907
    },
    {
      "epoch": 0.9896610749743239,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0768,
      "step": 28908
    },
    {
      "epoch": 0.9896953098254022,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0828,
      "step": 28909
    },
    {
      "epoch": 0.9897295446764807,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0629,
      "step": 28910
    },
    {
      "epoch": 0.989763779527559,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0768,
      "step": 28911
    },
    {
      "epoch": 0.9897980143786375,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0716,
      "step": 28912
    },
    {
      "epoch": 0.9898322492297158,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0674,
      "step": 28913
    },
    {
      "epoch": 0.9898664840807943,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0768,
      "step": 28914
    },
    {
      "epoch": 0.9899007189318726,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0695,
      "step": 28915
    },
    {
      "epoch": 0.989934953782951,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0882,
      "step": 28916
    },
    {
      "epoch": 0.9899691886340295,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0857,
      "step": 28917
    },
    {
      "epoch": 0.9900034234851078,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0718,
      "step": 28918
    },
    {
      "epoch": 0.9900376583361863,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0606,
      "step": 28919
    },
    {
      "epoch": 0.9900718931872646,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0689,
      "step": 28920
    },
    {
      "epoch": 0.9901061280383431,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0633,
      "step": 28921
    },
    {
      "epoch": 0.9901403628894214,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0868,
      "step": 28922
    },
    {
      "epoch": 0.9901745977404999,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0752,
      "step": 28923
    },
    {
      "epoch": 0.9902088325915782,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0743,
      "step": 28924
    },
    {
      "epoch": 0.9902430674426567,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0818,
      "step": 28925
    },
    {
      "epoch": 0.990277302293735,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.074,
      "step": 28926
    },
    {
      "epoch": 0.9903115371448135,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0696,
      "step": 28927
    },
    {
      "epoch": 0.9903457719958918,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0825,
      "step": 28928
    },
    {
      "epoch": 0.9903800068469703,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.1023,
      "step": 28929
    },
    {
      "epoch": 0.9904142416980486,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0702,
      "step": 28930
    },
    {
      "epoch": 0.990448476549127,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0677,
      "step": 28931
    },
    {
      "epoch": 0.9904827114002054,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0763,
      "step": 28932
    },
    {
      "epoch": 0.9905169462512838,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 28933
    },
    {
      "epoch": 0.9905511811023622,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0554,
      "step": 28934
    },
    {
      "epoch": 0.9905854159534406,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0711,
      "step": 28935
    },
    {
      "epoch": 0.990619650804519,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0717,
      "step": 28936
    },
    {
      "epoch": 0.9906538856555974,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0726,
      "step": 28937
    },
    {
      "epoch": 0.9906881205066758,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0684,
      "step": 28938
    },
    {
      "epoch": 0.9907223553577542,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0671,
      "step": 28939
    },
    {
      "epoch": 0.9907565902088326,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0824,
      "step": 28940
    },
    {
      "epoch": 0.990790825059911,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0603,
      "step": 28941
    },
    {
      "epoch": 0.9908250599109893,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0649,
      "step": 28942
    },
    {
      "epoch": 0.9908592947620678,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0807,
      "step": 28943
    },
    {
      "epoch": 0.9908935296131461,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0829,
      "step": 28944
    },
    {
      "epoch": 0.9909277644642246,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0578,
      "step": 28945
    },
    {
      "epoch": 0.9909619993153029,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0691,
      "step": 28946
    },
    {
      "epoch": 0.9909962341663814,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0754,
      "step": 28947
    },
    {
      "epoch": 0.9910304690174597,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0751,
      "step": 28948
    },
    {
      "epoch": 0.9910647038685382,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0751,
      "step": 28949
    },
    {
      "epoch": 0.9910989387196165,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0648,
      "step": 28950
    },
    {
      "epoch": 0.991133173570695,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0756,
      "step": 28951
    },
    {
      "epoch": 0.9911674084217733,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0692,
      "step": 28952
    },
    {
      "epoch": 0.9912016432728518,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0635,
      "step": 28953
    },
    {
      "epoch": 0.9912358781239302,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0632,
      "step": 28954
    },
    {
      "epoch": 0.9912701129750086,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0798,
      "step": 28955
    },
    {
      "epoch": 0.991304347826087,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.071,
      "step": 28956
    },
    {
      "epoch": 0.9913385826771653,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0692,
      "step": 28957
    },
    {
      "epoch": 0.9913728175282438,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0806,
      "step": 28958
    },
    {
      "epoch": 0.9914070523793221,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0796,
      "step": 28959
    },
    {
      "epoch": 0.9914412872304006,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0507,
      "step": 28960
    },
    {
      "epoch": 0.9914755220814789,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0832,
      "step": 28961
    },
    {
      "epoch": 0.9915097569325574,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0786,
      "step": 28962
    },
    {
      "epoch": 0.9915439917836357,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0535,
      "step": 28963
    },
    {
      "epoch": 0.9915782266347142,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0807,
      "step": 28964
    },
    {
      "epoch": 0.9916124614857925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0728,
      "step": 28965
    },
    {
      "epoch": 0.991646696336871,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0682,
      "step": 28966
    },
    {
      "epoch": 0.9916809311879493,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0825,
      "step": 28967
    },
    {
      "epoch": 0.9917151660390278,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0815,
      "step": 28968
    },
    {
      "epoch": 0.9917494008901061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0704,
      "step": 28969
    },
    {
      "epoch": 0.9917836357411846,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.077,
      "step": 28970
    },
    {
      "epoch": 0.9918178705922629,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0687,
      "step": 28971
    },
    {
      "epoch": 0.9918521054433413,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0818,
      "step": 28972
    },
    {
      "epoch": 0.9918863402944197,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.074,
      "step": 28973
    },
    {
      "epoch": 0.9919205751454981,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0661,
      "step": 28974
    },
    {
      "epoch": 0.9919548099965765,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0757,
      "step": 28975
    },
    {
      "epoch": 0.9919890448476549,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0576,
      "step": 28976
    },
    {
      "epoch": 0.9920232796987333,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0715,
      "step": 28977
    },
    {
      "epoch": 0.9920575145498117,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0728,
      "step": 28978
    },
    {
      "epoch": 0.9920917494008901,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0788,
      "step": 28979
    },
    {
      "epoch": 0.9921259842519685,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0987,
      "step": 28980
    },
    {
      "epoch": 0.9921602191030469,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0693,
      "step": 28981
    },
    {
      "epoch": 0.9921944539541253,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0739,
      "step": 28982
    },
    {
      "epoch": 0.9922286888052037,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0675,
      "step": 28983
    },
    {
      "epoch": 0.9922629236562821,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0763,
      "step": 28984
    },
    {
      "epoch": 0.9922971585073604,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0676,
      "step": 28985
    },
    {
      "epoch": 0.9923313933584389,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0462,
      "step": 28986
    },
    {
      "epoch": 0.9923656282095172,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0634,
      "step": 28987
    },
    {
      "epoch": 0.9923998630605957,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0611,
      "step": 28988
    },
    {
      "epoch": 0.992434097911674,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0629,
      "step": 28989
    },
    {
      "epoch": 0.9924683327627525,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0748,
      "step": 28990
    },
    {
      "epoch": 0.9925025676138309,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0734,
      "step": 28991
    },
    {
      "epoch": 0.9925368024649093,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0635,
      "step": 28992
    },
    {
      "epoch": 0.9925710373159877,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0727,
      "step": 28993
    },
    {
      "epoch": 0.9926052721670661,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0698,
      "step": 28994
    },
    {
      "epoch": 0.9926395070181445,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 28995
    },
    {
      "epoch": 0.9926737418692229,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0668,
      "step": 28996
    },
    {
      "epoch": 0.9927079767203013,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0698,
      "step": 28997
    },
    {
      "epoch": 0.9927422115713797,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0608,
      "step": 28998
    },
    {
      "epoch": 0.9927764464224581,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0651,
      "step": 28999
    },
    {
      "epoch": 0.9928106812735364,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.071,
      "step": 29000
    },
    {
      "epoch": 0.9928449161246149,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0713,
      "step": 29001
    },
    {
      "epoch": 0.9928791509756932,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0744,
      "step": 29002
    },
    {
      "epoch": 0.9929133858267717,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.072,
      "step": 29003
    },
    {
      "epoch": 0.99294762067785,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0725,
      "step": 29004
    },
    {
      "epoch": 0.9929818555289285,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0546,
      "step": 29005
    },
    {
      "epoch": 0.9930160903800068,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.068,
      "step": 29006
    },
    {
      "epoch": 0.9930503252310853,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0807,
      "step": 29007
    },
    {
      "epoch": 0.9930845600821636,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0602,
      "step": 29008
    },
    {
      "epoch": 0.9931187949332421,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0639,
      "step": 29009
    },
    {
      "epoch": 0.9931530297843204,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0793,
      "step": 29010
    },
    {
      "epoch": 0.9931872646353989,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0591,
      "step": 29011
    },
    {
      "epoch": 0.9932214994864772,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0602,
      "step": 29012
    },
    {
      "epoch": 0.9932557343375557,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0674,
      "step": 29013
    },
    {
      "epoch": 0.993289969188634,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0812,
      "step": 29014
    },
    {
      "epoch": 0.9933242040397124,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.059,
      "step": 29015
    },
    {
      "epoch": 0.9933584388907908,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0701,
      "step": 29016
    },
    {
      "epoch": 0.9933926737418692,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0722,
      "step": 29017
    },
    {
      "epoch": 0.9934269085929476,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0642,
      "step": 29018
    },
    {
      "epoch": 0.993461143444026,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0544,
      "step": 29019
    },
    {
      "epoch": 0.9934953782951044,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0795,
      "step": 29020
    },
    {
      "epoch": 0.9935296131461828,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0562,
      "step": 29021
    },
    {
      "epoch": 0.9935638479972612,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0663,
      "step": 29022
    },
    {
      "epoch": 0.9935980828483396,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0848,
      "step": 29023
    },
    {
      "epoch": 0.993632317699418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0611,
      "step": 29024
    },
    {
      "epoch": 0.9936665525504964,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0594,
      "step": 29025
    },
    {
      "epoch": 0.9937007874015747,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.064,
      "step": 29026
    },
    {
      "epoch": 0.9937350222526532,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0723,
      "step": 29027
    },
    {
      "epoch": 0.9937692571037317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.06,
      "step": 29028
    },
    {
      "epoch": 0.99380349195481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0787,
      "step": 29029
    },
    {
      "epoch": 0.9938377268058884,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0907,
      "step": 29030
    },
    {
      "epoch": 0.9938719616569668,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0547,
      "step": 29031
    },
    {
      "epoch": 0.9939061965080452,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0814,
      "step": 29032
    },
    {
      "epoch": 0.9939404313591236,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0544,
      "step": 29033
    },
    {
      "epoch": 0.993974666210202,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0708,
      "step": 29034
    },
    {
      "epoch": 0.9940089010612804,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0736,
      "step": 29035
    },
    {
      "epoch": 0.9940431359123588,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0737,
      "step": 29036
    },
    {
      "epoch": 0.9940773707634372,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0812,
      "step": 29037
    },
    {
      "epoch": 0.9941116056145156,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0716,
      "step": 29038
    },
    {
      "epoch": 0.994145840465594,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0882,
      "step": 29039
    },
    {
      "epoch": 0.9941800753166724,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0704,
      "step": 29040
    },
    {
      "epoch": 0.9942143101677507,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0691,
      "step": 29041
    },
    {
      "epoch": 0.9942485450188292,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0677,
      "step": 29042
    },
    {
      "epoch": 0.9942827798699075,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0721,
      "step": 29043
    },
    {
      "epoch": 0.994317014720986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0741,
      "step": 29044
    },
    {
      "epoch": 0.9943512495720643,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0854,
      "step": 29045
    },
    {
      "epoch": 0.9943854844231428,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0494,
      "step": 29046
    },
    {
      "epoch": 0.9944197192742211,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.053,
      "step": 29047
    },
    {
      "epoch": 0.9944539541252996,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0784,
      "step": 29048
    },
    {
      "epoch": 0.9944881889763779,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0795,
      "step": 29049
    },
    {
      "epoch": 0.9945224238274564,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0641,
      "step": 29050
    },
    {
      "epoch": 0.9945566586785347,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0792,
      "step": 29051
    },
    {
      "epoch": 0.9945908935296132,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0534,
      "step": 29052
    },
    {
      "epoch": 0.9946251283806915,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0726,
      "step": 29053
    },
    {
      "epoch": 0.99465936323177,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0764,
      "step": 29054
    },
    {
      "epoch": 0.9946935980828483,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0687,
      "step": 29055
    },
    {
      "epoch": 0.9947278329339267,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0533,
      "step": 29056
    },
    {
      "epoch": 0.9947620677850051,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0501,
      "step": 29057
    },
    {
      "epoch": 0.9947963026360835,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0746,
      "step": 29058
    },
    {
      "epoch": 0.9948305374871619,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0696,
      "step": 29059
    },
    {
      "epoch": 0.9948647723382403,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0668,
      "step": 29060
    },
    {
      "epoch": 0.9948990071893187,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0657,
      "step": 29061
    },
    {
      "epoch": 0.9949332420403971,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0717,
      "step": 29062
    },
    {
      "epoch": 0.9949674768914756,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0751,
      "step": 29063
    },
    {
      "epoch": 0.9950017117425539,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0742,
      "step": 29064
    },
    {
      "epoch": 0.9950359465936324,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.067,
      "step": 29065
    },
    {
      "epoch": 0.9950701814447107,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0767,
      "step": 29066
    },
    {
      "epoch": 0.9951044162957892,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0838,
      "step": 29067
    },
    {
      "epoch": 0.9951386511468675,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0724,
      "step": 29068
    },
    {
      "epoch": 0.995172885997946,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.064,
      "step": 29069
    },
    {
      "epoch": 0.9952071208490243,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0629,
      "step": 29070
    },
    {
      "epoch": 0.9952413557001027,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0663,
      "step": 29071
    },
    {
      "epoch": 0.9952755905511811,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0778,
      "step": 29072
    },
    {
      "epoch": 0.9953098254022595,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0719,
      "step": 29073
    },
    {
      "epoch": 0.9953440602533379,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0824,
      "step": 29074
    },
    {
      "epoch": 0.9953782951044163,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0888,
      "step": 29075
    },
    {
      "epoch": 0.9954125299554947,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0602,
      "step": 29076
    },
    {
      "epoch": 0.9954467648065731,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0649,
      "step": 29077
    },
    {
      "epoch": 0.9954809996576515,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.075,
      "step": 29078
    },
    {
      "epoch": 0.9955152345087299,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.075,
      "step": 29079
    },
    {
      "epoch": 0.9955494693598083,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0624,
      "step": 29080
    },
    {
      "epoch": 0.9955837042108867,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.066,
      "step": 29081
    },
    {
      "epoch": 0.995617939061965,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0719,
      "step": 29082
    },
    {
      "epoch": 0.9956521739130435,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0887,
      "step": 29083
    },
    {
      "epoch": 0.9956864087641218,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0598,
      "step": 29084
    },
    {
      "epoch": 0.9957206436152003,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0681,
      "step": 29085
    },
    {
      "epoch": 0.9957548784662786,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0924,
      "step": 29086
    },
    {
      "epoch": 0.9957891133173571,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0577,
      "step": 29087
    },
    {
      "epoch": 0.9958233481684354,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0744,
      "step": 29088
    },
    {
      "epoch": 0.9958575830195139,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0646,
      "step": 29089
    },
    {
      "epoch": 0.9958918178705922,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.068,
      "step": 29090
    },
    {
      "epoch": 0.9959260527216707,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0631,
      "step": 29091
    },
    {
      "epoch": 0.995960287572749,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.06,
      "step": 29092
    },
    {
      "epoch": 0.9959945224238275,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0805,
      "step": 29093
    },
    {
      "epoch": 0.9960287572749058,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.057,
      "step": 29094
    },
    {
      "epoch": 0.9960629921259843,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0702,
      "step": 29095
    },
    {
      "epoch": 0.9960972269770626,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.071,
      "step": 29096
    },
    {
      "epoch": 0.996131461828141,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0667,
      "step": 29097
    },
    {
      "epoch": 0.9961656966792194,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0611,
      "step": 29098
    },
    {
      "epoch": 0.9961999315302978,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0677,
      "step": 29099
    },
    {
      "epoch": 0.9962341663813763,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0694,
      "step": 29100
    },
    {
      "epoch": 0.9962684012324546,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0728,
      "step": 29101
    },
    {
      "epoch": 0.9963026360835331,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0778,
      "step": 29102
    },
    {
      "epoch": 0.9963368709346114,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0759,
      "step": 29103
    },
    {
      "epoch": 0.9963711057856899,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0635,
      "step": 29104
    },
    {
      "epoch": 0.9964053406367682,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0569,
      "step": 29105
    },
    {
      "epoch": 0.9964395754878467,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0711,
      "step": 29106
    },
    {
      "epoch": 0.996473810338925,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0665,
      "step": 29107
    },
    {
      "epoch": 0.9965080451900035,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0569,
      "step": 29108
    },
    {
      "epoch": 0.9965422800410818,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0658,
      "step": 29109
    },
    {
      "epoch": 0.9965765148921603,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0796,
      "step": 29110
    },
    {
      "epoch": 0.9966107497432386,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0689,
      "step": 29111
    },
    {
      "epoch": 0.996644984594317,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.056,
      "step": 29112
    },
    {
      "epoch": 0.9966792194453954,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0763,
      "step": 29113
    },
    {
      "epoch": 0.9967134542964738,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0845,
      "step": 29114
    },
    {
      "epoch": 0.9967476891475522,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0594,
      "step": 29115
    },
    {
      "epoch": 0.9967819239986306,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0684,
      "step": 29116
    },
    {
      "epoch": 0.996816158849709,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0497,
      "step": 29117
    },
    {
      "epoch": 0.9968503937007874,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0732,
      "step": 29118
    },
    {
      "epoch": 0.9968846285518658,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0751,
      "step": 29119
    },
    {
      "epoch": 0.9969188634029442,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0694,
      "step": 29120
    },
    {
      "epoch": 0.9969530982540226,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0863,
      "step": 29121
    },
    {
      "epoch": 0.996987333105101,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0593,
      "step": 29122
    },
    {
      "epoch": 0.9970215679561794,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0779,
      "step": 29123
    },
    {
      "epoch": 0.9970558028072578,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0756,
      "step": 29124
    },
    {
      "epoch": 0.9970900376583361,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.066,
      "step": 29125
    },
    {
      "epoch": 0.9971242725094146,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0704,
      "step": 29126
    },
    {
      "epoch": 0.9971585073604929,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0648,
      "step": 29127
    },
    {
      "epoch": 0.9971927422115714,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0501,
      "step": 29128
    },
    {
      "epoch": 0.9972269770626497,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0785,
      "step": 29129
    },
    {
      "epoch": 0.9972612119137282,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0672,
      "step": 29130
    },
    {
      "epoch": 0.9972954467648065,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0693,
      "step": 29131
    },
    {
      "epoch": 0.997329681615885,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.062,
      "step": 29132
    },
    {
      "epoch": 0.9973639164669633,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0737,
      "step": 29133
    },
    {
      "epoch": 0.9973981513180418,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0853,
      "step": 29134
    },
    {
      "epoch": 0.9974323861691201,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0646,
      "step": 29135
    },
    {
      "epoch": 0.9974666210201986,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0869,
      "step": 29136
    },
    {
      "epoch": 0.997500855871277,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0679,
      "step": 29137
    },
    {
      "epoch": 0.9975350907223554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.071,
      "step": 29138
    },
    {
      "epoch": 0.9975693255734338,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0821,
      "step": 29139
    },
    {
      "epoch": 0.9976035604245121,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0678,
      "step": 29140
    },
    {
      "epoch": 0.9976377952755906,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0624,
      "step": 29141
    },
    {
      "epoch": 0.9976720301266689,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0766,
      "step": 29142
    },
    {
      "epoch": 0.9977062649777474,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0737,
      "step": 29143
    },
    {
      "epoch": 0.9977404998288257,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0716,
      "step": 29144
    },
    {
      "epoch": 0.9977747346799042,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0573,
      "step": 29145
    },
    {
      "epoch": 0.9978089695309825,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0597,
      "step": 29146
    },
    {
      "epoch": 0.997843204382061,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0635,
      "step": 29147
    },
    {
      "epoch": 0.9978774392331393,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0781,
      "step": 29148
    },
    {
      "epoch": 0.9979116740842178,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0561,
      "step": 29149
    },
    {
      "epoch": 0.9979459089352961,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0653,
      "step": 29150
    },
    {
      "epoch": 0.9979801437863746,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0766,
      "step": 29151
    },
    {
      "epoch": 0.9980143786374529,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0702,
      "step": 29152
    },
    {
      "epoch": 0.9980486134885314,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.072,
      "step": 29153
    },
    {
      "epoch": 0.9980828483396097,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0706,
      "step": 29154
    },
    {
      "epoch": 0.9981170831906881,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0796,
      "step": 29155
    },
    {
      "epoch": 0.9981513180417665,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0685,
      "step": 29156
    },
    {
      "epoch": 0.9981855528928449,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0787,
      "step": 29157
    },
    {
      "epoch": 0.9982197877439233,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0691,
      "step": 29158
    },
    {
      "epoch": 0.9982540225950017,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0579,
      "step": 29159
    },
    {
      "epoch": 0.9982882574460801,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0709,
      "step": 29160
    },
    {
      "epoch": 0.9983224922971585,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0639,
      "step": 29161
    },
    {
      "epoch": 0.9983567271482369,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0649,
      "step": 29162
    },
    {
      "epoch": 0.9983909619993153,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0588,
      "step": 29163
    },
    {
      "epoch": 0.9984251968503937,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0671,
      "step": 29164
    },
    {
      "epoch": 0.9984594317014721,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0691,
      "step": 29165
    },
    {
      "epoch": 0.9984936665525505,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0845,
      "step": 29166
    },
    {
      "epoch": 0.9985279014036289,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0629,
      "step": 29167
    },
    {
      "epoch": 0.9985621362547072,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0731,
      "step": 29168
    },
    {
      "epoch": 0.9985963711057857,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0597,
      "step": 29169
    },
    {
      "epoch": 0.998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0635,
      "step": 29170
    },
    {
      "epoch": 0.9986648408079425,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.071,
      "step": 29171
    },
    {
      "epoch": 0.9986990756590208,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0728,
      "step": 29172
    },
    {
      "epoch": 0.9987333105100993,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0818,
      "step": 29173
    },
    {
      "epoch": 0.9987675453611777,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0547,
      "step": 29174
    },
    {
      "epoch": 0.9988017802122561,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0685,
      "step": 29175
    },
    {
      "epoch": 0.9988360150633345,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0544,
      "step": 29176
    },
    {
      "epoch": 0.9988702499144129,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0513,
      "step": 29177
    },
    {
      "epoch": 0.9989044847654913,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0885,
      "step": 29178
    },
    {
      "epoch": 0.9989387196165697,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0843,
      "step": 29179
    },
    {
      "epoch": 0.9989729544676481,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0828,
      "step": 29180
    },
    {
      "epoch": 0.9990071893187265,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0652,
      "step": 29181
    },
    {
      "epoch": 0.9990414241698049,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.06,
      "step": 29182
    },
    {
      "epoch": 0.9990756590208832,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0752,
      "step": 29183
    },
    {
      "epoch": 0.9991098938719617,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0787,
      "step": 29184
    },
    {
      "epoch": 0.99914412872304,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0772,
      "step": 29185
    },
    {
      "epoch": 0.9991783635741185,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0824,
      "step": 29186
    },
    {
      "epoch": 0.9992125984251968,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0656,
      "step": 29187
    },
    {
      "epoch": 0.9992468332762753,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0616,
      "step": 29188
    },
    {
      "epoch": 0.9992810681273536,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0649,
      "step": 29189
    },
    {
      "epoch": 0.9993153029784321,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0618,
      "step": 29190
    },
    {
      "epoch": 0.9993495378295104,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0698,
      "step": 29191
    },
    {
      "epoch": 0.9993837726805889,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0721,
      "step": 29192
    },
    {
      "epoch": 0.9994180075316672,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0674,
      "step": 29193
    },
    {
      "epoch": 0.9994522423827457,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0532,
      "step": 29194
    },
    {
      "epoch": 0.999486477233824,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0515,
      "step": 29195
    },
    {
      "epoch": 0.9995207120849025,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0798,
      "step": 29196
    },
    {
      "epoch": 0.9995549469359808,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0829,
      "step": 29197
    },
    {
      "epoch": 0.9995891817870592,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0797,
      "step": 29198
    },
    {
      "epoch": 0.9996234166381376,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0678,
      "step": 29199
    },
    {
      "epoch": 0.999657651489216,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0631,
      "step": 29200
    },
    {
      "epoch": 0.9996918863402944,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0696,
      "step": 29201
    },
    {
      "epoch": 0.9997261211913728,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0644,
      "step": 29202
    },
    {
      "epoch": 0.9997603560424512,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0614,
      "step": 29203
    },
    {
      "epoch": 0.9997945908935296,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0677,
      "step": 29204
    },
    {
      "epoch": 0.999828825744608,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0811,
      "step": 29205
    },
    {
      "epoch": 0.9998630605956864,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0609,
      "step": 29206
    },
    {
      "epoch": 0.9998972954467648,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0737,
      "step": 29207
    },
    {
      "epoch": 0.9999315302978432,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0757,
      "step": 29208
    },
    {
      "epoch": 0.9999657651489215,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0651,
      "step": 29209
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.0638,
      "step": 29210
    },
    {
      "epoch": 1.0,
      "step": 29210,
      "total_flos": 1.1304695744862513e+20,
      "train_loss": 1.6284227603675976,
      "train_runtime": 129717.5823,
      "train_samples_per_second": 115.291,
      "train_steps_per_second": 0.225
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 29210,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1304695744862513e+20,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}