{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.18321729571271528,
  "eval_steps": 500,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00018321729571271528,
      "grad_norm": 7.481241625973833,
      "learning_rate": 0.0,
      "loss": 0.764,
      "step": 1
    },
    {
      "epoch": 0.00036643459142543056,
      "grad_norm": 7.936839352385908,
      "learning_rate": 1.8315018315018317e-08,
      "loss": 0.7505,
      "step": 2
    },
    {
      "epoch": 0.0005496518871381459,
      "grad_norm": 8.381229660678587,
      "learning_rate": 3.6630036630036635e-08,
      "loss": 0.7767,
      "step": 3
    },
    {
      "epoch": 0.0007328691828508611,
      "grad_norm": 7.533434041856819,
      "learning_rate": 5.494505494505495e-08,
      "loss": 0.7545,
      "step": 4
    },
    {
      "epoch": 0.0009160864785635764,
      "grad_norm": 6.4876435204860545,
      "learning_rate": 7.326007326007327e-08,
      "loss": 0.7254,
      "step": 5
    },
    {
      "epoch": 0.0010993037742762918,
      "grad_norm": 6.984439562746247,
      "learning_rate": 9.157509157509159e-08,
      "loss": 0.7238,
      "step": 6
    },
    {
      "epoch": 0.001282521069989007,
      "grad_norm": 8.432795226889532,
      "learning_rate": 1.098901098901099e-07,
      "loss": 0.7667,
      "step": 7
    },
    {
      "epoch": 0.0014657383657017222,
      "grad_norm": 6.615389400475257,
      "learning_rate": 1.282051282051282e-07,
      "loss": 0.739,
      "step": 8
    },
    {
      "epoch": 0.0016489556614144375,
      "grad_norm": 7.043438048491804,
      "learning_rate": 1.4652014652014654e-07,
      "loss": 0.7412,
      "step": 9
    },
    {
      "epoch": 0.001832172957127153,
      "grad_norm": 7.763888342467084,
      "learning_rate": 1.6483516483516484e-07,
      "loss": 0.739,
      "step": 10
    },
    {
      "epoch": 0.002015390252839868,
      "grad_norm": 7.115680896600135,
      "learning_rate": 1.8315018315018317e-07,
      "loss": 0.6814,
      "step": 11
    },
    {
      "epoch": 0.0021986075485525836,
      "grad_norm": 9.472751553447468,
      "learning_rate": 2.014652014652015e-07,
      "loss": 0.7694,
      "step": 12
    },
    {
      "epoch": 0.0023818248442652986,
      "grad_norm": 9.164365397122241,
      "learning_rate": 2.197802197802198e-07,
      "loss": 0.8101,
      "step": 13
    },
    {
      "epoch": 0.002565042139978014,
      "grad_norm": 7.831934204465113,
      "learning_rate": 2.3809523809523811e-07,
      "loss": 0.7504,
      "step": 14
    },
    {
      "epoch": 0.002748259435690729,
      "grad_norm": 6.7423320112288865,
      "learning_rate": 2.564102564102564e-07,
      "loss": 0.7442,
      "step": 15
    },
    {
      "epoch": 0.0029314767314034445,
      "grad_norm": 7.8103949840893145,
      "learning_rate": 2.7472527472527475e-07,
      "loss": 0.7782,
      "step": 16
    },
    {
      "epoch": 0.00311469402711616,
      "grad_norm": 6.1519268993583465,
      "learning_rate": 2.930402930402931e-07,
      "loss": 0.663,
      "step": 17
    },
    {
      "epoch": 0.003297911322828875,
      "grad_norm": 8.049047835270914,
      "learning_rate": 3.113553113553114e-07,
      "loss": 0.7934,
      "step": 18
    },
    {
      "epoch": 0.0034811286185415903,
      "grad_norm": 6.388709624292669,
      "learning_rate": 3.296703296703297e-07,
      "loss": 0.7241,
      "step": 19
    },
    {
      "epoch": 0.003664345914254306,
      "grad_norm": 7.039170133500172,
      "learning_rate": 3.47985347985348e-07,
      "loss": 0.7133,
      "step": 20
    },
    {
      "epoch": 0.003847563209967021,
      "grad_norm": 8.416142106876258,
      "learning_rate": 3.6630036630036635e-07,
      "loss": 0.7787,
      "step": 21
    },
    {
      "epoch": 0.004030780505679736,
      "grad_norm": 5.942542945633484,
      "learning_rate": 3.846153846153847e-07,
      "loss": 0.692,
      "step": 22
    },
    {
      "epoch": 0.004213997801392451,
      "grad_norm": 6.397620608740201,
      "learning_rate": 4.02930402930403e-07,
      "loss": 0.6885,
      "step": 23
    },
    {
      "epoch": 0.004397215097105167,
      "grad_norm": 4.789506080162222,
      "learning_rate": 4.212454212454213e-07,
      "loss": 0.6818,
      "step": 24
    },
    {
      "epoch": 0.004580432392817882,
      "grad_norm": 5.4323837663846275,
      "learning_rate": 4.395604395604396e-07,
      "loss": 0.7202,
      "step": 25
    },
    {
      "epoch": 0.004763649688530597,
      "grad_norm": 6.001418955741802,
      "learning_rate": 4.578754578754579e-07,
      "loss": 0.6915,
      "step": 26
    },
    {
      "epoch": 0.004946866984243312,
      "grad_norm": 6.931290264891086,
      "learning_rate": 4.7619047619047623e-07,
      "loss": 0.7214,
      "step": 27
    },
    {
      "epoch": 0.005130084279956028,
      "grad_norm": 5.423177642103934,
      "learning_rate": 4.945054945054946e-07,
      "loss": 0.6627,
      "step": 28
    },
    {
      "epoch": 0.005313301575668743,
      "grad_norm": 5.824082478920574,
      "learning_rate": 5.128205128205128e-07,
      "loss": 0.7128,
      "step": 29
    },
    {
      "epoch": 0.005496518871381458,
      "grad_norm": 4.837515053542817,
      "learning_rate": 5.311355311355311e-07,
      "loss": 0.7063,
      "step": 30
    },
    {
      "epoch": 0.005679736167094174,
      "grad_norm": 4.2178492878732,
      "learning_rate": 5.494505494505495e-07,
      "loss": 0.6667,
      "step": 31
    },
    {
      "epoch": 0.005862953462806889,
      "grad_norm": 3.7310273500960283,
      "learning_rate": 5.677655677655678e-07,
      "loss": 0.6724,
      "step": 32
    },
    {
      "epoch": 0.006046170758519604,
      "grad_norm": 3.4436166557785013,
      "learning_rate": 5.860805860805862e-07,
      "loss": 0.6158,
      "step": 33
    },
    {
      "epoch": 0.00622938805423232,
      "grad_norm": 2.6914219359650264,
      "learning_rate": 6.043956043956044e-07,
      "loss": 0.6384,
      "step": 34
    },
    {
      "epoch": 0.006412605349945035,
      "grad_norm": 2.8611154351289803,
      "learning_rate": 6.227106227106228e-07,
      "loss": 0.6564,
      "step": 35
    },
    {
      "epoch": 0.00659582264565775,
      "grad_norm": 2.915608387835296,
      "learning_rate": 6.41025641025641e-07,
      "loss": 0.6556,
      "step": 36
    },
    {
      "epoch": 0.006779039941370466,
      "grad_norm": 3.104210670394775,
      "learning_rate": 6.593406593406594e-07,
      "loss": 0.6571,
      "step": 37
    },
    {
      "epoch": 0.006962257237083181,
      "grad_norm": 2.416024185643152,
      "learning_rate": 6.776556776556777e-07,
      "loss": 0.6118,
      "step": 38
    },
    {
      "epoch": 0.007145474532795896,
      "grad_norm": 2.2840327527366187,
      "learning_rate": 6.95970695970696e-07,
      "loss": 0.5773,
      "step": 39
    },
    {
      "epoch": 0.007328691828508612,
      "grad_norm": 2.994955066139323,
      "learning_rate": 7.142857142857143e-07,
      "loss": 0.6621,
      "step": 40
    },
    {
      "epoch": 0.007511909124221327,
      "grad_norm": 2.4977146662969956,
      "learning_rate": 7.326007326007327e-07,
      "loss": 0.5878,
      "step": 41
    },
    {
      "epoch": 0.007695126419934042,
      "grad_norm": 2.2697538551105443,
      "learning_rate": 7.50915750915751e-07,
      "loss": 0.6161,
      "step": 42
    },
    {
      "epoch": 0.007878343715646757,
      "grad_norm": 2.040084870467292,
      "learning_rate": 7.692307692307694e-07,
      "loss": 0.625,
      "step": 43
    },
    {
      "epoch": 0.008061561011359472,
      "grad_norm": 1.4264410515550054,
      "learning_rate": 7.875457875457876e-07,
      "loss": 0.5991,
      "step": 44
    },
    {
      "epoch": 0.008244778307072188,
      "grad_norm": 1.3307329745484617,
      "learning_rate": 8.05860805860806e-07,
      "loss": 0.5802,
      "step": 45
    },
    {
      "epoch": 0.008427995602784902,
      "grad_norm": 1.3186002922814006,
      "learning_rate": 8.241758241758242e-07,
      "loss": 0.5971,
      "step": 46
    },
    {
      "epoch": 0.008611212898497618,
      "grad_norm": 1.4454419288227918,
      "learning_rate": 8.424908424908426e-07,
      "loss": 0.6152,
      "step": 47
    },
    {
      "epoch": 0.008794430194210334,
      "grad_norm": 1.2728865809977838,
      "learning_rate": 8.608058608058609e-07,
      "loss": 0.6176,
      "step": 48
    },
    {
      "epoch": 0.008977647489923048,
      "grad_norm": 1.2387163683388445,
      "learning_rate": 8.791208791208792e-07,
      "loss": 0.6233,
      "step": 49
    },
    {
      "epoch": 0.009160864785635764,
      "grad_norm": 1.1465107255795282,
      "learning_rate": 8.974358974358975e-07,
      "loss": 0.5795,
      "step": 50
    },
    {
      "epoch": 0.00934408208134848,
      "grad_norm": 1.264265250343769,
      "learning_rate": 9.157509157509158e-07,
      "loss": 0.6262,
      "step": 51
    },
    {
      "epoch": 0.009527299377061194,
      "grad_norm": 0.919312835052228,
      "learning_rate": 9.340659340659342e-07,
      "loss": 0.5535,
      "step": 52
    },
    {
      "epoch": 0.00971051667277391,
      "grad_norm": 0.8672735287012353,
      "learning_rate": 9.523809523809525e-07,
      "loss": 0.5327,
      "step": 53
    },
    {
      "epoch": 0.009893733968486624,
      "grad_norm": 0.8288169490471791,
      "learning_rate": 9.706959706959708e-07,
      "loss": 0.5673,
      "step": 54
    },
    {
      "epoch": 0.01007695126419934,
      "grad_norm": 0.8603030677007734,
      "learning_rate": 9.890109890109891e-07,
      "loss": 0.5547,
      "step": 55
    },
    {
      "epoch": 0.010260168559912056,
      "grad_norm": 0.7482006424960487,
      "learning_rate": 1.0073260073260074e-06,
      "loss": 0.6012,
      "step": 56
    },
    {
      "epoch": 0.01044338585562477,
      "grad_norm": 0.6144742944660406,
      "learning_rate": 1.0256410256410257e-06,
      "loss": 0.5231,
      "step": 57
    },
    {
      "epoch": 0.010626603151337486,
      "grad_norm": 0.634897676727318,
      "learning_rate": 1.0439560439560442e-06,
      "loss": 0.5726,
      "step": 58
    },
    {
      "epoch": 0.010809820447050202,
      "grad_norm": 0.6410545094813134,
      "learning_rate": 1.0622710622710622e-06,
      "loss": 0.5695,
      "step": 59
    },
    {
      "epoch": 0.010993037742762916,
      "grad_norm": 0.6609456014192859,
      "learning_rate": 1.0805860805860807e-06,
      "loss": 0.5805,
      "step": 60
    },
    {
      "epoch": 0.011176255038475632,
      "grad_norm": 0.7416419125176478,
      "learning_rate": 1.098901098901099e-06,
      "loss": 0.634,
      "step": 61
    },
    {
      "epoch": 0.011359472334188348,
      "grad_norm": 0.6851293138177442,
      "learning_rate": 1.1172161172161173e-06,
      "loss": 0.5544,
      "step": 62
    },
    {
      "epoch": 0.011542689629901062,
      "grad_norm": 0.6383351643137376,
      "learning_rate": 1.1355311355311355e-06,
      "loss": 0.557,
      "step": 63
    },
    {
      "epoch": 0.011725906925613778,
      "grad_norm": 0.6348085314283121,
      "learning_rate": 1.153846153846154e-06,
      "loss": 0.5632,
      "step": 64
    },
    {
      "epoch": 0.011909124221326494,
      "grad_norm": 0.7024363928163538,
      "learning_rate": 1.1721611721611723e-06,
      "loss": 0.5839,
      "step": 65
    },
    {
      "epoch": 0.012092341517039208,
      "grad_norm": 0.5900911656210056,
      "learning_rate": 1.1904761904761906e-06,
      "loss": 0.5344,
      "step": 66
    },
    {
      "epoch": 0.012275558812751924,
      "grad_norm": 0.6275963867777459,
      "learning_rate": 1.2087912087912089e-06,
      "loss": 0.5584,
      "step": 67
    },
    {
      "epoch": 0.01245877610846464,
      "grad_norm": 0.5865171487667508,
      "learning_rate": 1.2271062271062271e-06,
      "loss": 0.575,
      "step": 68
    },
    {
      "epoch": 0.012641993404177354,
      "grad_norm": 0.674046870317235,
      "learning_rate": 1.2454212454212456e-06,
      "loss": 0.5744,
      "step": 69
    },
    {
      "epoch": 0.01282521069989007,
      "grad_norm": 0.60918911566989,
      "learning_rate": 1.2637362637362637e-06,
      "loss": 0.5423,
      "step": 70
    },
    {
      "epoch": 0.013008427995602785,
      "grad_norm": 0.6221860528344966,
      "learning_rate": 1.282051282051282e-06,
      "loss": 0.5652,
      "step": 71
    },
    {
      "epoch": 0.0131916452913155,
      "grad_norm": 0.606665778835579,
      "learning_rate": 1.3003663003663005e-06,
      "loss": 0.6123,
      "step": 72
    },
    {
      "epoch": 0.013374862587028215,
      "grad_norm": 0.5884723188789052,
      "learning_rate": 1.3186813186813187e-06,
      "loss": 0.5863,
      "step": 73
    },
    {
      "epoch": 0.013558079882740931,
      "grad_norm": 0.5824062487652404,
      "learning_rate": 1.336996336996337e-06,
      "loss": 0.5918,
      "step": 74
    },
    {
      "epoch": 0.013741297178453645,
      "grad_norm": 0.5460196146561194,
      "learning_rate": 1.3553113553113553e-06,
      "loss": 0.5912,
      "step": 75
    },
    {
      "epoch": 0.013924514474166361,
      "grad_norm": 0.49213553293102813,
      "learning_rate": 1.3736263736263738e-06,
      "loss": 0.5421,
      "step": 76
    },
    {
      "epoch": 0.014107731769879077,
      "grad_norm": 0.5594168254817149,
      "learning_rate": 1.391941391941392e-06,
      "loss": 0.5122,
      "step": 77
    },
    {
      "epoch": 0.014290949065591791,
      "grad_norm": 0.5277088034821339,
      "learning_rate": 1.4102564102564104e-06,
      "loss": 0.5678,
      "step": 78
    },
    {
      "epoch": 0.014474166361304507,
      "grad_norm": 0.49574136870511754,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.5498,
      "step": 79
    },
    {
      "epoch": 0.014657383657017223,
      "grad_norm": 0.5944104688680958,
      "learning_rate": 1.4468864468864471e-06,
      "loss": 0.5469,
      "step": 80
    },
    {
      "epoch": 0.014840600952729937,
      "grad_norm": 0.4732970613815555,
      "learning_rate": 1.4652014652014654e-06,
      "loss": 0.5233,
      "step": 81
    },
    {
      "epoch": 0.015023818248442653,
      "grad_norm": 0.49591541086638596,
      "learning_rate": 1.4835164835164837e-06,
      "loss": 0.5367,
      "step": 82
    },
    {
      "epoch": 0.015207035544155369,
      "grad_norm": 0.4883648219378977,
      "learning_rate": 1.501831501831502e-06,
      "loss": 0.5782,
      "step": 83
    },
    {
      "epoch": 0.015390252839868083,
      "grad_norm": 0.48559293967760114,
      "learning_rate": 1.5201465201465202e-06,
      "loss": 0.5578,
      "step": 84
    },
    {
      "epoch": 0.015573470135580799,
      "grad_norm": 0.5802435990379926,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 0.5562,
      "step": 85
    },
    {
      "epoch": 0.015756687431293513,
      "grad_norm": 0.6655689083082568,
      "learning_rate": 1.556776556776557e-06,
      "loss": 0.5916,
      "step": 86
    },
    {
      "epoch": 0.01593990472700623,
      "grad_norm": 0.5272919501805459,
      "learning_rate": 1.5750915750915753e-06,
      "loss": 0.5462,
      "step": 87
    },
    {
      "epoch": 0.016123122022718945,
      "grad_norm": 0.5025768009972991,
      "learning_rate": 1.5934065934065933e-06,
      "loss": 0.564,
      "step": 88
    },
    {
      "epoch": 0.01630633931843166,
      "grad_norm": 0.4868371012830415,
      "learning_rate": 1.611721611721612e-06,
      "loss": 0.557,
      "step": 89
    },
    {
      "epoch": 0.016489556614144377,
      "grad_norm": 0.4724914851279723,
      "learning_rate": 1.6300366300366301e-06,
      "loss": 0.534,
      "step": 90
    },
    {
      "epoch": 0.01667277390985709,
      "grad_norm": 0.4933114328584066,
      "learning_rate": 1.6483516483516484e-06,
      "loss": 0.5628,
      "step": 91
    },
    {
      "epoch": 0.016855991205569805,
      "grad_norm": 0.5429724547645147,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.5693,
      "step": 92
    },
    {
      "epoch": 0.017039208501282523,
      "grad_norm": 0.46970029586030615,
      "learning_rate": 1.6849816849816852e-06,
      "loss": 0.5761,
      "step": 93
    },
    {
      "epoch": 0.017222425796995237,
      "grad_norm": 0.5466204682716642,
      "learning_rate": 1.7032967032967034e-06,
      "loss": 0.5407,
      "step": 94
    },
    {
      "epoch": 0.01740564309270795,
      "grad_norm": 0.46448768993410167,
      "learning_rate": 1.7216117216117217e-06,
      "loss": 0.5784,
      "step": 95
    },
    {
      "epoch": 0.01758886038842067,
      "grad_norm": 0.46769210989227256,
      "learning_rate": 1.73992673992674e-06,
      "loss": 0.5262,
      "step": 96
    },
    {
      "epoch": 0.017772077684133383,
      "grad_norm": 0.4412027593954725,
      "learning_rate": 1.7582417582417585e-06,
      "loss": 0.5608,
      "step": 97
    },
    {
      "epoch": 0.017955294979846097,
      "grad_norm": 0.5975733130220022,
      "learning_rate": 1.7765567765567768e-06,
      "loss": 0.5633,
      "step": 98
    },
    {
      "epoch": 0.018138512275558814,
      "grad_norm": 0.46156861116006753,
      "learning_rate": 1.794871794871795e-06,
      "loss": 0.5576,
      "step": 99
    },
    {
      "epoch": 0.01832172957127153,
      "grad_norm": 0.5193644526534718,
      "learning_rate": 1.8131868131868133e-06,
      "loss": 0.5533,
      "step": 100
    },
    {
      "epoch": 0.018504946866984243,
      "grad_norm": 0.479596247036775,
      "learning_rate": 1.8315018315018316e-06,
      "loss": 0.5337,
      "step": 101
    },
    {
      "epoch": 0.01868816416269696,
      "grad_norm": 0.4610500867184236,
      "learning_rate": 1.84981684981685e-06,
      "loss": 0.5314,
      "step": 102
    },
    {
      "epoch": 0.018871381458409674,
      "grad_norm": 0.481950984787821,
      "learning_rate": 1.8681318681318684e-06,
      "loss": 0.5865,
      "step": 103
    },
    {
      "epoch": 0.01905459875412239,
      "grad_norm": 0.5225151739123198,
      "learning_rate": 1.8864468864468866e-06,
      "loss": 0.5694,
      "step": 104
    },
    {
      "epoch": 0.019237816049835106,
      "grad_norm": 0.46381468108353424,
      "learning_rate": 1.904761904761905e-06,
      "loss": 0.557,
      "step": 105
    },
    {
      "epoch": 0.01942103334554782,
      "grad_norm": 0.45411743556679485,
      "learning_rate": 1.9230769230769234e-06,
      "loss": 0.5648,
      "step": 106
    },
    {
      "epoch": 0.019604250641260534,
      "grad_norm": 0.4859210831259208,
      "learning_rate": 1.9413919413919417e-06,
      "loss": 0.5685,
      "step": 107
    },
    {
      "epoch": 0.01978746793697325,
      "grad_norm": 0.4205701080898951,
      "learning_rate": 1.95970695970696e-06,
      "loss": 0.5426,
      "step": 108
    },
    {
      "epoch": 0.019970685232685966,
      "grad_norm": 0.5044757429436201,
      "learning_rate": 1.9780219780219782e-06,
      "loss": 0.5509,
      "step": 109
    },
    {
      "epoch": 0.02015390252839868,
      "grad_norm": 0.47571345015861244,
      "learning_rate": 1.9963369963369965e-06,
      "loss": 0.5281,
      "step": 110
    },
    {
      "epoch": 0.020337119824111394,
      "grad_norm": 0.5681283820477757,
      "learning_rate": 2.0146520146520148e-06,
      "loss": 0.5398,
      "step": 111
    },
    {
      "epoch": 0.020520337119824112,
      "grad_norm": 0.48199138528425167,
      "learning_rate": 2.032967032967033e-06,
      "loss": 0.56,
      "step": 112
    },
    {
      "epoch": 0.020703554415536826,
      "grad_norm": 0.4841461198775233,
      "learning_rate": 2.0512820512820513e-06,
      "loss": 0.5454,
      "step": 113
    },
    {
      "epoch": 0.02088677171124954,
      "grad_norm": 0.4680411378560794,
      "learning_rate": 2.0695970695970696e-06,
      "loss": 0.516,
      "step": 114
    },
    {
      "epoch": 0.021069989006962258,
      "grad_norm": 0.4426302566203345,
      "learning_rate": 2.0879120879120883e-06,
      "loss": 0.5493,
      "step": 115
    },
    {
      "epoch": 0.021253206302674972,
      "grad_norm": 0.5379521772056074,
      "learning_rate": 2.1062271062271066e-06,
      "loss": 0.5642,
      "step": 116
    },
    {
      "epoch": 0.021436423598387686,
      "grad_norm": 0.42644965401865687,
      "learning_rate": 2.1245421245421245e-06,
      "loss": 0.564,
      "step": 117
    },
    {
      "epoch": 0.021619640894100404,
      "grad_norm": 1.3341683924694292,
      "learning_rate": 2.1428571428571427e-06,
      "loss": 0.5831,
      "step": 118
    },
    {
      "epoch": 0.021802858189813118,
      "grad_norm": 0.5613781478585862,
      "learning_rate": 2.1611721611721614e-06,
      "loss": 0.5598,
      "step": 119
    },
    {
      "epoch": 0.021986075485525832,
      "grad_norm": 0.4389899980262906,
      "learning_rate": 2.1794871794871797e-06,
      "loss": 0.5498,
      "step": 120
    },
    {
      "epoch": 0.02216929278123855,
      "grad_norm": 0.5433793107490897,
      "learning_rate": 2.197802197802198e-06,
      "loss": 0.5838,
      "step": 121
    },
    {
      "epoch": 0.022352510076951264,
      "grad_norm": 0.49992126724035435,
      "learning_rate": 2.2161172161172163e-06,
      "loss": 0.5581,
      "step": 122
    },
    {
      "epoch": 0.022535727372663978,
      "grad_norm": 0.5192253683114394,
      "learning_rate": 2.2344322344322345e-06,
      "loss": 0.5963,
      "step": 123
    },
    {
      "epoch": 0.022718944668376696,
      "grad_norm": 0.5218720529273816,
      "learning_rate": 2.252747252747253e-06,
      "loss": 0.5568,
      "step": 124
    },
    {
      "epoch": 0.02290216196408941,
      "grad_norm": 0.41135006043138017,
      "learning_rate": 2.271062271062271e-06,
      "loss": 0.5256,
      "step": 125
    },
    {
      "epoch": 0.023085379259802124,
      "grad_norm": 0.4960132808732071,
      "learning_rate": 2.2893772893772894e-06,
      "loss": 0.5636,
      "step": 126
    },
    {
      "epoch": 0.02326859655551484,
      "grad_norm": 0.40294874528313,
      "learning_rate": 2.307692307692308e-06,
      "loss": 0.5373,
      "step": 127
    },
    {
      "epoch": 0.023451813851227556,
      "grad_norm": 0.49129951716432685,
      "learning_rate": 2.3260073260073264e-06,
      "loss": 0.5571,
      "step": 128
    },
    {
      "epoch": 0.02363503114694027,
      "grad_norm": 0.4541626724950761,
      "learning_rate": 2.3443223443223446e-06,
      "loss": 0.5407,
      "step": 129
    },
    {
      "epoch": 0.023818248442652987,
      "grad_norm": 0.44854257290105054,
      "learning_rate": 2.362637362637363e-06,
      "loss": 0.5516,
      "step": 130
    },
    {
      "epoch": 0.0240014657383657,
      "grad_norm": 0.4439039076319426,
      "learning_rate": 2.380952380952381e-06,
      "loss": 0.5192,
      "step": 131
    },
    {
      "epoch": 0.024184683034078416,
      "grad_norm": 0.4410959524536652,
      "learning_rate": 2.3992673992673995e-06,
      "loss": 0.5678,
      "step": 132
    },
    {
      "epoch": 0.024367900329791133,
      "grad_norm": 0.49361868947998544,
      "learning_rate": 2.4175824175824177e-06,
      "loss": 0.5591,
      "step": 133
    },
    {
      "epoch": 0.024551117625503847,
      "grad_norm": 0.5023022664764156,
      "learning_rate": 2.435897435897436e-06,
      "loss": 0.5818,
      "step": 134
    },
    {
      "epoch": 0.02473433492121656,
      "grad_norm": 0.4287249843742935,
      "learning_rate": 2.4542124542124543e-06,
      "loss": 0.5532,
      "step": 135
    },
    {
      "epoch": 0.02491755221692928,
      "grad_norm": 0.43595844702277614,
      "learning_rate": 2.472527472527473e-06,
      "loss": 0.534,
      "step": 136
    },
    {
      "epoch": 0.025100769512641993,
      "grad_norm": 0.5546618594341888,
      "learning_rate": 2.4908424908424913e-06,
      "loss": 0.5058,
      "step": 137
    },
    {
      "epoch": 0.025283986808354707,
      "grad_norm": 0.5017549598118495,
      "learning_rate": 2.509157509157509e-06,
      "loss": 0.5416,
      "step": 138
    },
    {
      "epoch": 0.025467204104067425,
      "grad_norm": 0.47047331776441903,
      "learning_rate": 2.5274725274725274e-06,
      "loss": 0.5557,
      "step": 139
    },
    {
      "epoch": 0.02565042139978014,
      "grad_norm": 0.4881037742451912,
      "learning_rate": 2.5457875457875457e-06,
      "loss": 0.5494,
      "step": 140
    },
    {
      "epoch": 0.025833638695492853,
      "grad_norm": 0.41933017046045795,
      "learning_rate": 2.564102564102564e-06,
      "loss": 0.5729,
      "step": 141
    },
    {
      "epoch": 0.02601685599120557,
      "grad_norm": 0.39974209370594543,
      "learning_rate": 2.582417582417583e-06,
      "loss": 0.5381,
      "step": 142
    },
    {
      "epoch": 0.026200073286918285,
      "grad_norm": 0.4473481461077892,
      "learning_rate": 2.600732600732601e-06,
      "loss": 0.5336,
      "step": 143
    },
    {
      "epoch": 0.026383290582631,
      "grad_norm": 0.45197895778035446,
      "learning_rate": 2.6190476190476192e-06,
      "loss": 0.5271,
      "step": 144
    },
    {
      "epoch": 0.026566507878343717,
      "grad_norm": 0.49529182062519755,
      "learning_rate": 2.6373626373626375e-06,
      "loss": 0.5468,
      "step": 145
    },
    {
      "epoch": 0.02674972517405643,
      "grad_norm": 0.5413416851523152,
      "learning_rate": 2.6556776556776558e-06,
      "loss": 0.5807,
      "step": 146
    },
    {
      "epoch": 0.026932942469769145,
      "grad_norm": 0.4703582853460863,
      "learning_rate": 2.673992673992674e-06,
      "loss": 0.538,
      "step": 147
    },
    {
      "epoch": 0.027116159765481863,
      "grad_norm": 0.4332295907365602,
      "learning_rate": 2.6923076923076923e-06,
      "loss": 0.5167,
      "step": 148
    },
    {
      "epoch": 0.027299377061194577,
      "grad_norm": 0.458616583095158,
      "learning_rate": 2.7106227106227106e-06,
      "loss": 0.5226,
      "step": 149
    },
    {
      "epoch": 0.02748259435690729,
      "grad_norm": 0.43740744109233864,
      "learning_rate": 2.728937728937729e-06,
      "loss": 0.541,
      "step": 150
    },
    {
      "epoch": 0.02766581165262001,
      "grad_norm": 0.5735574364330706,
      "learning_rate": 2.7472527472527476e-06,
      "loss": 0.5835,
      "step": 151
    },
    {
      "epoch": 0.027849028948332723,
      "grad_norm": 0.5055518766772779,
      "learning_rate": 2.765567765567766e-06,
      "loss": 0.5631,
      "step": 152
    },
    {
      "epoch": 0.028032246244045437,
      "grad_norm": 0.43271616931082146,
      "learning_rate": 2.783882783882784e-06,
      "loss": 0.5426,
      "step": 153
    },
    {
      "epoch": 0.028215463539758154,
      "grad_norm": 0.4971463373765023,
      "learning_rate": 2.8021978021978024e-06,
      "loss": 0.5389,
      "step": 154
    },
    {
      "epoch": 0.02839868083547087,
      "grad_norm": 0.4591734918280654,
      "learning_rate": 2.8205128205128207e-06,
      "loss": 0.5309,
      "step": 155
    },
    {
      "epoch": 0.028581898131183583,
      "grad_norm": 0.45247087926212365,
      "learning_rate": 2.838827838827839e-06,
      "loss": 0.5536,
      "step": 156
    },
    {
      "epoch": 0.0287651154268963,
      "grad_norm": 0.46626015341742694,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.559,
      "step": 157
    },
    {
      "epoch": 0.028948332722609015,
      "grad_norm": 0.4834854871636071,
      "learning_rate": 2.8754578754578755e-06,
      "loss": 0.5476,
      "step": 158
    },
    {
      "epoch": 0.02913155001832173,
      "grad_norm": 0.570546858013933,
      "learning_rate": 2.8937728937728942e-06,
      "loss": 0.5394,
      "step": 159
    },
    {
      "epoch": 0.029314767314034446,
      "grad_norm": 0.45441480104130544,
      "learning_rate": 2.9120879120879125e-06,
      "loss": 0.5029,
      "step": 160
    },
    {
      "epoch": 0.02949798460974716,
      "grad_norm": 0.5206871098196969,
      "learning_rate": 2.930402930402931e-06,
      "loss": 0.528,
      "step": 161
    },
    {
      "epoch": 0.029681201905459875,
      "grad_norm": 0.43359542229542136,
      "learning_rate": 2.948717948717949e-06,
      "loss": 0.5502,
      "step": 162
    },
    {
      "epoch": 0.029864419201172592,
      "grad_norm": 0.46302617065984364,
      "learning_rate": 2.9670329670329673e-06,
      "loss": 0.5447,
      "step": 163
    },
    {
      "epoch": 0.030047636496885306,
      "grad_norm": 0.3915642271916536,
      "learning_rate": 2.9853479853479856e-06,
      "loss": 0.503,
      "step": 164
    },
    {
      "epoch": 0.03023085379259802,
      "grad_norm": 0.42991246482419715,
      "learning_rate": 3.003663003663004e-06,
      "loss": 0.521,
      "step": 165
    },
    {
      "epoch": 0.030414071088310738,
      "grad_norm": 0.468175424095518,
      "learning_rate": 3.021978021978022e-06,
      "loss": 0.5101,
      "step": 166
    },
    {
      "epoch": 0.030597288384023452,
      "grad_norm": 0.6468735604295471,
      "learning_rate": 3.0402930402930405e-06,
      "loss": 0.5617,
      "step": 167
    },
    {
      "epoch": 0.030780505679736166,
      "grad_norm": 0.5058923699848836,
      "learning_rate": 3.058608058608059e-06,
      "loss": 0.5154,
      "step": 168
    },
    {
      "epoch": 0.030963722975448884,
      "grad_norm": 0.45437537978064513,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 0.4993,
      "step": 169
    },
    {
      "epoch": 0.031146940271161598,
      "grad_norm": 0.467183819204265,
      "learning_rate": 3.0952380952380957e-06,
      "loss": 0.5701,
      "step": 170
    },
    {
      "epoch": 0.031330157566874316,
      "grad_norm": 0.3835674459614267,
      "learning_rate": 3.113553113553114e-06,
      "loss": 0.4902,
      "step": 171
    },
    {
      "epoch": 0.031513374862587026,
      "grad_norm": 0.4292795112150875,
      "learning_rate": 3.1318681318681323e-06,
      "loss": 0.5237,
      "step": 172
    },
    {
      "epoch": 0.031696592158299744,
      "grad_norm": 0.6593051731116806,
      "learning_rate": 3.1501831501831505e-06,
      "loss": 0.5361,
      "step": 173
    },
    {
      "epoch": 0.03187980945401246,
      "grad_norm": 0.4718436483558741,
      "learning_rate": 3.1684981684981684e-06,
      "loss": 0.5324,
      "step": 174
    },
    {
      "epoch": 0.03206302674972517,
      "grad_norm": 0.48018416222395494,
      "learning_rate": 3.1868131868131867e-06,
      "loss": 0.5273,
      "step": 175
    },
    {
      "epoch": 0.03224624404543789,
      "grad_norm": 0.6316028239985647,
      "learning_rate": 3.205128205128206e-06,
      "loss": 0.5346,
      "step": 176
    },
    {
      "epoch": 0.03242946134115061,
      "grad_norm": 0.4596646339699305,
      "learning_rate": 3.223443223443224e-06,
      "loss": 0.5263,
      "step": 177
    },
    {
      "epoch": 0.03261267863686332,
      "grad_norm": 0.6529252419894329,
      "learning_rate": 3.2417582417582424e-06,
      "loss": 0.5442,
      "step": 178
    },
    {
      "epoch": 0.032795895932576036,
      "grad_norm": 0.5538108341969676,
      "learning_rate": 3.2600732600732602e-06,
      "loss": 0.543,
      "step": 179
    },
    {
      "epoch": 0.03297911322828875,
      "grad_norm": 0.46058359751530825,
      "learning_rate": 3.2783882783882785e-06,
      "loss": 0.5101,
      "step": 180
    },
    {
      "epoch": 0.033162330524001464,
      "grad_norm": 0.42238032667898895,
      "learning_rate": 3.2967032967032968e-06,
      "loss": 0.5487,
      "step": 181
    },
    {
      "epoch": 0.03334554781971418,
      "grad_norm": 0.49973876221605035,
      "learning_rate": 3.315018315018315e-06,
      "loss": 0.5333,
      "step": 182
    },
    {
      "epoch": 0.0335287651154269,
      "grad_norm": 0.49185079032879564,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.5733,
      "step": 183
    },
    {
      "epoch": 0.03371198241113961,
      "grad_norm": 0.41357500203470565,
      "learning_rate": 3.3516483516483516e-06,
      "loss": 0.5108,
      "step": 184
    },
    {
      "epoch": 0.03389519970685233,
      "grad_norm": 0.45597659979596383,
      "learning_rate": 3.3699633699633703e-06,
      "loss": 0.5316,
      "step": 185
    },
    {
      "epoch": 0.034078417002565045,
      "grad_norm": 0.523873134306111,
      "learning_rate": 3.3882783882783886e-06,
      "loss": 0.5516,
      "step": 186
    },
    {
      "epoch": 0.034261634298277756,
      "grad_norm": 0.4604705346503309,
      "learning_rate": 3.406593406593407e-06,
      "loss": 0.5463,
      "step": 187
    },
    {
      "epoch": 0.03444485159399047,
      "grad_norm": 0.4494660179181277,
      "learning_rate": 3.424908424908425e-06,
      "loss": 0.5195,
      "step": 188
    },
    {
      "epoch": 0.03462806888970319,
      "grad_norm": 0.47161535336220833,
      "learning_rate": 3.4432234432234434e-06,
      "loss": 0.5495,
      "step": 189
    },
    {
      "epoch": 0.0348112861854159,
      "grad_norm": 0.41422931205807795,
      "learning_rate": 3.4615384615384617e-06,
      "loss": 0.5165,
      "step": 190
    },
    {
      "epoch": 0.03499450348112862,
      "grad_norm": 0.48552414748511474,
      "learning_rate": 3.47985347985348e-06,
      "loss": 0.5207,
      "step": 191
    },
    {
      "epoch": 0.03517772077684134,
      "grad_norm": 0.49918231460984896,
      "learning_rate": 3.4981684981684982e-06,
      "loss": 0.5055,
      "step": 192
    },
    {
      "epoch": 0.03536093807255405,
      "grad_norm": 0.5047784031000427,
      "learning_rate": 3.516483516483517e-06,
      "loss": 0.5541,
      "step": 193
    },
    {
      "epoch": 0.035544155368266765,
      "grad_norm": 0.45515393167769386,
      "learning_rate": 3.5347985347985352e-06,
      "loss": 0.5324,
      "step": 194
    },
    {
      "epoch": 0.03572737266397948,
      "grad_norm": 0.4697755766255993,
      "learning_rate": 3.5531135531135535e-06,
      "loss": 0.5317,
      "step": 195
    },
    {
      "epoch": 0.035910589959692193,
      "grad_norm": 0.47670842684148323,
      "learning_rate": 3.5714285714285718e-06,
      "loss": 0.5331,
      "step": 196
    },
    {
      "epoch": 0.03609380725540491,
      "grad_norm": 0.46160363428859325,
      "learning_rate": 3.58974358974359e-06,
      "loss": 0.5645,
      "step": 197
    },
    {
      "epoch": 0.03627702455111763,
      "grad_norm": 0.5456279444043086,
      "learning_rate": 3.6080586080586083e-06,
      "loss": 0.4923,
      "step": 198
    },
    {
      "epoch": 0.03646024184683034,
      "grad_norm": 0.45679324633890633,
      "learning_rate": 3.6263736263736266e-06,
      "loss": 0.5408,
      "step": 199
    },
    {
      "epoch": 0.03664345914254306,
      "grad_norm": 0.4038620329916062,
      "learning_rate": 3.644688644688645e-06,
      "loss": 0.5542,
      "step": 200
    },
    {
      "epoch": 0.036826676438255775,
      "grad_norm": 0.42338470275172285,
      "learning_rate": 3.663003663003663e-06,
      "loss": 0.5509,
      "step": 201
    },
    {
      "epoch": 0.037009893733968485,
      "grad_norm": 0.48862631901451187,
      "learning_rate": 3.681318681318682e-06,
      "loss": 0.533,
      "step": 202
    },
    {
      "epoch": 0.0371931110296812,
      "grad_norm": 0.4339739485667441,
      "learning_rate": 3.6996336996337e-06,
      "loss": 0.5331,
      "step": 203
    },
    {
      "epoch": 0.03737632832539392,
      "grad_norm": 0.4040593227125272,
      "learning_rate": 3.7179487179487184e-06,
      "loss": 0.568,
      "step": 204
    },
    {
      "epoch": 0.03755954562110663,
      "grad_norm": 0.5025361443024049,
      "learning_rate": 3.7362637362637367e-06,
      "loss": 0.5616,
      "step": 205
    },
    {
      "epoch": 0.03774276291681935,
      "grad_norm": 0.43671528407867277,
      "learning_rate": 3.754578754578755e-06,
      "loss": 0.5468,
      "step": 206
    },
    {
      "epoch": 0.037925980212532066,
      "grad_norm": 0.5426409729355149,
      "learning_rate": 3.7728937728937733e-06,
      "loss": 0.5274,
      "step": 207
    },
    {
      "epoch": 0.03810919750824478,
      "grad_norm": 0.4383066270524436,
      "learning_rate": 3.7912087912087915e-06,
      "loss": 0.5491,
      "step": 208
    },
    {
      "epoch": 0.038292414803957495,
      "grad_norm": 0.4541571165503901,
      "learning_rate": 3.80952380952381e-06,
      "loss": 0.511,
      "step": 209
    },
    {
      "epoch": 0.03847563209967021,
      "grad_norm": 0.42914694472434756,
      "learning_rate": 3.827838827838828e-06,
      "loss": 0.4891,
      "step": 210
    },
    {
      "epoch": 0.03865884939538292,
      "grad_norm": 0.4583586099579229,
      "learning_rate": 3.846153846153847e-06,
      "loss": 0.5359,
      "step": 211
    },
    {
      "epoch": 0.03884206669109564,
      "grad_norm": 0.46332307938408596,
      "learning_rate": 3.864468864468865e-06,
      "loss": 0.506,
      "step": 212
    },
    {
      "epoch": 0.03902528398680836,
      "grad_norm": 0.4979093495563886,
      "learning_rate": 3.882783882783883e-06,
      "loss": 0.5523,
      "step": 213
    },
    {
      "epoch": 0.03920850128252107,
      "grad_norm": 0.524225251704003,
      "learning_rate": 3.901098901098901e-06,
      "loss": 0.543,
      "step": 214
    },
    {
      "epoch": 0.039391718578233786,
      "grad_norm": 0.5147235908704881,
      "learning_rate": 3.91941391941392e-06,
      "loss": 0.5465,
      "step": 215
    },
    {
      "epoch": 0.0395749358739465,
      "grad_norm": 0.4170039710750516,
      "learning_rate": 3.937728937728938e-06,
      "loss": 0.557,
      "step": 216
    },
    {
      "epoch": 0.039758153169659215,
      "grad_norm": 0.4459151957597281,
      "learning_rate": 3.9560439560439565e-06,
      "loss": 0.5176,
      "step": 217
    },
    {
      "epoch": 0.03994137046537193,
      "grad_norm": 0.4387028825678643,
      "learning_rate": 3.974358974358974e-06,
      "loss": 0.5433,
      "step": 218
    },
    {
      "epoch": 0.04012458776108464,
      "grad_norm": 0.4606220492260897,
      "learning_rate": 3.992673992673993e-06,
      "loss": 0.5558,
      "step": 219
    },
    {
      "epoch": 0.04030780505679736,
      "grad_norm": 0.43876788524484817,
      "learning_rate": 4.010989010989012e-06,
      "loss": 0.496,
      "step": 220
    },
    {
      "epoch": 0.04049102235251008,
      "grad_norm": 0.5354442317270937,
      "learning_rate": 4.0293040293040296e-06,
      "loss": 0.4883,
      "step": 221
    },
    {
      "epoch": 0.04067423964822279,
      "grad_norm": 0.4471338769246311,
      "learning_rate": 4.047619047619048e-06,
      "loss": 0.5248,
      "step": 222
    },
    {
      "epoch": 0.040857456943935506,
      "grad_norm": 0.4728450428012797,
      "learning_rate": 4.065934065934066e-06,
      "loss": 0.5412,
      "step": 223
    },
    {
      "epoch": 0.041040674239648224,
      "grad_norm": 0.5021628515290991,
      "learning_rate": 4.084249084249085e-06,
      "loss": 0.5398,
      "step": 224
    },
    {
      "epoch": 0.041223891535360935,
      "grad_norm": 0.379469323319607,
      "learning_rate": 4.102564102564103e-06,
      "loss": 0.5484,
      "step": 225
    },
    {
      "epoch": 0.04140710883107365,
      "grad_norm": 0.4682923741442823,
      "learning_rate": 4.120879120879121e-06,
      "loss": 0.5221,
      "step": 226
    },
    {
      "epoch": 0.04159032612678637,
      "grad_norm": 0.5298230539031403,
      "learning_rate": 4.139194139194139e-06,
      "loss": 0.5248,
      "step": 227
    },
    {
      "epoch": 0.04177354342249908,
      "grad_norm": 0.448919135267925,
      "learning_rate": 4.157509157509158e-06,
      "loss": 0.5492,
      "step": 228
    },
    {
      "epoch": 0.0419567607182118,
      "grad_norm": 0.4651550006247672,
      "learning_rate": 4.175824175824177e-06,
      "loss": 0.5244,
      "step": 229
    },
    {
      "epoch": 0.042139978013924516,
      "grad_norm": 0.46885786974024124,
      "learning_rate": 4.1941391941391945e-06,
      "loss": 0.5145,
      "step": 230
    },
    {
      "epoch": 0.04232319530963723,
      "grad_norm": 0.9939588576814024,
      "learning_rate": 4.212454212454213e-06,
      "loss": 0.507,
      "step": 231
    },
    {
      "epoch": 0.042506412605349944,
      "grad_norm": 0.42445387506164906,
      "learning_rate": 4.230769230769231e-06,
      "loss": 0.528,
      "step": 232
    },
    {
      "epoch": 0.04268962990106266,
      "grad_norm": 0.4386077325175301,
      "learning_rate": 4.249084249084249e-06,
      "loss": 0.5235,
      "step": 233
    },
    {
      "epoch": 0.04287284719677537,
      "grad_norm": 0.4275787850644743,
      "learning_rate": 4.267399267399268e-06,
      "loss": 0.5268,
      "step": 234
    },
    {
      "epoch": 0.04305606449248809,
      "grad_norm": 0.3819534900324145,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.5108,
      "step": 235
    },
    {
      "epoch": 0.04323928178820081,
      "grad_norm": 0.5069196499403993,
      "learning_rate": 4.304029304029305e-06,
      "loss": 0.4914,
      "step": 236
    },
    {
      "epoch": 0.04342249908391352,
      "grad_norm": 0.4224671659155117,
      "learning_rate": 4.322344322344323e-06,
      "loss": 0.531,
      "step": 237
    },
    {
      "epoch": 0.043605716379626236,
      "grad_norm": 0.49892435842001814,
      "learning_rate": 4.340659340659341e-06,
      "loss": 0.5309,
      "step": 238
    },
    {
      "epoch": 0.043788933675338954,
      "grad_norm": 0.4435906661481072,
      "learning_rate": 4.358974358974359e-06,
      "loss": 0.5386,
      "step": 239
    },
    {
      "epoch": 0.043972150971051664,
      "grad_norm": 0.4646701721197805,
      "learning_rate": 4.377289377289377e-06,
      "loss": 0.5445,
      "step": 240
    },
    {
      "epoch": 0.04415536826676438,
      "grad_norm": 0.42345412346700445,
      "learning_rate": 4.395604395604396e-06,
      "loss": 0.5171,
      "step": 241
    },
    {
      "epoch": 0.0443385855624771,
      "grad_norm": 0.4664513677280782,
      "learning_rate": 4.413919413919414e-06,
      "loss": 0.5471,
      "step": 242
    },
    {
      "epoch": 0.04452180285818981,
      "grad_norm": 0.5007403600133091,
      "learning_rate": 4.4322344322344325e-06,
      "loss": 0.5224,
      "step": 243
    },
    {
      "epoch": 0.04470502015390253,
      "grad_norm": 0.4202775330369337,
      "learning_rate": 4.45054945054945e-06,
      "loss": 0.5283,
      "step": 244
    },
    {
      "epoch": 0.044888237449615245,
      "grad_norm": 0.485312343325994,
      "learning_rate": 4.468864468864469e-06,
      "loss": 0.5365,
      "step": 245
    },
    {
      "epoch": 0.045071454745327956,
      "grad_norm": 0.4796744192562041,
      "learning_rate": 4.487179487179488e-06,
      "loss": 0.5481,
      "step": 246
    },
    {
      "epoch": 0.045254672041040674,
      "grad_norm": 0.4699525411145978,
      "learning_rate": 4.505494505494506e-06,
      "loss": 0.526,
      "step": 247
    },
    {
      "epoch": 0.04543788933675339,
      "grad_norm": 0.37235134296143163,
      "learning_rate": 4.523809523809524e-06,
      "loss": 0.4929,
      "step": 248
    },
    {
      "epoch": 0.0456211066324661,
      "grad_norm": 0.4594065464998237,
      "learning_rate": 4.542124542124542e-06,
      "loss": 0.5435,
      "step": 249
    },
    {
      "epoch": 0.04580432392817882,
      "grad_norm": 0.4812140084395847,
      "learning_rate": 4.560439560439561e-06,
      "loss": 0.4714,
      "step": 250
    },
    {
      "epoch": 0.04598754122389154,
      "grad_norm": 0.44275937617791644,
      "learning_rate": 4.578754578754579e-06,
      "loss": 0.5289,
      "step": 251
    },
    {
      "epoch": 0.04617075851960425,
      "grad_norm": 0.543477861772032,
      "learning_rate": 4.5970695970695975e-06,
      "loss": 0.5176,
      "step": 252
    },
    {
      "epoch": 0.046353975815316965,
      "grad_norm": 0.9195336625704912,
      "learning_rate": 4.615384615384616e-06,
      "loss": 0.5315,
      "step": 253
    },
    {
      "epoch": 0.04653719311102968,
      "grad_norm": 0.4922315670719196,
      "learning_rate": 4.633699633699634e-06,
      "loss": 0.5364,
      "step": 254
    },
    {
      "epoch": 0.046720410406742394,
      "grad_norm": 0.47361690353516367,
      "learning_rate": 4.652014652014653e-06,
      "loss": 0.5591,
      "step": 255
    },
    {
      "epoch": 0.04690362770245511,
      "grad_norm": 0.5164779414047217,
      "learning_rate": 4.6703296703296706e-06,
      "loss": 0.5538,
      "step": 256
    },
    {
      "epoch": 0.04708684499816783,
      "grad_norm": 0.421397036777767,
      "learning_rate": 4.688644688644689e-06,
      "loss": 0.5338,
      "step": 257
    },
    {
      "epoch": 0.04727006229388054,
      "grad_norm": 0.5053871259325204,
      "learning_rate": 4.706959706959707e-06,
      "loss": 0.544,
      "step": 258
    },
    {
      "epoch": 0.04745327958959326,
      "grad_norm": 0.4607447877406368,
      "learning_rate": 4.725274725274726e-06,
      "loss": 0.4812,
      "step": 259
    },
    {
      "epoch": 0.047636496885305975,
      "grad_norm": 0.4875422302168998,
      "learning_rate": 4.743589743589744e-06,
      "loss": 0.5614,
      "step": 260
    },
    {
      "epoch": 0.047819714181018685,
      "grad_norm": 0.41579068200919733,
      "learning_rate": 4.761904761904762e-06,
      "loss": 0.4729,
      "step": 261
    },
    {
      "epoch": 0.0480029314767314,
      "grad_norm": 0.46282242693556186,
      "learning_rate": 4.780219780219781e-06,
      "loss": 0.5224,
      "step": 262
    },
    {
      "epoch": 0.04818614877244412,
      "grad_norm": 0.4482830307148575,
      "learning_rate": 4.798534798534799e-06,
      "loss": 0.5326,
      "step": 263
    },
    {
      "epoch": 0.04836936606815683,
      "grad_norm": 0.4342637761169385,
      "learning_rate": 4.816849816849818e-06,
      "loss": 0.5056,
      "step": 264
    },
    {
      "epoch": 0.04855258336386955,
      "grad_norm": 0.42342307710917526,
      "learning_rate": 4.8351648351648355e-06,
      "loss": 0.5049,
      "step": 265
    },
    {
      "epoch": 0.04873580065958227,
      "grad_norm": 0.41567729548709964,
      "learning_rate": 4.853479853479854e-06,
      "loss": 0.5313,
      "step": 266
    },
    {
      "epoch": 0.04891901795529498,
      "grad_norm": 0.44877730848158315,
      "learning_rate": 4.871794871794872e-06,
      "loss": 0.5027,
      "step": 267
    },
    {
      "epoch": 0.049102235251007695,
      "grad_norm": 0.46298601715996757,
      "learning_rate": 4.890109890109891e-06,
      "loss": 0.5418,
      "step": 268
    },
    {
      "epoch": 0.04928545254672041,
      "grad_norm": 0.4233152082129357,
      "learning_rate": 4.908424908424909e-06,
      "loss": 0.534,
      "step": 269
    },
    {
      "epoch": 0.04946866984243312,
      "grad_norm": 0.42327618076780654,
      "learning_rate": 4.926739926739927e-06,
      "loss": 0.5073,
      "step": 270
    },
    {
      "epoch": 0.04965188713814584,
      "grad_norm": 0.47132160003804374,
      "learning_rate": 4.945054945054946e-06,
      "loss": 0.564,
      "step": 271
    },
    {
      "epoch": 0.04983510443385856,
      "grad_norm": 0.467395758671848,
      "learning_rate": 4.963369963369964e-06,
      "loss": 0.5194,
      "step": 272
    },
    {
      "epoch": 0.05001832172957127,
      "grad_norm": 0.4377241688268797,
      "learning_rate": 4.9816849816849826e-06,
      "loss": 0.5473,
      "step": 273
    },
    {
      "epoch": 0.05020153902528399,
      "grad_norm": 0.4344310773487788,
      "learning_rate": 5e-06,
      "loss": 0.5198,
      "step": 274
    },
    {
      "epoch": 0.050384756320996704,
      "grad_norm": 0.505691771505538,
      "learning_rate": 5.018315018315018e-06,
      "loss": 0.5001,
      "step": 275
    },
    {
      "epoch": 0.050567973616709415,
      "grad_norm": 0.4741702696479342,
      "learning_rate": 5.036630036630037e-06,
      "loss": 0.5269,
      "step": 276
    },
    {
      "epoch": 0.05075119091242213,
      "grad_norm": 0.4123760853657366,
      "learning_rate": 5.054945054945055e-06,
      "loss": 0.5366,
      "step": 277
    },
    {
      "epoch": 0.05093440820813485,
      "grad_norm": 0.5165952852785715,
      "learning_rate": 5.0732600732600735e-06,
      "loss": 0.5629,
      "step": 278
    },
    {
      "epoch": 0.05111762550384756,
      "grad_norm": 0.6018743338441076,
      "learning_rate": 5.091575091575091e-06,
      "loss": 0.5268,
      "step": 279
    },
    {
      "epoch": 0.05130084279956028,
      "grad_norm": 0.4647905088113548,
      "learning_rate": 5.10989010989011e-06,
      "loss": 0.5353,
      "step": 280
    },
    {
      "epoch": 0.051484060095272996,
      "grad_norm": 0.4666862939383661,
      "learning_rate": 5.128205128205128e-06,
      "loss": 0.5582,
      "step": 281
    },
    {
      "epoch": 0.05166727739098571,
      "grad_norm": 0.42295571485014016,
      "learning_rate": 5.146520146520147e-06,
      "loss": 0.5389,
      "step": 282
    },
    {
      "epoch": 0.051850494686698424,
      "grad_norm": 0.5088982589681916,
      "learning_rate": 5.164835164835166e-06,
      "loss": 0.5114,
      "step": 283
    },
    {
      "epoch": 0.05203371198241114,
      "grad_norm": 0.39252264391052066,
      "learning_rate": 5.183150183150184e-06,
      "loss": 0.5078,
      "step": 284
    },
    {
      "epoch": 0.05221692927812385,
      "grad_norm": 0.47464881831711925,
      "learning_rate": 5.201465201465202e-06,
      "loss": 0.5719,
      "step": 285
    },
    {
      "epoch": 0.05240014657383657,
      "grad_norm": 0.42811181510690394,
      "learning_rate": 5.219780219780221e-06,
      "loss": 0.536,
      "step": 286
    },
    {
      "epoch": 0.05258336386954929,
      "grad_norm": 0.4983761768332983,
      "learning_rate": 5.2380952380952384e-06,
      "loss": 0.5454,
      "step": 287
    },
    {
      "epoch": 0.052766581165262,
      "grad_norm": 0.3871752405078846,
      "learning_rate": 5.256410256410257e-06,
      "loss": 0.5172,
      "step": 288
    },
    {
      "epoch": 0.052949798460974716,
      "grad_norm": 0.44878294094458826,
      "learning_rate": 5.274725274725275e-06,
      "loss": 0.4965,
      "step": 289
    },
    {
      "epoch": 0.053133015756687434,
      "grad_norm": 0.4193937449864018,
      "learning_rate": 5.293040293040294e-06,
      "loss": 0.5418,
      "step": 290
    },
    {
      "epoch": 0.053316233052400144,
      "grad_norm": 0.47326184468203625,
      "learning_rate": 5.3113553113553116e-06,
      "loss": 0.5465,
      "step": 291
    },
    {
      "epoch": 0.05349945034811286,
      "grad_norm": 0.46890596965933473,
      "learning_rate": 5.32967032967033e-06,
      "loss": 0.5427,
      "step": 292
    },
    {
      "epoch": 0.05368266764382558,
      "grad_norm": 0.4718404244115825,
      "learning_rate": 5.347985347985348e-06,
      "loss": 0.5315,
      "step": 293
    },
    {
      "epoch": 0.05386588493953829,
      "grad_norm": 0.5313479853203268,
      "learning_rate": 5.366300366300367e-06,
      "loss": 0.5261,
      "step": 294
    },
    {
      "epoch": 0.05404910223525101,
      "grad_norm": 0.4619862699623299,
      "learning_rate": 5.384615384615385e-06,
      "loss": 0.5224,
      "step": 295
    },
    {
      "epoch": 0.054232319530963725,
      "grad_norm": 0.4235637361399875,
      "learning_rate": 5.402930402930403e-06,
      "loss": 0.5484,
      "step": 296
    },
    {
      "epoch": 0.054415536826676436,
      "grad_norm": 0.4826898937600368,
      "learning_rate": 5.421245421245421e-06,
      "loss": 0.524,
      "step": 297
    },
    {
      "epoch": 0.054598754122389154,
      "grad_norm": 0.43904512964940123,
      "learning_rate": 5.43956043956044e-06,
      "loss": 0.5486,
      "step": 298
    },
    {
      "epoch": 0.05478197141810187,
      "grad_norm": 0.5045757484045217,
      "learning_rate": 5.457875457875458e-06,
      "loss": 0.5407,
      "step": 299
    },
    {
      "epoch": 0.05496518871381458,
      "grad_norm": 0.47829971819207484,
      "learning_rate": 5.476190476190477e-06,
      "loss": 0.5344,
      "step": 300
    },
    {
      "epoch": 0.0551484060095273,
      "grad_norm": 0.416644246441645,
      "learning_rate": 5.494505494505495e-06,
      "loss": 0.5111,
      "step": 301
    },
    {
      "epoch": 0.05533162330524002,
      "grad_norm": 0.488275746902462,
      "learning_rate": 5.512820512820514e-06,
      "loss": 0.54,
      "step": 302
    },
    {
      "epoch": 0.05551484060095273,
      "grad_norm": 0.43082352297647686,
      "learning_rate": 5.531135531135532e-06,
      "loss": 0.5219,
      "step": 303
    },
    {
      "epoch": 0.055698057896665445,
      "grad_norm": 0.41708996725660685,
      "learning_rate": 5.5494505494505504e-06,
      "loss": 0.5272,
      "step": 304
    },
    {
      "epoch": 0.05588127519237816,
      "grad_norm": 0.4748217492221608,
      "learning_rate": 5.567765567765568e-06,
      "loss": 0.5439,
      "step": 305
    },
    {
      "epoch": 0.056064492488090874,
      "grad_norm": 0.5257169187612324,
      "learning_rate": 5.586080586080587e-06,
      "loss": 0.5671,
      "step": 306
    },
    {
      "epoch": 0.05624770978380359,
      "grad_norm": 0.4243472668028098,
      "learning_rate": 5.604395604395605e-06,
      "loss": 0.5044,
      "step": 307
    },
    {
      "epoch": 0.05643092707951631,
      "grad_norm": 0.43876355547814727,
      "learning_rate": 5.6227106227106235e-06,
      "loss": 0.4815,
      "step": 308
    },
    {
      "epoch": 0.05661414437522902,
      "grad_norm": 0.43342398130791976,
      "learning_rate": 5.641025641025641e-06,
      "loss": 0.5308,
      "step": 309
    },
    {
      "epoch": 0.05679736167094174,
      "grad_norm": 0.4660705177035744,
      "learning_rate": 5.65934065934066e-06,
      "loss": 0.5686,
      "step": 310
    },
    {
      "epoch": 0.056980578966654455,
      "grad_norm": 0.4170384834874546,
      "learning_rate": 5.677655677655678e-06,
      "loss": 0.4965,
      "step": 311
    },
    {
      "epoch": 0.057163796262367166,
      "grad_norm": 0.4568771189697619,
      "learning_rate": 5.695970695970696e-06,
      "loss": 0.5074,
      "step": 312
    },
    {
      "epoch": 0.05734701355807988,
      "grad_norm": 0.4666988354433752,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.5225,
      "step": 313
    },
    {
      "epoch": 0.0575302308537926,
      "grad_norm": 0.38305953336818827,
      "learning_rate": 5.732600732600732e-06,
      "loss": 0.5057,
      "step": 314
    },
    {
      "epoch": 0.05771344814950531,
      "grad_norm": 0.8971752127635111,
      "learning_rate": 5.750915750915751e-06,
      "loss": 0.5556,
      "step": 315
    },
    {
      "epoch": 0.05789666544521803,
      "grad_norm": 0.4580428245754137,
      "learning_rate": 5.769230769230769e-06,
      "loss": 0.5344,
      "step": 316
    },
    {
      "epoch": 0.05807988274093075,
      "grad_norm": 0.414647309534276,
      "learning_rate": 5.7875457875457885e-06,
      "loss": 0.5177,
      "step": 317
    },
    {
      "epoch": 0.05826310003664346,
      "grad_norm": 0.45495854796733415,
      "learning_rate": 5.805860805860807e-06,
      "loss": 0.5214,
      "step": 318
    },
    {
      "epoch": 0.058446317332356175,
      "grad_norm": 0.47098300998769715,
      "learning_rate": 5.824175824175825e-06,
      "loss": 0.5519,
      "step": 319
    },
    {
      "epoch": 0.05862953462806889,
      "grad_norm": 0.43335718768494447,
      "learning_rate": 5.842490842490844e-06,
      "loss": 0.5282,
      "step": 320
    },
    {
      "epoch": 0.0588127519237816,
      "grad_norm": 0.4649559782288625,
      "learning_rate": 5.860805860805862e-06,
      "loss": 0.5163,
      "step": 321
    },
    {
      "epoch": 0.05899596921949432,
      "grad_norm": 0.44623794630893965,
      "learning_rate": 5.8791208791208794e-06,
      "loss": 0.514,
      "step": 322
    },
    {
      "epoch": 0.05917918651520704,
      "grad_norm": 0.4521538486948526,
      "learning_rate": 5.897435897435898e-06,
      "loss": 0.5248,
      "step": 323
    },
    {
      "epoch": 0.05936240381091975,
      "grad_norm": 0.453853173278213,
      "learning_rate": 5.915750915750916e-06,
      "loss": 0.5105,
      "step": 324
    },
    {
      "epoch": 0.05954562110663247,
      "grad_norm": 0.45615939787874893,
      "learning_rate": 5.934065934065935e-06,
      "loss": 0.5321,
      "step": 325
    },
    {
      "epoch": 0.059728838402345184,
      "grad_norm": 0.5103507598238842,
      "learning_rate": 5.9523809523809525e-06,
      "loss": 0.4889,
      "step": 326
    },
    {
      "epoch": 0.059912055698057895,
      "grad_norm": 0.4755868373014201,
      "learning_rate": 5.970695970695971e-06,
      "loss": 0.5233,
      "step": 327
    },
    {
      "epoch": 0.06009527299377061,
      "grad_norm": 0.48186711326734255,
      "learning_rate": 5.989010989010989e-06,
      "loss": 0.5216,
      "step": 328
    },
    {
      "epoch": 0.06027849028948333,
      "grad_norm": 0.406290487409726,
      "learning_rate": 6.007326007326008e-06,
      "loss": 0.5048,
      "step": 329
    },
    {
      "epoch": 0.06046170758519604,
      "grad_norm": 0.47965200235867606,
      "learning_rate": 6.025641025641026e-06,
      "loss": 0.513,
      "step": 330
    },
    {
      "epoch": 0.06064492488090876,
      "grad_norm": 0.492103246179344,
      "learning_rate": 6.043956043956044e-06,
      "loss": 0.5553,
      "step": 331
    },
    {
      "epoch": 0.060828142176621476,
      "grad_norm": 0.49171785130479284,
      "learning_rate": 6.062271062271062e-06,
      "loss": 0.5289,
      "step": 332
    },
    {
      "epoch": 0.06101135947233419,
      "grad_norm": 0.4659535600554528,
      "learning_rate": 6.080586080586081e-06,
      "loss": 0.5143,
      "step": 333
    },
    {
      "epoch": 0.061194576768046904,
      "grad_norm": 0.43136677550682173,
      "learning_rate": 6.0989010989011e-06,
      "loss": 0.469,
      "step": 334
    },
    {
      "epoch": 0.06137779406375962,
      "grad_norm": 0.6094192173431031,
      "learning_rate": 6.117216117216118e-06,
      "loss": 0.5121,
      "step": 335
    },
    {
      "epoch": 0.06156101135947233,
      "grad_norm": 0.45352827270619606,
      "learning_rate": 6.135531135531136e-06,
      "loss": 0.5097,
      "step": 336
    },
    {
      "epoch": 0.06174422865518505,
      "grad_norm": 0.42626268219917746,
      "learning_rate": 6.153846153846155e-06,
      "loss": 0.5234,
      "step": 337
    },
    {
      "epoch": 0.06192744595089777,
      "grad_norm": 0.4438479757326601,
      "learning_rate": 6.172161172161173e-06,
      "loss": 0.4852,
      "step": 338
    },
    {
      "epoch": 0.06211066324661048,
      "grad_norm": 0.688188614062373,
      "learning_rate": 6.1904761904761914e-06,
      "loss": 0.502,
      "step": 339
    },
    {
      "epoch": 0.062293880542323196,
      "grad_norm": 0.5160627484540854,
      "learning_rate": 6.208791208791209e-06,
      "loss": 0.521,
      "step": 340
    },
    {
      "epoch": 0.062477097838035914,
      "grad_norm": 0.4356067335955304,
      "learning_rate": 6.227106227106228e-06,
      "loss": 0.4918,
      "step": 341
    },
    {
      "epoch": 0.06266031513374863,
      "grad_norm": 0.5001791021027777,
      "learning_rate": 6.245421245421246e-06,
      "loss": 0.5173,
      "step": 342
    },
    {
      "epoch": 0.06284353242946134,
      "grad_norm": 0.45621279623031163,
      "learning_rate": 6.2637362637362645e-06,
      "loss": 0.5532,
      "step": 343
    },
    {
      "epoch": 0.06302674972517405,
      "grad_norm": 0.4686583276600699,
      "learning_rate": 6.282051282051282e-06,
      "loss": 0.5544,
      "step": 344
    },
    {
      "epoch": 0.06320996702088677,
      "grad_norm": 0.4500457038114704,
      "learning_rate": 6.300366300366301e-06,
      "loss": 0.5094,
      "step": 345
    },
    {
      "epoch": 0.06339318431659949,
      "grad_norm": 0.4707435610591325,
      "learning_rate": 6.318681318681319e-06,
      "loss": 0.5317,
      "step": 346
    },
    {
      "epoch": 0.0635764016123122,
      "grad_norm": 0.44910822533973516,
      "learning_rate": 6.336996336996337e-06,
      "loss": 0.5063,
      "step": 347
    },
    {
      "epoch": 0.06375961890802492,
      "grad_norm": 0.4262957774336457,
      "learning_rate": 6.3553113553113555e-06,
      "loss": 0.5389,
      "step": 348
    },
    {
      "epoch": 0.06394283620373763,
      "grad_norm": 0.4958715530578741,
      "learning_rate": 6.373626373626373e-06,
      "loss": 0.5253,
      "step": 349
    },
    {
      "epoch": 0.06412605349945034,
      "grad_norm": 0.4706503863353741,
      "learning_rate": 6.391941391941392e-06,
      "loss": 0.5078,
      "step": 350
    },
    {
      "epoch": 0.06430927079516306,
      "grad_norm": 0.5025375698465898,
      "learning_rate": 6.410256410256412e-06,
      "loss": 0.5366,
      "step": 351
    },
    {
      "epoch": 0.06449248809087578,
      "grad_norm": 0.43307402952194485,
      "learning_rate": 6.4285714285714295e-06,
      "loss": 0.5053,
      "step": 352
    },
    {
      "epoch": 0.0646757053865885,
      "grad_norm": 0.4145957842766705,
      "learning_rate": 6.446886446886448e-06,
      "loss": 0.5134,
      "step": 353
    },
    {
      "epoch": 0.06485892268230121,
      "grad_norm": 0.42558146560441634,
      "learning_rate": 6.465201465201466e-06,
      "loss": 0.498,
      "step": 354
    },
    {
      "epoch": 0.06504213997801392,
      "grad_norm": 0.6808736114735602,
      "learning_rate": 6.483516483516485e-06,
      "loss": 0.4865,
      "step": 355
    },
    {
      "epoch": 0.06522535727372664,
      "grad_norm": 0.4858578720351213,
      "learning_rate": 6.5018315018315026e-06,
      "loss": 0.4977,
      "step": 356
    },
    {
      "epoch": 0.06540857456943935,
      "grad_norm": 0.5523209851617275,
      "learning_rate": 6.5201465201465204e-06,
      "loss": 0.5278,
      "step": 357
    },
    {
      "epoch": 0.06559179186515207,
      "grad_norm": 0.4756243917905379,
      "learning_rate": 6.538461538461539e-06,
      "loss": 0.5018,
      "step": 358
    },
    {
      "epoch": 0.06577500916086479,
      "grad_norm": 0.5270789359516691,
      "learning_rate": 6.556776556776557e-06,
      "loss": 0.5127,
      "step": 359
    },
    {
      "epoch": 0.0659582264565775,
      "grad_norm": 0.5213313487503423,
      "learning_rate": 6.575091575091576e-06,
      "loss": 0.504,
      "step": 360
    },
    {
      "epoch": 0.06614144375229021,
      "grad_norm": 0.47699985237826076,
      "learning_rate": 6.5934065934065935e-06,
      "loss": 0.5356,
      "step": 361
    },
    {
      "epoch": 0.06632466104800293,
      "grad_norm": 0.4398115760336562,
      "learning_rate": 6.611721611721612e-06,
      "loss": 0.5038,
      "step": 362
    },
    {
      "epoch": 0.06650787834371565,
      "grad_norm": 0.5848808647247892,
      "learning_rate": 6.63003663003663e-06,
      "loss": 0.5393,
      "step": 363
    },
    {
      "epoch": 0.06669109563942836,
      "grad_norm": 0.40297449884724584,
      "learning_rate": 6.648351648351649e-06,
      "loss": 0.5231,
      "step": 364
    },
    {
      "epoch": 0.06687431293514108,
      "grad_norm": 0.4615954062532406,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.5017,
      "step": 365
    },
    {
      "epoch": 0.0670575302308538,
      "grad_norm": 0.4204668889117905,
      "learning_rate": 6.684981684981685e-06,
      "loss": 0.5651,
      "step": 366
    },
    {
      "epoch": 0.0672407475265665,
      "grad_norm": 0.4765525618556606,
      "learning_rate": 6.703296703296703e-06,
      "loss": 0.5253,
      "step": 367
    },
    {
      "epoch": 0.06742396482227922,
      "grad_norm": 0.431663431335195,
      "learning_rate": 6.721611721611723e-06,
      "loss": 0.5398,
      "step": 368
    },
    {
      "epoch": 0.06760718211799194,
      "grad_norm": 0.4403652862248983,
      "learning_rate": 6.739926739926741e-06,
      "loss": 0.5252,
      "step": 369
    },
    {
      "epoch": 0.06779039941370466,
      "grad_norm": 0.483725795531489,
      "learning_rate": 6.758241758241759e-06,
      "loss": 0.5349,
      "step": 370
    },
    {
      "epoch": 0.06797361670941737,
      "grad_norm": 0.8892216780264888,
      "learning_rate": 6.776556776556777e-06,
      "loss": 0.5184,
      "step": 371
    },
    {
      "epoch": 0.06815683400513009,
      "grad_norm": 0.48775432933817636,
      "learning_rate": 6.794871794871796e-06,
      "loss": 0.5275,
      "step": 372
    },
    {
      "epoch": 0.0683400513008428,
      "grad_norm": 0.4772597017011689,
      "learning_rate": 6.813186813186814e-06,
      "loss": 0.5517,
      "step": 373
    },
    {
      "epoch": 0.06852326859655551,
      "grad_norm": 0.46418441159832313,
      "learning_rate": 6.831501831501832e-06,
      "loss": 0.5429,
      "step": 374
    },
    {
      "epoch": 0.06870648589226823,
      "grad_norm": 0.45474945807496187,
      "learning_rate": 6.84981684981685e-06,
      "loss": 0.4999,
      "step": 375
    },
    {
      "epoch": 0.06888970318798095,
      "grad_norm": 0.5576640330118253,
      "learning_rate": 6.868131868131869e-06,
      "loss": 0.5198,
      "step": 376
    },
    {
      "epoch": 0.06907292048369366,
      "grad_norm": 0.4921371781399045,
      "learning_rate": 6.886446886446887e-06,
      "loss": 0.5379,
      "step": 377
    },
    {
      "epoch": 0.06925613777940638,
      "grad_norm": 0.41639097386056484,
      "learning_rate": 6.9047619047619055e-06,
      "loss": 0.519,
      "step": 378
    },
    {
      "epoch": 0.06943935507511909,
      "grad_norm": 0.4778909271586564,
      "learning_rate": 6.923076923076923e-06,
      "loss": 0.5336,
      "step": 379
    },
    {
      "epoch": 0.0696225723708318,
      "grad_norm": 0.4735414706699025,
      "learning_rate": 6.941391941391942e-06,
      "loss": 0.5,
      "step": 380
    },
    {
      "epoch": 0.06980578966654452,
      "grad_norm": 0.5646322357431656,
      "learning_rate": 6.95970695970696e-06,
      "loss": 0.5073,
      "step": 381
    },
    {
      "epoch": 0.06998900696225724,
      "grad_norm": 0.48231469572609614,
      "learning_rate": 6.978021978021979e-06,
      "loss": 0.519,
      "step": 382
    },
    {
      "epoch": 0.07017222425796996,
      "grad_norm": 0.5051213072897166,
      "learning_rate": 6.9963369963369965e-06,
      "loss": 0.5315,
      "step": 383
    },
    {
      "epoch": 0.07035544155368267,
      "grad_norm": 0.5166250853237789,
      "learning_rate": 7.014652014652014e-06,
      "loss": 0.4978,
      "step": 384
    },
    {
      "epoch": 0.07053865884939538,
      "grad_norm": 0.43900982727967125,
      "learning_rate": 7.032967032967034e-06,
      "loss": 0.4913,
      "step": 385
    },
    {
      "epoch": 0.0707218761451081,
      "grad_norm": 0.5262819691743885,
      "learning_rate": 7.051282051282053e-06,
      "loss": 0.5189,
      "step": 386
    },
    {
      "epoch": 0.07090509344082081,
      "grad_norm": 0.4906825009780104,
      "learning_rate": 7.0695970695970705e-06,
      "loss": 0.5586,
      "step": 387
    },
    {
      "epoch": 0.07108831073653353,
      "grad_norm": 0.48825547184252527,
      "learning_rate": 7.087912087912089e-06,
      "loss": 0.5218,
      "step": 388
    },
    {
      "epoch": 0.07127152803224625,
      "grad_norm": 0.5005437189224704,
      "learning_rate": 7.106227106227107e-06,
      "loss": 0.5088,
      "step": 389
    },
    {
      "epoch": 0.07145474532795897,
      "grad_norm": 0.4820090004987099,
      "learning_rate": 7.124542124542126e-06,
      "loss": 0.5132,
      "step": 390
    },
    {
      "epoch": 0.07163796262367167,
      "grad_norm": 0.46585258023856246,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.5198,
      "step": 391
    },
    {
      "epoch": 0.07182117991938439,
      "grad_norm": 0.5351346796257885,
      "learning_rate": 7.161172161172162e-06,
      "loss": 0.5174,
      "step": 392
    },
    {
      "epoch": 0.0720043972150971,
      "grad_norm": 0.505980784032641,
      "learning_rate": 7.17948717948718e-06,
      "loss": 0.517,
      "step": 393
    },
    {
      "epoch": 0.07218761451080982,
      "grad_norm": 0.3991493735258601,
      "learning_rate": 7.197802197802198e-06,
      "loss": 0.5193,
      "step": 394
    },
    {
      "epoch": 0.07237083180652254,
      "grad_norm": 0.47050897190858826,
      "learning_rate": 7.216117216117217e-06,
      "loss": 0.5278,
      "step": 395
    },
    {
      "epoch": 0.07255404910223526,
      "grad_norm": 0.42709223237684557,
      "learning_rate": 7.2344322344322345e-06,
      "loss": 0.5233,
      "step": 396
    },
    {
      "epoch": 0.07273726639794796,
      "grad_norm": 0.40701893214503493,
      "learning_rate": 7.252747252747253e-06,
      "loss": 0.4594,
      "step": 397
    },
    {
      "epoch": 0.07292048369366068,
      "grad_norm": 0.4793723507064191,
      "learning_rate": 7.271062271062271e-06,
      "loss": 0.5022,
      "step": 398
    },
    {
      "epoch": 0.0731037009893734,
      "grad_norm": 0.4084855491847111,
      "learning_rate": 7.28937728937729e-06,
      "loss": 0.5177,
      "step": 399
    },
    {
      "epoch": 0.07328691828508611,
      "grad_norm": 0.48791204362548757,
      "learning_rate": 7.307692307692308e-06,
      "loss": 0.5336,
      "step": 400
    },
    {
      "epoch": 0.07347013558079883,
      "grad_norm": 0.4721917787188149,
      "learning_rate": 7.326007326007326e-06,
      "loss": 0.5291,
      "step": 401
    },
    {
      "epoch": 0.07365335287651155,
      "grad_norm": 0.4482564665450733,
      "learning_rate": 7.344322344322346e-06,
      "loss": 0.516,
      "step": 402
    },
    {
      "epoch": 0.07383657017222425,
      "grad_norm": 0.503350703275625,
      "learning_rate": 7.362637362637364e-06,
      "loss": 0.5232,
      "step": 403
    },
    {
      "epoch": 0.07401978746793697,
      "grad_norm": 0.5152368271826678,
      "learning_rate": 7.380952380952382e-06,
      "loss": 0.5258,
      "step": 404
    },
    {
      "epoch": 0.07420300476364969,
      "grad_norm": 0.4523690267979874,
      "learning_rate": 7.3992673992674e-06,
      "loss": 0.5007,
      "step": 405
    },
    {
      "epoch": 0.0743862220593624,
      "grad_norm": 0.44122338596098665,
      "learning_rate": 7.417582417582418e-06,
      "loss": 0.5325,
      "step": 406
    },
    {
      "epoch": 0.07456943935507512,
      "grad_norm": 0.418021168922302,
      "learning_rate": 7.435897435897437e-06,
      "loss": 0.5048,
      "step": 407
    },
    {
      "epoch": 0.07475265665078784,
      "grad_norm": 0.44537098554090276,
      "learning_rate": 7.454212454212455e-06,
      "loss": 0.5294,
      "step": 408
    },
    {
      "epoch": 0.07493587394650054,
      "grad_norm": 0.41384101894590325,
      "learning_rate": 7.472527472527473e-06,
      "loss": 0.5119,
      "step": 409
    },
    {
      "epoch": 0.07511909124221326,
      "grad_norm": 0.4647470180215214,
      "learning_rate": 7.490842490842491e-06,
      "loss": 0.5202,
      "step": 410
    },
    {
      "epoch": 0.07530230853792598,
      "grad_norm": 0.4836409216489063,
      "learning_rate": 7.50915750915751e-06,
      "loss": 0.5056,
      "step": 411
    },
    {
      "epoch": 0.0754855258336387,
      "grad_norm": 0.45195196209865357,
      "learning_rate": 7.527472527472528e-06,
      "loss": 0.5213,
      "step": 412
    },
    {
      "epoch": 0.07566874312935142,
      "grad_norm": 0.5036829282617328,
      "learning_rate": 7.5457875457875465e-06,
      "loss": 0.5347,
      "step": 413
    },
    {
      "epoch": 0.07585196042506413,
      "grad_norm": 0.4666038283310089,
      "learning_rate": 7.564102564102564e-06,
      "loss": 0.5151,
      "step": 414
    },
    {
      "epoch": 0.07603517772077684,
      "grad_norm": 0.5067769344340564,
      "learning_rate": 7.582417582417583e-06,
      "loss": 0.5171,
      "step": 415
    },
    {
      "epoch": 0.07621839501648955,
      "grad_norm": 0.4093230258713508,
      "learning_rate": 7.600732600732601e-06,
      "loss": 0.5039,
      "step": 416
    },
    {
      "epoch": 0.07640161231220227,
      "grad_norm": 0.6498006485071868,
      "learning_rate": 7.61904761904762e-06,
      "loss": 0.5232,
      "step": 417
    },
    {
      "epoch": 0.07658482960791499,
      "grad_norm": 0.4558152414900924,
      "learning_rate": 7.637362637362638e-06,
      "loss": 0.5232,
      "step": 418
    },
    {
      "epoch": 0.0767680469036277,
      "grad_norm": 0.4804930639066785,
      "learning_rate": 7.655677655677656e-06,
      "loss": 0.5088,
      "step": 419
    },
    {
      "epoch": 0.07695126419934042,
      "grad_norm": 0.5081120095319347,
      "learning_rate": 7.673992673992676e-06,
      "loss": 0.5308,
      "step": 420
    },
    {
      "epoch": 0.07713448149505313,
      "grad_norm": 0.42393212923875745,
      "learning_rate": 7.692307692307694e-06,
      "loss": 0.5084,
      "step": 421
    },
    {
      "epoch": 0.07731769879076585,
      "grad_norm": 0.44672075335716155,
      "learning_rate": 7.710622710622711e-06,
      "loss": 0.483,
      "step": 422
    },
    {
      "epoch": 0.07750091608647856,
      "grad_norm": 0.4942211324726324,
      "learning_rate": 7.72893772893773e-06,
      "loss": 0.5236,
      "step": 423
    },
    {
      "epoch": 0.07768413338219128,
      "grad_norm": 0.39817279842841763,
      "learning_rate": 7.747252747252749e-06,
      "loss": 0.5332,
      "step": 424
    },
    {
      "epoch": 0.077867350677904,
      "grad_norm": 0.4608671155196199,
      "learning_rate": 7.765567765567767e-06,
      "loss": 0.523,
      "step": 425
    },
    {
      "epoch": 0.07805056797361672,
      "grad_norm": 0.5225991254379231,
      "learning_rate": 7.783882783882785e-06,
      "loss": 0.5171,
      "step": 426
    },
    {
      "epoch": 0.07823378526932942,
      "grad_norm": 0.5423490196220466,
      "learning_rate": 7.802197802197802e-06,
      "loss": 0.4985,
      "step": 427
    },
    {
      "epoch": 0.07841700256504214,
      "grad_norm": 0.502419506962013,
      "learning_rate": 7.820512820512822e-06,
      "loss": 0.5087,
      "step": 428
    },
    {
      "epoch": 0.07860021986075486,
      "grad_norm": 0.5102984182578104,
      "learning_rate": 7.83882783882784e-06,
      "loss": 0.5358,
      "step": 429
    },
    {
      "epoch": 0.07878343715646757,
      "grad_norm": 0.4742126281062651,
      "learning_rate": 7.857142857142858e-06,
      "loss": 0.5585,
      "step": 430
    },
    {
      "epoch": 0.07896665445218029,
      "grad_norm": 0.508787409495283,
      "learning_rate": 7.875457875457876e-06,
      "loss": 0.5411,
      "step": 431
    },
    {
      "epoch": 0.079149871747893,
      "grad_norm": 0.4559258370924022,
      "learning_rate": 7.893772893772893e-06,
      "loss": 0.5229,
      "step": 432
    },
    {
      "epoch": 0.07933308904360571,
      "grad_norm": 0.6320276655690735,
      "learning_rate": 7.912087912087913e-06,
      "loss": 0.4688,
      "step": 433
    },
    {
      "epoch": 0.07951630633931843,
      "grad_norm": 0.9797165625621964,
      "learning_rate": 7.93040293040293e-06,
      "loss": 0.5372,
      "step": 434
    },
    {
      "epoch": 0.07969952363503115,
      "grad_norm": 0.43648526458810954,
      "learning_rate": 7.948717948717949e-06,
      "loss": 0.4528,
      "step": 435
    },
    {
      "epoch": 0.07988274093074386,
      "grad_norm": 0.46101717931557235,
      "learning_rate": 7.967032967032966e-06,
      "loss": 0.4836,
      "step": 436
    },
    {
      "epoch": 0.08006595822645658,
      "grad_norm": 0.46019219132505085,
      "learning_rate": 7.985347985347986e-06,
      "loss": 0.4874,
      "step": 437
    },
    {
      "epoch": 0.08024917552216929,
      "grad_norm": 0.4228905699110607,
      "learning_rate": 8.003663003663006e-06,
      "loss": 0.483,
      "step": 438
    },
    {
      "epoch": 0.080432392817882,
      "grad_norm": 0.4920935432281776,
      "learning_rate": 8.021978021978023e-06,
      "loss": 0.489,
      "step": 439
    },
    {
      "epoch": 0.08061561011359472,
      "grad_norm": 0.45385267234753407,
      "learning_rate": 8.040293040293041e-06,
      "loss": 0.5262,
      "step": 440
    },
    {
      "epoch": 0.08079882740930744,
      "grad_norm": 0.48390990285633756,
      "learning_rate": 8.058608058608059e-06,
      "loss": 0.4969,
      "step": 441
    },
    {
      "epoch": 0.08098204470502016,
      "grad_norm": 0.4424118373260341,
      "learning_rate": 8.076923076923077e-06,
      "loss": 0.5217,
      "step": 442
    },
    {
      "epoch": 0.08116526200073287,
      "grad_norm": 0.4647327426734162,
      "learning_rate": 8.095238095238097e-06,
      "loss": 0.5103,
      "step": 443
    },
    {
      "epoch": 0.08134847929644558,
      "grad_norm": 0.42019112334691044,
      "learning_rate": 8.113553113553114e-06,
      "loss": 0.5378,
      "step": 444
    },
    {
      "epoch": 0.0815316965921583,
      "grad_norm": 0.40127483204178854,
      "learning_rate": 8.131868131868132e-06,
      "loss": 0.4813,
      "step": 445
    },
    {
      "epoch": 0.08171491388787101,
      "grad_norm": 0.4606579465800615,
      "learning_rate": 8.15018315018315e-06,
      "loss": 0.5067,
      "step": 446
    },
    {
      "epoch": 0.08189813118358373,
      "grad_norm": 0.4715535246345655,
      "learning_rate": 8.16849816849817e-06,
      "loss": 0.5181,
      "step": 447
    },
    {
      "epoch": 0.08208134847929645,
      "grad_norm": 0.5386430386856308,
      "learning_rate": 8.186813186813188e-06,
      "loss": 0.5188,
      "step": 448
    },
    {
      "epoch": 0.08226456577500917,
      "grad_norm": 0.5389030998275688,
      "learning_rate": 8.205128205128205e-06,
      "loss": 0.516,
      "step": 449
    },
    {
      "epoch": 0.08244778307072187,
      "grad_norm": 0.5181954163327931,
      "learning_rate": 8.223443223443223e-06,
      "loss": 0.5099,
      "step": 450
    },
    {
      "epoch": 0.08263100036643459,
      "grad_norm": 0.5425277965290457,
      "learning_rate": 8.241758241758243e-06,
      "loss": 0.5508,
      "step": 451
    },
    {
      "epoch": 0.0828142176621473,
      "grad_norm": 0.4366152129754919,
      "learning_rate": 8.26007326007326e-06,
      "loss": 0.4772,
      "step": 452
    },
    {
      "epoch": 0.08299743495786002,
      "grad_norm": 0.45607475441878936,
      "learning_rate": 8.278388278388278e-06,
      "loss": 0.5319,
      "step": 453
    },
    {
      "epoch": 0.08318065225357274,
      "grad_norm": 0.4712644997156548,
      "learning_rate": 8.296703296703298e-06,
      "loss": 0.5477,
      "step": 454
    },
    {
      "epoch": 0.08336386954928546,
      "grad_norm": 0.6856852588986854,
      "learning_rate": 8.315018315018316e-06,
      "loss": 0.5235,
      "step": 455
    },
    {
      "epoch": 0.08354708684499816,
      "grad_norm": 0.5019345602931549,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.5101,
      "step": 456
    },
    {
      "epoch": 0.08373030414071088,
      "grad_norm": 0.4155972768165925,
      "learning_rate": 8.351648351648353e-06,
      "loss": 0.4986,
      "step": 457
    },
    {
      "epoch": 0.0839135214364236,
      "grad_norm": 0.45725631484653695,
      "learning_rate": 8.369963369963371e-06,
      "loss": 0.5233,
      "step": 458
    },
    {
      "epoch": 0.08409673873213631,
      "grad_norm": 0.5500464841633228,
      "learning_rate": 8.388278388278389e-06,
      "loss": 0.5225,
      "step": 459
    },
    {
      "epoch": 0.08427995602784903,
      "grad_norm": 0.5210381864086956,
      "learning_rate": 8.406593406593407e-06,
      "loss": 0.5509,
      "step": 460
    },
    {
      "epoch": 0.08446317332356175,
      "grad_norm": 0.4559821107303101,
      "learning_rate": 8.424908424908426e-06,
      "loss": 0.5118,
      "step": 461
    },
    {
      "epoch": 0.08464639061927445,
      "grad_norm": 0.48323545775926524,
      "learning_rate": 8.443223443223444e-06,
      "loss": 0.5049,
      "step": 462
    },
    {
      "epoch": 0.08482960791498717,
      "grad_norm": 0.4812859935236779,
      "learning_rate": 8.461538461538462e-06,
      "loss": 0.5204,
      "step": 463
    },
    {
      "epoch": 0.08501282521069989,
      "grad_norm": 0.454474341410942,
      "learning_rate": 8.47985347985348e-06,
      "loss": 0.5387,
      "step": 464
    },
    {
      "epoch": 0.0851960425064126,
      "grad_norm": 0.6615650181808158,
      "learning_rate": 8.498168498168498e-06,
      "loss": 0.5393,
      "step": 465
    },
    {
      "epoch": 0.08537925980212532,
      "grad_norm": 0.4355325124699567,
      "learning_rate": 8.516483516483517e-06,
      "loss": 0.5027,
      "step": 466
    },
    {
      "epoch": 0.08556247709783804,
      "grad_norm": 0.43977659231205646,
      "learning_rate": 8.534798534798535e-06,
      "loss": 0.4851,
      "step": 467
    },
    {
      "epoch": 0.08574569439355074,
      "grad_norm": 0.5490378411527629,
      "learning_rate": 8.553113553113553e-06,
      "loss": 0.5087,
      "step": 468
    },
    {
      "epoch": 0.08592891168926346,
      "grad_norm": 0.4334113007101785,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.4874,
      "step": 469
    },
    {
      "epoch": 0.08611212898497618,
      "grad_norm": 0.48093462345535853,
      "learning_rate": 8.58974358974359e-06,
      "loss": 0.5103,
      "step": 470
    },
    {
      "epoch": 0.0862953462806889,
      "grad_norm": 0.509104778628553,
      "learning_rate": 8.60805860805861e-06,
      "loss": 0.5125,
      "step": 471
    },
    {
      "epoch": 0.08647856357640162,
      "grad_norm": 0.46329459466937095,
      "learning_rate": 8.626373626373628e-06,
      "loss": 0.4862,
      "step": 472
    },
    {
      "epoch": 0.08666178087211433,
      "grad_norm": 0.41221910536829426,
      "learning_rate": 8.644688644688646e-06,
      "loss": 0.5058,
      "step": 473
    },
    {
      "epoch": 0.08684499816782704,
      "grad_norm": 0.4965957595147397,
      "learning_rate": 8.663003663003664e-06,
      "loss": 0.5045,
      "step": 474
    },
    {
      "epoch": 0.08702821546353975,
      "grad_norm": 0.6831024596734047,
      "learning_rate": 8.681318681318681e-06,
      "loss": 0.533,
      "step": 475
    },
    {
      "epoch": 0.08721143275925247,
      "grad_norm": 0.44117689667679433,
      "learning_rate": 8.699633699633701e-06,
      "loss": 0.494,
      "step": 476
    },
    {
      "epoch": 0.08739465005496519,
      "grad_norm": 0.49552902332130444,
      "learning_rate": 8.717948717948719e-06,
      "loss": 0.5295,
      "step": 477
    },
    {
      "epoch": 0.08757786735067791,
      "grad_norm": 0.5202864223018397,
      "learning_rate": 8.736263736263737e-06,
      "loss": 0.5315,
      "step": 478
    },
    {
      "epoch": 0.08776108464639062,
      "grad_norm": 0.4531456591891388,
      "learning_rate": 8.754578754578755e-06,
      "loss": 0.5177,
      "step": 479
    },
    {
      "epoch": 0.08794430194210333,
      "grad_norm": 0.5060531781359298,
      "learning_rate": 8.772893772893774e-06,
      "loss": 0.5422,
      "step": 480
    },
    {
      "epoch": 0.08812751923781605,
      "grad_norm": 0.44222377285427994,
      "learning_rate": 8.791208791208792e-06,
      "loss": 0.4679,
      "step": 481
    },
    {
      "epoch": 0.08831073653352876,
      "grad_norm": 0.4308135949373974,
      "learning_rate": 8.80952380952381e-06,
      "loss": 0.5057,
      "step": 482
    },
    {
      "epoch": 0.08849395382924148,
      "grad_norm": 0.5107147323097014,
      "learning_rate": 8.827838827838828e-06,
      "loss": 0.5147,
      "step": 483
    },
    {
      "epoch": 0.0886771711249542,
      "grad_norm": 0.4722359038239213,
      "learning_rate": 8.846153846153847e-06,
      "loss": 0.5069,
      "step": 484
    },
    {
      "epoch": 0.08886038842066692,
      "grad_norm": 0.4913307139634766,
      "learning_rate": 8.864468864468865e-06,
      "loss": 0.4943,
      "step": 485
    },
    {
      "epoch": 0.08904360571637962,
      "grad_norm": 0.47381589638250404,
      "learning_rate": 8.882783882783883e-06,
      "loss": 0.5432,
      "step": 486
    },
    {
      "epoch": 0.08922682301209234,
      "grad_norm": 0.4763727220823097,
      "learning_rate": 8.9010989010989e-06,
      "loss": 0.4765,
      "step": 487
    },
    {
      "epoch": 0.08941004030780506,
      "grad_norm": 0.48581942940784734,
      "learning_rate": 8.91941391941392e-06,
      "loss": 0.4918,
      "step": 488
    },
    {
      "epoch": 0.08959325760351777,
      "grad_norm": 0.4771435540271998,
      "learning_rate": 8.937728937728938e-06,
      "loss": 0.5146,
      "step": 489
    },
    {
      "epoch": 0.08977647489923049,
      "grad_norm": 0.5554558590150881,
      "learning_rate": 8.956043956043958e-06,
      "loss": 0.5139,
      "step": 490
    },
    {
      "epoch": 0.08995969219494321,
      "grad_norm": 0.4409852140988399,
      "learning_rate": 8.974358974358976e-06,
      "loss": 0.494,
      "step": 491
    },
    {
      "epoch": 0.09014290949065591,
      "grad_norm": 0.470617109732078,
      "learning_rate": 8.992673992673993e-06,
      "loss": 0.5293,
      "step": 492
    },
    {
      "epoch": 0.09032612678636863,
      "grad_norm": 0.48719044521659705,
      "learning_rate": 9.010989010989011e-06,
      "loss": 0.5111,
      "step": 493
    },
    {
      "epoch": 0.09050934408208135,
      "grad_norm": 0.46876439756978366,
      "learning_rate": 9.02930402930403e-06,
      "loss": 0.4942,
      "step": 494
    },
    {
      "epoch": 0.09069256137779406,
      "grad_norm": 0.46671821534033575,
      "learning_rate": 9.047619047619049e-06,
      "loss": 0.5617,
      "step": 495
    },
    {
      "epoch": 0.09087577867350678,
      "grad_norm": 0.4868372988815388,
      "learning_rate": 9.065934065934067e-06,
      "loss": 0.5195,
      "step": 496
    },
    {
      "epoch": 0.0910589959692195,
      "grad_norm": 0.47526685976460153,
      "learning_rate": 9.084249084249084e-06,
      "loss": 0.4561,
      "step": 497
    },
    {
      "epoch": 0.0912422132649322,
      "grad_norm": 0.526795639153496,
      "learning_rate": 9.102564102564104e-06,
      "loss": 0.4905,
      "step": 498
    },
    {
      "epoch": 0.09142543056064492,
      "grad_norm": 0.47935527575760656,
      "learning_rate": 9.120879120879122e-06,
      "loss": 0.5129,
      "step": 499
    },
    {
      "epoch": 0.09160864785635764,
      "grad_norm": 0.47020582683735346,
      "learning_rate": 9.13919413919414e-06,
      "loss": 0.511,
      "step": 500
    },
    {
      "epoch": 0.09179186515207036,
      "grad_norm": 0.46760420885699033,
      "learning_rate": 9.157509157509158e-06,
      "loss": 0.4789,
      "step": 501
    },
    {
      "epoch": 0.09197508244778307,
      "grad_norm": 0.4695251283934706,
      "learning_rate": 9.175824175824175e-06,
      "loss": 0.5351,
      "step": 502
    },
    {
      "epoch": 0.09215829974349579,
      "grad_norm": 0.5153360795669553,
      "learning_rate": 9.194139194139195e-06,
      "loss": 0.5522,
      "step": 503
    },
    {
      "epoch": 0.0923415170392085,
      "grad_norm": 0.44806081150815513,
      "learning_rate": 9.212454212454213e-06,
      "loss": 0.5173,
      "step": 504
    },
    {
      "epoch": 0.09252473433492121,
      "grad_norm": 0.451536504838503,
      "learning_rate": 9.230769230769232e-06,
      "loss": 0.5082,
      "step": 505
    },
    {
      "epoch": 0.09270795163063393,
      "grad_norm": 0.48436384147084766,
      "learning_rate": 9.24908424908425e-06,
      "loss": 0.5388,
      "step": 506
    },
    {
      "epoch": 0.09289116892634665,
      "grad_norm": 0.4609113225702784,
      "learning_rate": 9.267399267399268e-06,
      "loss": 0.4892,
      "step": 507
    },
    {
      "epoch": 0.09307438622205937,
      "grad_norm": 0.4098711492197863,
      "learning_rate": 9.285714285714288e-06,
      "loss": 0.4856,
      "step": 508
    },
    {
      "epoch": 0.09325760351777208,
      "grad_norm": 0.4715675907322317,
      "learning_rate": 9.304029304029305e-06,
      "loss": 0.53,
      "step": 509
    },
    {
      "epoch": 0.09344082081348479,
      "grad_norm": 0.5026574189450673,
      "learning_rate": 9.322344322344323e-06,
      "loss": 0.5314,
      "step": 510
    },
    {
      "epoch": 0.0936240381091975,
      "grad_norm": 0.4253199666429807,
      "learning_rate": 9.340659340659341e-06,
      "loss": 0.5068,
      "step": 511
    },
    {
      "epoch": 0.09380725540491022,
      "grad_norm": 0.45894485399853974,
      "learning_rate": 9.358974358974359e-06,
      "loss": 0.5235,
      "step": 512
    },
    {
      "epoch": 0.09399047270062294,
      "grad_norm": 0.7513225377427756,
      "learning_rate": 9.377289377289379e-06,
      "loss": 0.5071,
      "step": 513
    },
    {
      "epoch": 0.09417368999633566,
      "grad_norm": 0.5231017027547264,
      "learning_rate": 9.395604395604396e-06,
      "loss": 0.5339,
      "step": 514
    },
    {
      "epoch": 0.09435690729204838,
      "grad_norm": 0.5146997140217462,
      "learning_rate": 9.413919413919414e-06,
      "loss": 0.5052,
      "step": 515
    },
    {
      "epoch": 0.09454012458776108,
      "grad_norm": 0.47760493192836756,
      "learning_rate": 9.432234432234432e-06,
      "loss": 0.5102,
      "step": 516
    },
    {
      "epoch": 0.0947233418834738,
      "grad_norm": 0.44117590454697736,
      "learning_rate": 9.450549450549452e-06,
      "loss": 0.5237,
      "step": 517
    },
    {
      "epoch": 0.09490655917918651,
      "grad_norm": 0.5004680185091399,
      "learning_rate": 9.46886446886447e-06,
      "loss": 0.5243,
      "step": 518
    },
    {
      "epoch": 0.09508977647489923,
      "grad_norm": 0.4549800276149351,
      "learning_rate": 9.487179487179487e-06,
      "loss": 0.5139,
      "step": 519
    },
    {
      "epoch": 0.09527299377061195,
      "grad_norm": 0.4263327588755031,
      "learning_rate": 9.505494505494505e-06,
      "loss": 0.4877,
      "step": 520
    },
    {
      "epoch": 0.09545621106632467,
      "grad_norm": 0.4484968491218323,
      "learning_rate": 9.523809523809525e-06,
      "loss": 0.4986,
      "step": 521
    },
    {
      "epoch": 0.09563942836203737,
      "grad_norm": 0.39390345192807547,
      "learning_rate": 9.542124542124543e-06,
      "loss": 0.5116,
      "step": 522
    },
    {
      "epoch": 0.09582264565775009,
      "grad_norm": 0.4529147798983982,
      "learning_rate": 9.560439560439562e-06,
      "loss": 0.5312,
      "step": 523
    },
    {
      "epoch": 0.0960058629534628,
      "grad_norm": 0.49701803503982106,
      "learning_rate": 9.57875457875458e-06,
      "loss": 0.529,
      "step": 524
    },
    {
      "epoch": 0.09618908024917552,
      "grad_norm": 0.4097214793479841,
      "learning_rate": 9.597069597069598e-06,
      "loss": 0.5005,
      "step": 525
    },
    {
      "epoch": 0.09637229754488824,
      "grad_norm": 0.5148151884655592,
      "learning_rate": 9.615384615384616e-06,
      "loss": 0.53,
      "step": 526
    },
    {
      "epoch": 0.09655551484060096,
      "grad_norm": 0.4264857545219357,
      "learning_rate": 9.633699633699635e-06,
      "loss": 0.5136,
      "step": 527
    },
    {
      "epoch": 0.09673873213631366,
      "grad_norm": 0.45426464252638443,
      "learning_rate": 9.652014652014653e-06,
      "loss": 0.5117,
      "step": 528
    },
    {
      "epoch": 0.09692194943202638,
      "grad_norm": 0.47571176071645493,
      "learning_rate": 9.670329670329671e-06,
      "loss": 0.5185,
      "step": 529
    },
    {
      "epoch": 0.0971051667277391,
      "grad_norm": 0.42418752426373113,
      "learning_rate": 9.688644688644689e-06,
      "loss": 0.5285,
      "step": 530
    },
    {
      "epoch": 0.09728838402345182,
      "grad_norm": 0.48590348616099827,
      "learning_rate": 9.706959706959708e-06,
      "loss": 0.5339,
      "step": 531
    },
    {
      "epoch": 0.09747160131916453,
      "grad_norm": 0.463774513888958,
      "learning_rate": 9.725274725274726e-06,
      "loss": 0.5386,
      "step": 532
    },
    {
      "epoch": 0.09765481861487725,
      "grad_norm": 0.4891808396070614,
      "learning_rate": 9.743589743589744e-06,
      "loss": 0.5326,
      "step": 533
    },
    {
      "epoch": 0.09783803591058995,
      "grad_norm": 0.40841004082693305,
      "learning_rate": 9.761904761904762e-06,
      "loss": 0.5085,
      "step": 534
    },
    {
      "epoch": 0.09802125320630267,
      "grad_norm": 0.42494728872814735,
      "learning_rate": 9.780219780219781e-06,
      "loss": 0.4751,
      "step": 535
    },
    {
      "epoch": 0.09820447050201539,
      "grad_norm": 0.4594283822364781,
      "learning_rate": 9.7985347985348e-06,
      "loss": 0.5275,
      "step": 536
    },
    {
      "epoch": 0.09838768779772811,
      "grad_norm": 0.4064302259117676,
      "learning_rate": 9.816849816849817e-06,
      "loss": 0.5089,
      "step": 537
    },
    {
      "epoch": 0.09857090509344082,
      "grad_norm": 0.4860027724973198,
      "learning_rate": 9.835164835164835e-06,
      "loss": 0.505,
      "step": 538
    },
    {
      "epoch": 0.09875412238915354,
      "grad_norm": 0.47625362870716265,
      "learning_rate": 9.853479853479855e-06,
      "loss": 0.4954,
      "step": 539
    },
    {
      "epoch": 0.09893733968486625,
      "grad_norm": 0.4592940212317861,
      "learning_rate": 9.871794871794872e-06,
      "loss": 0.4871,
      "step": 540
    },
    {
      "epoch": 0.09912055698057896,
      "grad_norm": 0.42717127053329573,
      "learning_rate": 9.890109890109892e-06,
      "loss": 0.5261,
      "step": 541
    },
    {
      "epoch": 0.09930377427629168,
      "grad_norm": 0.43970535809042904,
      "learning_rate": 9.90842490842491e-06,
      "loss": 0.4866,
      "step": 542
    },
    {
      "epoch": 0.0994869915720044,
      "grad_norm": 0.42143360797263807,
      "learning_rate": 9.926739926739928e-06,
      "loss": 0.5113,
      "step": 543
    },
    {
      "epoch": 0.09967020886771712,
      "grad_norm": 0.4416711016318173,
      "learning_rate": 9.945054945054946e-06,
      "loss": 0.5116,
      "step": 544
    },
    {
      "epoch": 0.09985342616342983,
      "grad_norm": 0.45325267149265236,
      "learning_rate": 9.963369963369965e-06,
      "loss": 0.5283,
      "step": 545
    },
    {
      "epoch": 0.10003664345914254,
      "grad_norm": 0.3948707773420248,
      "learning_rate": 9.981684981684983e-06,
      "loss": 0.5114,
      "step": 546
    },
    {
      "epoch": 0.10021986075485526,
      "grad_norm": 0.42983426021717475,
      "learning_rate": 1e-05,
      "loss": 0.5169,
      "step": 547
    },
    {
      "epoch": 0.10040307805056797,
      "grad_norm": 0.5119731461361727,
      "learning_rate": 9.999998977359419e-06,
      "loss": 0.518,
      "step": 548
    },
    {
      "epoch": 0.10058629534628069,
      "grad_norm": 0.48479775231872485,
      "learning_rate": 9.999995909438092e-06,
      "loss": 0.5207,
      "step": 549
    },
    {
      "epoch": 0.10076951264199341,
      "grad_norm": 0.45005420614602115,
      "learning_rate": 9.999990796237274e-06,
      "loss": 0.4878,
      "step": 550
    },
    {
      "epoch": 0.10095272993770613,
      "grad_norm": 0.40595587220813967,
      "learning_rate": 9.999983637759059e-06,
      "loss": 0.4712,
      "step": 551
    },
    {
      "epoch": 0.10113594723341883,
      "grad_norm": 0.43893870910937327,
      "learning_rate": 9.999974434006372e-06,
      "loss": 0.4964,
      "step": 552
    },
    {
      "epoch": 0.10131916452913155,
      "grad_norm": 0.4638079942473788,
      "learning_rate": 9.99996318498298e-06,
      "loss": 0.4834,
      "step": 553
    },
    {
      "epoch": 0.10150238182484426,
      "grad_norm": 0.45078724585599195,
      "learning_rate": 9.999949890693484e-06,
      "loss": 0.4828,
      "step": 554
    },
    {
      "epoch": 0.10168559912055698,
      "grad_norm": 0.4968230440871285,
      "learning_rate": 9.999934551143319e-06,
      "loss": 0.5146,
      "step": 555
    },
    {
      "epoch": 0.1018688164162697,
      "grad_norm": 0.4431824586567993,
      "learning_rate": 9.999917166338767e-06,
      "loss": 0.5117,
      "step": 556
    },
    {
      "epoch": 0.10205203371198242,
      "grad_norm": 0.40809319706771474,
      "learning_rate": 9.999897736286932e-06,
      "loss": 0.4942,
      "step": 557
    },
    {
      "epoch": 0.10223525100769512,
      "grad_norm": 0.46904649406632837,
      "learning_rate": 9.999876260995767e-06,
      "loss": 0.5223,
      "step": 558
    },
    {
      "epoch": 0.10241846830340784,
      "grad_norm": 0.45993616230767786,
      "learning_rate": 9.999852740474054e-06,
      "loss": 0.498,
      "step": 559
    },
    {
      "epoch": 0.10260168559912056,
      "grad_norm": 0.41328723379038895,
      "learning_rate": 9.999827174731414e-06,
      "loss": 0.5294,
      "step": 560
    },
    {
      "epoch": 0.10278490289483327,
      "grad_norm": 0.4584774156426854,
      "learning_rate": 9.999799563778307e-06,
      "loss": 0.5138,
      "step": 561
    },
    {
      "epoch": 0.10296812019054599,
      "grad_norm": 0.4626480547701558,
      "learning_rate": 9.999769907626024e-06,
      "loss": 0.5122,
      "step": 562
    },
    {
      "epoch": 0.10315133748625871,
      "grad_norm": 0.4647555669122866,
      "learning_rate": 9.9997382062867e-06,
      "loss": 0.5276,
      "step": 563
    },
    {
      "epoch": 0.10333455478197141,
      "grad_norm": 0.41824885855037686,
      "learning_rate": 9.9997044597733e-06,
      "loss": 0.4976,
      "step": 564
    },
    {
      "epoch": 0.10351777207768413,
      "grad_norm": 0.41187305924650414,
      "learning_rate": 9.999668668099628e-06,
      "loss": 0.4953,
      "step": 565
    },
    {
      "epoch": 0.10370098937339685,
      "grad_norm": 0.42660595756816655,
      "learning_rate": 9.999630831280329e-06,
      "loss": 0.514,
      "step": 566
    },
    {
      "epoch": 0.10388420666910957,
      "grad_norm": 0.5099786508796531,
      "learning_rate": 9.999590949330876e-06,
      "loss": 0.5038,
      "step": 567
    },
    {
      "epoch": 0.10406742396482228,
      "grad_norm": 0.4926825490754442,
      "learning_rate": 9.999549022267582e-06,
      "loss": 0.5535,
      "step": 568
    },
    {
      "epoch": 0.104250641260535,
      "grad_norm": 0.4578490443823955,
      "learning_rate": 9.999505050107601e-06,
      "loss": 0.5075,
      "step": 569
    },
    {
      "epoch": 0.1044338585562477,
      "grad_norm": 0.4733323605035433,
      "learning_rate": 9.99945903286892e-06,
      "loss": 0.4747,
      "step": 570
    },
    {
      "epoch": 0.10461707585196042,
      "grad_norm": 0.44893796479946446,
      "learning_rate": 9.999410970570358e-06,
      "loss": 0.4711,
      "step": 571
    },
    {
      "epoch": 0.10480029314767314,
      "grad_norm": 0.5254914900315981,
      "learning_rate": 9.99936086323158e-06,
      "loss": 0.5358,
      "step": 572
    },
    {
      "epoch": 0.10498351044338586,
      "grad_norm": 0.537923540220477,
      "learning_rate": 9.999308710873083e-06,
      "loss": 0.5079,
      "step": 573
    },
    {
      "epoch": 0.10516672773909858,
      "grad_norm": 0.45536567675754,
      "learning_rate": 9.999254513516196e-06,
      "loss": 0.4908,
      "step": 574
    },
    {
      "epoch": 0.10534994503481128,
      "grad_norm": 0.38323869490085266,
      "learning_rate": 9.999198271183094e-06,
      "loss": 0.4983,
      "step": 575
    },
    {
      "epoch": 0.105533162330524,
      "grad_norm": 0.43580085679521197,
      "learning_rate": 9.999139983896779e-06,
      "loss": 0.5031,
      "step": 576
    },
    {
      "epoch": 0.10571637962623671,
      "grad_norm": 0.4193926406628939,
      "learning_rate": 9.999079651681096e-06,
      "loss": 0.5026,
      "step": 577
    },
    {
      "epoch": 0.10589959692194943,
      "grad_norm": 0.5149327291476714,
      "learning_rate": 9.999017274560722e-06,
      "loss": 0.5414,
      "step": 578
    },
    {
      "epoch": 0.10608281421766215,
      "grad_norm": 0.44098660936462153,
      "learning_rate": 9.998952852561176e-06,
      "loss": 0.5032,
      "step": 579
    },
    {
      "epoch": 0.10626603151337487,
      "grad_norm": 0.4413320145497695,
      "learning_rate": 9.998886385708807e-06,
      "loss": 0.497,
      "step": 580
    },
    {
      "epoch": 0.10644924880908757,
      "grad_norm": 0.45575842795199323,
      "learning_rate": 9.998817874030808e-06,
      "loss": 0.5207,
      "step": 581
    },
    {
      "epoch": 0.10663246610480029,
      "grad_norm": 0.532654166955093,
      "learning_rate": 9.9987473175552e-06,
      "loss": 0.5212,
      "step": 582
    },
    {
      "epoch": 0.106815683400513,
      "grad_norm": 0.4157619606008366,
      "learning_rate": 9.998674716310846e-06,
      "loss": 0.5085,
      "step": 583
    },
    {
      "epoch": 0.10699890069622572,
      "grad_norm": 0.4868388397165033,
      "learning_rate": 9.998600070327444e-06,
      "loss": 0.5306,
      "step": 584
    },
    {
      "epoch": 0.10718211799193844,
      "grad_norm": 0.49401937562080056,
      "learning_rate": 9.998523379635527e-06,
      "loss": 0.5429,
      "step": 585
    },
    {
      "epoch": 0.10736533528765116,
      "grad_norm": 0.3986626914030523,
      "learning_rate": 9.99844464426647e-06,
      "loss": 0.4695,
      "step": 586
    },
    {
      "epoch": 0.10754855258336386,
      "grad_norm": 0.49524254606268836,
      "learning_rate": 9.998363864252474e-06,
      "loss": 0.5226,
      "step": 587
    },
    {
      "epoch": 0.10773176987907658,
      "grad_norm": 0.46186447850660334,
      "learning_rate": 9.998281039626588e-06,
      "loss": 0.5294,
      "step": 588
    },
    {
      "epoch": 0.1079149871747893,
      "grad_norm": 0.4207082230286084,
      "learning_rate": 9.99819617042269e-06,
      "loss": 0.5096,
      "step": 589
    },
    {
      "epoch": 0.10809820447050202,
      "grad_norm": 0.3955565788721964,
      "learning_rate": 9.998109256675496e-06,
      "loss": 0.4931,
      "step": 590
    },
    {
      "epoch": 0.10828142176621473,
      "grad_norm": 0.36723747707336185,
      "learning_rate": 9.998020298420559e-06,
      "loss": 0.4964,
      "step": 591
    },
    {
      "epoch": 0.10846463906192745,
      "grad_norm": 0.4185544526751163,
      "learning_rate": 9.997929295694266e-06,
      "loss": 0.4801,
      "step": 592
    },
    {
      "epoch": 0.10864785635764015,
      "grad_norm": 0.42948593775837285,
      "learning_rate": 9.997836248533845e-06,
      "loss": 0.5197,
      "step": 593
    },
    {
      "epoch": 0.10883107365335287,
      "grad_norm": 0.43966759837587943,
      "learning_rate": 9.997741156977356e-06,
      "loss": 0.5254,
      "step": 594
    },
    {
      "epoch": 0.10901429094906559,
      "grad_norm": 0.4632445570564666,
      "learning_rate": 9.997644021063698e-06,
      "loss": 0.4802,
      "step": 595
    },
    {
      "epoch": 0.10919750824477831,
      "grad_norm": 0.46480621716730125,
      "learning_rate": 9.997544840832604e-06,
      "loss": 0.5219,
      "step": 596
    },
    {
      "epoch": 0.10938072554049103,
      "grad_norm": 0.44567173311984926,
      "learning_rate": 9.997443616324645e-06,
      "loss": 0.5105,
      "step": 597
    },
    {
      "epoch": 0.10956394283620374,
      "grad_norm": 0.4534263401515789,
      "learning_rate": 9.997340347581226e-06,
      "loss": 0.5212,
      "step": 598
    },
    {
      "epoch": 0.10974716013191645,
      "grad_norm": 0.40359525147680225,
      "learning_rate": 9.99723503464459e-06,
      "loss": 0.5261,
      "step": 599
    },
    {
      "epoch": 0.10993037742762916,
      "grad_norm": 0.8305073137318505,
      "learning_rate": 9.99712767755782e-06,
      "loss": 0.5173,
      "step": 600
    },
    {
      "epoch": 0.11011359472334188,
      "grad_norm": 0.5039765167575108,
      "learning_rate": 9.997018276364825e-06,
      "loss": 0.4983,
      "step": 601
    },
    {
      "epoch": 0.1102968120190546,
      "grad_norm": 0.4000444070826362,
      "learning_rate": 9.99690683111036e-06,
      "loss": 0.4876,
      "step": 602
    },
    {
      "epoch": 0.11048002931476732,
      "grad_norm": 0.45535979366815893,
      "learning_rate": 9.99679334184001e-06,
      "loss": 0.5116,
      "step": 603
    },
    {
      "epoch": 0.11066324661048003,
      "grad_norm": 0.5208092576973751,
      "learning_rate": 9.996677808600202e-06,
      "loss": 0.5079,
      "step": 604
    },
    {
      "epoch": 0.11084646390619274,
      "grad_norm": 0.47721138790862244,
      "learning_rate": 9.996560231438193e-06,
      "loss": 0.5031,
      "step": 605
    },
    {
      "epoch": 0.11102968120190546,
      "grad_norm": 0.4106250945716929,
      "learning_rate": 9.996440610402078e-06,
      "loss": 0.4789,
      "step": 606
    },
    {
      "epoch": 0.11121289849761817,
      "grad_norm": 0.4981443340009068,
      "learning_rate": 9.996318945540792e-06,
      "loss": 0.4749,
      "step": 607
    },
    {
      "epoch": 0.11139611579333089,
      "grad_norm": 0.4365748987869686,
      "learning_rate": 9.996195236904098e-06,
      "loss": 0.503,
      "step": 608
    },
    {
      "epoch": 0.11157933308904361,
      "grad_norm": 0.42769357659836243,
      "learning_rate": 9.996069484542605e-06,
      "loss": 0.4883,
      "step": 609
    },
    {
      "epoch": 0.11176255038475633,
      "grad_norm": 0.4468425715852285,
      "learning_rate": 9.995941688507749e-06,
      "loss": 0.528,
      "step": 610
    },
    {
      "epoch": 0.11194576768046903,
      "grad_norm": 0.44305501127316244,
      "learning_rate": 9.995811848851807e-06,
      "loss": 0.5192,
      "step": 611
    },
    {
      "epoch": 0.11212898497618175,
      "grad_norm": 0.4430509247761743,
      "learning_rate": 9.995679965627891e-06,
      "loss": 0.4879,
      "step": 612
    },
    {
      "epoch": 0.11231220227189447,
      "grad_norm": 0.4330736232909904,
      "learning_rate": 9.995546038889948e-06,
      "loss": 0.5315,
      "step": 613
    },
    {
      "epoch": 0.11249541956760718,
      "grad_norm": 0.38536887827486094,
      "learning_rate": 9.995410068692763e-06,
      "loss": 0.4789,
      "step": 614
    },
    {
      "epoch": 0.1126786368633199,
      "grad_norm": 0.5420414741691238,
      "learning_rate": 9.995272055091954e-06,
      "loss": 0.4863,
      "step": 615
    },
    {
      "epoch": 0.11286185415903262,
      "grad_norm": 0.4166977498062619,
      "learning_rate": 9.995131998143976e-06,
      "loss": 0.4863,
      "step": 616
    },
    {
      "epoch": 0.11304507145474532,
      "grad_norm": 0.40681536551456327,
      "learning_rate": 9.99498989790612e-06,
      "loss": 0.5158,
      "step": 617
    },
    {
      "epoch": 0.11322828875045804,
      "grad_norm": 0.4164498167643854,
      "learning_rate": 9.994845754436516e-06,
      "loss": 0.5287,
      "step": 618
    },
    {
      "epoch": 0.11341150604617076,
      "grad_norm": 0.430875770675555,
      "learning_rate": 9.994699567794123e-06,
      "loss": 0.4949,
      "step": 619
    },
    {
      "epoch": 0.11359472334188347,
      "grad_norm": 0.47209089186963143,
      "learning_rate": 9.994551338038742e-06,
      "loss": 0.5212,
      "step": 620
    },
    {
      "epoch": 0.11377794063759619,
      "grad_norm": 0.4693460118731517,
      "learning_rate": 9.994401065231008e-06,
      "loss": 0.5218,
      "step": 621
    },
    {
      "epoch": 0.11396115793330891,
      "grad_norm": 0.43645941737097255,
      "learning_rate": 9.994248749432388e-06,
      "loss": 0.5355,
      "step": 622
    },
    {
      "epoch": 0.11414437522902161,
      "grad_norm": 0.45174829183811943,
      "learning_rate": 9.994094390705189e-06,
      "loss": 0.4941,
      "step": 623
    },
    {
      "epoch": 0.11432759252473433,
      "grad_norm": 0.4609885421062077,
      "learning_rate": 9.993937989112554e-06,
      "loss": 0.5424,
      "step": 624
    },
    {
      "epoch": 0.11451080982044705,
      "grad_norm": 0.4711205796072519,
      "learning_rate": 9.993779544718459e-06,
      "loss": 0.5054,
      "step": 625
    },
    {
      "epoch": 0.11469402711615977,
      "grad_norm": 0.4692458224302816,
      "learning_rate": 9.993619057587714e-06,
      "loss": 0.5013,
      "step": 626
    },
    {
      "epoch": 0.11487724441187248,
      "grad_norm": 0.44543459536907126,
      "learning_rate": 9.99345652778597e-06,
      "loss": 0.4882,
      "step": 627
    },
    {
      "epoch": 0.1150604617075852,
      "grad_norm": 0.43461884283611496,
      "learning_rate": 9.993291955379713e-06,
      "loss": 0.5352,
      "step": 628
    },
    {
      "epoch": 0.1152436790032979,
      "grad_norm": 0.41500076390959145,
      "learning_rate": 9.993125340436258e-06,
      "loss": 0.4895,
      "step": 629
    },
    {
      "epoch": 0.11542689629901062,
      "grad_norm": 0.4307377636082482,
      "learning_rate": 9.992956683023762e-06,
      "loss": 0.5149,
      "step": 630
    },
    {
      "epoch": 0.11561011359472334,
      "grad_norm": 0.45085961488644083,
      "learning_rate": 9.992785983211214e-06,
      "loss": 0.5061,
      "step": 631
    },
    {
      "epoch": 0.11579333089043606,
      "grad_norm": 0.45702457348234676,
      "learning_rate": 9.992613241068444e-06,
      "loss": 0.4947,
      "step": 632
    },
    {
      "epoch": 0.11597654818614878,
      "grad_norm": 0.46046219980195896,
      "learning_rate": 9.992438456666108e-06,
      "loss": 0.4727,
      "step": 633
    },
    {
      "epoch": 0.1161597654818615,
      "grad_norm": 0.5109084834664002,
      "learning_rate": 9.992261630075704e-06,
      "loss": 0.5015,
      "step": 634
    },
    {
      "epoch": 0.1163429827775742,
      "grad_norm": 0.4605028589793491,
      "learning_rate": 9.992082761369567e-06,
      "loss": 0.518,
      "step": 635
    },
    {
      "epoch": 0.11652620007328691,
      "grad_norm": 0.4135254020313275,
      "learning_rate": 9.991901850620861e-06,
      "loss": 0.4833,
      "step": 636
    },
    {
      "epoch": 0.11670941736899963,
      "grad_norm": 0.5226822189268372,
      "learning_rate": 9.99171889790359e-06,
      "loss": 0.5118,
      "step": 637
    },
    {
      "epoch": 0.11689263466471235,
      "grad_norm": 0.4061164579366039,
      "learning_rate": 9.991533903292592e-06,
      "loss": 0.4953,
      "step": 638
    },
    {
      "epoch": 0.11707585196042507,
      "grad_norm": 2.3765281557608486,
      "learning_rate": 9.99134686686354e-06,
      "loss": 0.4708,
      "step": 639
    },
    {
      "epoch": 0.11725906925613779,
      "grad_norm": 0.4675538148921928,
      "learning_rate": 9.991157788692942e-06,
      "loss": 0.4994,
      "step": 640
    },
    {
      "epoch": 0.11744228655185049,
      "grad_norm": 0.4467597960418954,
      "learning_rate": 9.990966668858144e-06,
      "loss": 0.4982,
      "step": 641
    },
    {
      "epoch": 0.1176255038475632,
      "grad_norm": 0.46222076234046117,
      "learning_rate": 9.99077350743732e-06,
      "loss": 0.5208,
      "step": 642
    },
    {
      "epoch": 0.11780872114327592,
      "grad_norm": 0.45856899421216063,
      "learning_rate": 9.990578304509488e-06,
      "loss": 0.5109,
      "step": 643
    },
    {
      "epoch": 0.11799193843898864,
      "grad_norm": 0.5006843127431095,
      "learning_rate": 9.990381060154496e-06,
      "loss": 0.4941,
      "step": 644
    },
    {
      "epoch": 0.11817515573470136,
      "grad_norm": 0.4638055815431298,
      "learning_rate": 9.990181774453028e-06,
      "loss": 0.509,
      "step": 645
    },
    {
      "epoch": 0.11835837303041408,
      "grad_norm": 0.4429725714841303,
      "learning_rate": 9.989980447486601e-06,
      "loss": 0.5196,
      "step": 646
    },
    {
      "epoch": 0.11854159032612678,
      "grad_norm": 0.44051330732830757,
      "learning_rate": 9.989777079337572e-06,
      "loss": 0.5316,
      "step": 647
    },
    {
      "epoch": 0.1187248076218395,
      "grad_norm": 0.41555588890083767,
      "learning_rate": 9.989571670089129e-06,
      "loss": 0.5126,
      "step": 648
    },
    {
      "epoch": 0.11890802491755222,
      "grad_norm": 0.46239837428682623,
      "learning_rate": 9.989364219825295e-06,
      "loss": 0.506,
      "step": 649
    },
    {
      "epoch": 0.11909124221326493,
      "grad_norm": 0.416093035128269,
      "learning_rate": 9.98915472863093e-06,
      "loss": 0.5194,
      "step": 650
    },
    {
      "epoch": 0.11927445950897765,
      "grad_norm": 0.43847114193541153,
      "learning_rate": 9.988943196591727e-06,
      "loss": 0.4994,
      "step": 651
    },
    {
      "epoch": 0.11945767680469037,
      "grad_norm": 0.44451525143619486,
      "learning_rate": 9.988729623794215e-06,
      "loss": 0.5283,
      "step": 652
    },
    {
      "epoch": 0.11964089410040307,
      "grad_norm": 0.4041297930697904,
      "learning_rate": 9.988514010325758e-06,
      "loss": 0.5049,
      "step": 653
    },
    {
      "epoch": 0.11982411139611579,
      "grad_norm": 0.4075716530480146,
      "learning_rate": 9.988296356274551e-06,
      "loss": 0.4921,
      "step": 654
    },
    {
      "epoch": 0.12000732869182851,
      "grad_norm": 0.40736195443308915,
      "learning_rate": 9.988076661729631e-06,
      "loss": 0.4805,
      "step": 655
    },
    {
      "epoch": 0.12019054598754123,
      "grad_norm": 0.4545077167148167,
      "learning_rate": 9.987854926780863e-06,
      "loss": 0.4789,
      "step": 656
    },
    {
      "epoch": 0.12037376328325394,
      "grad_norm": 0.5314832542562551,
      "learning_rate": 9.987631151518948e-06,
      "loss": 0.5263,
      "step": 657
    },
    {
      "epoch": 0.12055698057896666,
      "grad_norm": 0.4967524623940482,
      "learning_rate": 9.987405336035425e-06,
      "loss": 0.5106,
      "step": 658
    },
    {
      "epoch": 0.12074019787467936,
      "grad_norm": 0.4306664623952609,
      "learning_rate": 9.987177480422663e-06,
      "loss": 0.5376,
      "step": 659
    },
    {
      "epoch": 0.12092341517039208,
      "grad_norm": 0.4158801663482348,
      "learning_rate": 9.98694758477387e-06,
      "loss": 0.4838,
      "step": 660
    },
    {
      "epoch": 0.1211066324661048,
      "grad_norm": 0.40605820771792417,
      "learning_rate": 9.986715649183084e-06,
      "loss": 0.505,
      "step": 661
    },
    {
      "epoch": 0.12128984976181752,
      "grad_norm": 0.4085721106839356,
      "learning_rate": 9.986481673745183e-06,
      "loss": 0.4993,
      "step": 662
    },
    {
      "epoch": 0.12147306705753023,
      "grad_norm": 0.4785768918832099,
      "learning_rate": 9.986245658555873e-06,
      "loss": 0.5387,
      "step": 663
    },
    {
      "epoch": 0.12165628435324295,
      "grad_norm": 0.4175528791505364,
      "learning_rate": 9.986007603711698e-06,
      "loss": 0.4976,
      "step": 664
    },
    {
      "epoch": 0.12183950164895566,
      "grad_norm": 0.4323331836919794,
      "learning_rate": 9.985767509310035e-06,
      "loss": 0.5346,
      "step": 665
    },
    {
      "epoch": 0.12202271894466837,
      "grad_norm": 0.4868011056584651,
      "learning_rate": 9.9855253754491e-06,
      "loss": 0.5309,
      "step": 666
    },
    {
      "epoch": 0.12220593624038109,
      "grad_norm": 0.46372133085485157,
      "learning_rate": 9.985281202227936e-06,
      "loss": 0.5053,
      "step": 667
    },
    {
      "epoch": 0.12238915353609381,
      "grad_norm": 0.48326269099644514,
      "learning_rate": 9.985034989746423e-06,
      "loss": 0.4941,
      "step": 668
    },
    {
      "epoch": 0.12257237083180653,
      "grad_norm": 0.4557552947362274,
      "learning_rate": 9.984786738105279e-06,
      "loss": 0.5121,
      "step": 669
    },
    {
      "epoch": 0.12275558812751924,
      "grad_norm": 0.39704238565295197,
      "learning_rate": 9.98453644740605e-06,
      "loss": 0.4962,
      "step": 670
    },
    {
      "epoch": 0.12293880542323195,
      "grad_norm": 0.40645243334504044,
      "learning_rate": 9.98428411775112e-06,
      "loss": 0.5046,
      "step": 671
    },
    {
      "epoch": 0.12312202271894467,
      "grad_norm": 0.42629021258457467,
      "learning_rate": 9.984029749243707e-06,
      "loss": 0.5084,
      "step": 672
    },
    {
      "epoch": 0.12330524001465738,
      "grad_norm": 0.45965819318406503,
      "learning_rate": 9.98377334198786e-06,
      "loss": 0.4759,
      "step": 673
    },
    {
      "epoch": 0.1234884573103701,
      "grad_norm": 0.41967629717267835,
      "learning_rate": 9.983514896088466e-06,
      "loss": 0.498,
      "step": 674
    },
    {
      "epoch": 0.12367167460608282,
      "grad_norm": 0.4295939596150219,
      "learning_rate": 9.983254411651242e-06,
      "loss": 0.5243,
      "step": 675
    },
    {
      "epoch": 0.12385489190179554,
      "grad_norm": 0.4808089074794197,
      "learning_rate": 9.982991888782742e-06,
      "loss": 0.5,
      "step": 676
    },
    {
      "epoch": 0.12403810919750824,
      "grad_norm": 0.48728725704379316,
      "learning_rate": 9.982727327590352e-06,
      "loss": 0.5459,
      "step": 677
    },
    {
      "epoch": 0.12422132649322096,
      "grad_norm": 0.42799437772375803,
      "learning_rate": 9.982460728182292e-06,
      "loss": 0.5488,
      "step": 678
    },
    {
      "epoch": 0.12440454378893367,
      "grad_norm": 0.4334349975707624,
      "learning_rate": 9.982192090667618e-06,
      "loss": 0.501,
      "step": 679
    },
    {
      "epoch": 0.12458776108464639,
      "grad_norm": 0.45671378695636383,
      "learning_rate": 9.981921415156217e-06,
      "loss": 0.5236,
      "step": 680
    },
    {
      "epoch": 0.12477097838035911,
      "grad_norm": 0.49531406109254905,
      "learning_rate": 9.98164870175881e-06,
      "loss": 0.5224,
      "step": 681
    },
    {
      "epoch": 0.12495419567607183,
      "grad_norm": 0.3902067301024602,
      "learning_rate": 9.981373950586952e-06,
      "loss": 0.4835,
      "step": 682
    },
    {
      "epoch": 0.12513741297178455,
      "grad_norm": 0.46995820446689096,
      "learning_rate": 9.981097161753032e-06,
      "loss": 0.527,
      "step": 683
    },
    {
      "epoch": 0.12532063026749726,
      "grad_norm": 0.43642001976560174,
      "learning_rate": 9.980818335370273e-06,
      "loss": 0.514,
      "step": 684
    },
    {
      "epoch": 0.12550384756320998,
      "grad_norm": 0.5074521053811396,
      "learning_rate": 9.980537471552728e-06,
      "loss": 0.4897,
      "step": 685
    },
    {
      "epoch": 0.12568706485892267,
      "grad_norm": 0.43625566798430576,
      "learning_rate": 9.98025457041529e-06,
      "loss": 0.5174,
      "step": 686
    },
    {
      "epoch": 0.1258702821546354,
      "grad_norm": 0.4129578447344159,
      "learning_rate": 9.979969632073678e-06,
      "loss": 0.5135,
      "step": 687
    },
    {
      "epoch": 0.1260534994503481,
      "grad_norm": 0.4553566655837448,
      "learning_rate": 9.97968265664445e-06,
      "loss": 0.5228,
      "step": 688
    },
    {
      "epoch": 0.12623671674606082,
      "grad_norm": 0.4682536997157449,
      "learning_rate": 9.979393644244992e-06,
      "loss": 0.5039,
      "step": 689
    },
    {
      "epoch": 0.12641993404177354,
      "grad_norm": 0.42839595687214543,
      "learning_rate": 9.979102594993533e-06,
      "loss": 0.5128,
      "step": 690
    },
    {
      "epoch": 0.12660315133748626,
      "grad_norm": 0.46092632925079513,
      "learning_rate": 9.978809509009121e-06,
      "loss": 0.5011,
      "step": 691
    },
    {
      "epoch": 0.12678636863319898,
      "grad_norm": 0.4486176186332823,
      "learning_rate": 9.97851438641165e-06,
      "loss": 0.5169,
      "step": 692
    },
    {
      "epoch": 0.1269695859289117,
      "grad_norm": 0.47419826995019465,
      "learning_rate": 9.978217227321837e-06,
      "loss": 0.5131,
      "step": 693
    },
    {
      "epoch": 0.1271528032246244,
      "grad_norm": 0.4004091408746546,
      "learning_rate": 9.97791803186124e-06,
      "loss": 0.4743,
      "step": 694
    },
    {
      "epoch": 0.12733602052033713,
      "grad_norm": 0.38849935615066844,
      "learning_rate": 9.977616800152248e-06,
      "loss": 0.4944,
      "step": 695
    },
    {
      "epoch": 0.12751923781604985,
      "grad_norm": 0.44216099670793213,
      "learning_rate": 9.977313532318078e-06,
      "loss": 0.5072,
      "step": 696
    },
    {
      "epoch": 0.12770245511176256,
      "grad_norm": 0.4462646919434802,
      "learning_rate": 9.977008228482785e-06,
      "loss": 0.4901,
      "step": 697
    },
    {
      "epoch": 0.12788567240747525,
      "grad_norm": 0.44769942596462214,
      "learning_rate": 9.976700888771259e-06,
      "loss": 0.5263,
      "step": 698
    },
    {
      "epoch": 0.12806888970318797,
      "grad_norm": 0.5113829977752261,
      "learning_rate": 9.976391513309212e-06,
      "loss": 0.5202,
      "step": 699
    },
    {
      "epoch": 0.1282521069989007,
      "grad_norm": 0.4245344457429005,
      "learning_rate": 9.976080102223202e-06,
      "loss": 0.5182,
      "step": 700
    },
    {
      "epoch": 0.1284353242946134,
      "grad_norm": 0.4305381223047699,
      "learning_rate": 9.97576665564061e-06,
      "loss": 0.507,
      "step": 701
    },
    {
      "epoch": 0.12861854159032612,
      "grad_norm": 0.47072785404283757,
      "learning_rate": 9.975451173689658e-06,
      "loss": 0.5166,
      "step": 702
    },
    {
      "epoch": 0.12880175888603884,
      "grad_norm": 0.6675845713373886,
      "learning_rate": 9.975133656499392e-06,
      "loss": 0.47,
      "step": 703
    },
    {
      "epoch": 0.12898497618175156,
      "grad_norm": 0.4533261672443661,
      "learning_rate": 9.974814104199694e-06,
      "loss": 0.5362,
      "step": 704
    },
    {
      "epoch": 0.12916819347746428,
      "grad_norm": 0.4503943455070683,
      "learning_rate": 9.97449251692128e-06,
      "loss": 0.5134,
      "step": 705
    },
    {
      "epoch": 0.129351410773177,
      "grad_norm": 0.44545859946020006,
      "learning_rate": 9.974168894795698e-06,
      "loss": 0.4981,
      "step": 706
    },
    {
      "epoch": 0.1295346280688897,
      "grad_norm": 0.430864283858029,
      "learning_rate": 9.973843237955328e-06,
      "loss": 0.5329,
      "step": 707
    },
    {
      "epoch": 0.12971784536460243,
      "grad_norm": 0.7865986917626206,
      "learning_rate": 9.973515546533379e-06,
      "loss": 0.504,
      "step": 708
    },
    {
      "epoch": 0.12990106266031515,
      "grad_norm": 0.44976211133712124,
      "learning_rate": 9.973185820663897e-06,
      "loss": 0.4737,
      "step": 709
    },
    {
      "epoch": 0.13008427995602784,
      "grad_norm": 0.4549304394815992,
      "learning_rate": 9.97285406048176e-06,
      "loss": 0.5425,
      "step": 710
    },
    {
      "epoch": 0.13026749725174055,
      "grad_norm": 0.5049990473394447,
      "learning_rate": 9.972520266122676e-06,
      "loss": 0.5117,
      "step": 711
    },
    {
      "epoch": 0.13045071454745327,
      "grad_norm": 0.42657669286902156,
      "learning_rate": 9.972184437723182e-06,
      "loss": 0.5057,
      "step": 712
    },
    {
      "epoch": 0.130633931843166,
      "grad_norm": 0.5299863700854671,
      "learning_rate": 9.971846575420656e-06,
      "loss": 0.5262,
      "step": 713
    },
    {
      "epoch": 0.1308171491388787,
      "grad_norm": 0.45345740582998295,
      "learning_rate": 9.9715066793533e-06,
      "loss": 0.4997,
      "step": 714
    },
    {
      "epoch": 0.13100036643459143,
      "grad_norm": 0.39393086616308026,
      "learning_rate": 9.971164749660149e-06,
      "loss": 0.4996,
      "step": 715
    },
    {
      "epoch": 0.13118358373030414,
      "grad_norm": 0.4577611248522203,
      "learning_rate": 9.970820786481075e-06,
      "loss": 0.5018,
      "step": 716
    },
    {
      "epoch": 0.13136680102601686,
      "grad_norm": 0.43279352533859367,
      "learning_rate": 9.970474789956775e-06,
      "loss": 0.4914,
      "step": 717
    },
    {
      "epoch": 0.13155001832172958,
      "grad_norm": 0.4366448132250816,
      "learning_rate": 9.970126760228785e-06,
      "loss": 0.4864,
      "step": 718
    },
    {
      "epoch": 0.1317332356174423,
      "grad_norm": 0.43998158412052296,
      "learning_rate": 9.969776697439463e-06,
      "loss": 0.4834,
      "step": 719
    },
    {
      "epoch": 0.131916452913155,
      "grad_norm": 0.46940772392002544,
      "learning_rate": 9.969424601732011e-06,
      "loss": 0.501,
      "step": 720
    },
    {
      "epoch": 0.1320996702088677,
      "grad_norm": 0.43117110167391715,
      "learning_rate": 9.96907047325045e-06,
      "loss": 0.5562,
      "step": 721
    },
    {
      "epoch": 0.13228288750458042,
      "grad_norm": 0.4119352967972661,
      "learning_rate": 9.968714312139642e-06,
      "loss": 0.4748,
      "step": 722
    },
    {
      "epoch": 0.13246610480029314,
      "grad_norm": 0.5028443530663289,
      "learning_rate": 9.968356118545277e-06,
      "loss": 0.5344,
      "step": 723
    },
    {
      "epoch": 0.13264932209600586,
      "grad_norm": 0.43089419048593447,
      "learning_rate": 9.967995892613875e-06,
      "loss": 0.4704,
      "step": 724
    },
    {
      "epoch": 0.13283253939171857,
      "grad_norm": 0.5044933769783081,
      "learning_rate": 9.967633634492788e-06,
      "loss": 0.4996,
      "step": 725
    },
    {
      "epoch": 0.1330157566874313,
      "grad_norm": 0.442554097189932,
      "learning_rate": 9.967269344330201e-06,
      "loss": 0.5278,
      "step": 726
    },
    {
      "epoch": 0.133198973983144,
      "grad_norm": 0.431094495901828,
      "learning_rate": 9.966903022275131e-06,
      "loss": 0.4943,
      "step": 727
    },
    {
      "epoch": 0.13338219127885673,
      "grad_norm": 0.6222105275898999,
      "learning_rate": 9.966534668477421e-06,
      "loss": 0.5215,
      "step": 728
    },
    {
      "epoch": 0.13356540857456944,
      "grad_norm": 0.4562449049230116,
      "learning_rate": 9.96616428308775e-06,
      "loss": 0.5112,
      "step": 729
    },
    {
      "epoch": 0.13374862587028216,
      "grad_norm": 0.4249667668518143,
      "learning_rate": 9.965791866257626e-06,
      "loss": 0.5083,
      "step": 730
    },
    {
      "epoch": 0.13393184316599488,
      "grad_norm": 0.42736565388331654,
      "learning_rate": 9.96541741813939e-06,
      "loss": 0.5078,
      "step": 731
    },
    {
      "epoch": 0.1341150604617076,
      "grad_norm": 0.42789645341508464,
      "learning_rate": 9.96504093888621e-06,
      "loss": 0.5148,
      "step": 732
    },
    {
      "epoch": 0.1342982777574203,
      "grad_norm": 0.5018533252539279,
      "learning_rate": 9.964662428652088e-06,
      "loss": 0.4994,
      "step": 733
    },
    {
      "epoch": 0.134481495053133,
      "grad_norm": 0.4561559749816332,
      "learning_rate": 9.964281887591856e-06,
      "loss": 0.5036,
      "step": 734
    },
    {
      "epoch": 0.13466471234884572,
      "grad_norm": 0.43874828747054045,
      "learning_rate": 9.963899315861176e-06,
      "loss": 0.4976,
      "step": 735
    },
    {
      "epoch": 0.13484792964455844,
      "grad_norm": 0.44683273370213666,
      "learning_rate": 9.963514713616544e-06,
      "loss": 0.4837,
      "step": 736
    },
    {
      "epoch": 0.13503114694027116,
      "grad_norm": 0.4575940082218496,
      "learning_rate": 9.963128081015282e-06,
      "loss": 0.531,
      "step": 737
    },
    {
      "epoch": 0.13521436423598387,
      "grad_norm": 0.4190671986873532,
      "learning_rate": 9.962739418215545e-06,
      "loss": 0.525,
      "step": 738
    },
    {
      "epoch": 0.1353975815316966,
      "grad_norm": 0.41284775197321,
      "learning_rate": 9.962348725376318e-06,
      "loss": 0.519,
      "step": 739
    },
    {
      "epoch": 0.1355807988274093,
      "grad_norm": 0.41220938724190687,
      "learning_rate": 9.961956002657414e-06,
      "loss": 0.535,
      "step": 740
    },
    {
      "epoch": 0.13576401612312203,
      "grad_norm": 0.4428272636340964,
      "learning_rate": 9.961561250219482e-06,
      "loss": 0.5287,
      "step": 741
    },
    {
      "epoch": 0.13594723341883475,
      "grad_norm": 0.42565025293852765,
      "learning_rate": 9.961164468223996e-06,
      "loss": 0.5329,
      "step": 742
    },
    {
      "epoch": 0.13613045071454746,
      "grad_norm": 0.5963216630193675,
      "learning_rate": 9.960765656833263e-06,
      "loss": 0.5036,
      "step": 743
    },
    {
      "epoch": 0.13631366801026018,
      "grad_norm": 0.42494573499796934,
      "learning_rate": 9.96036481621042e-06,
      "loss": 0.4936,
      "step": 744
    },
    {
      "epoch": 0.13649688530597287,
      "grad_norm": 0.39224716002216886,
      "learning_rate": 9.959961946519431e-06,
      "loss": 0.4704,
      "step": 745
    },
    {
      "epoch": 0.1366801026016856,
      "grad_norm": 0.42574265198042266,
      "learning_rate": 9.959557047925095e-06,
      "loss": 0.4964,
      "step": 746
    },
    {
      "epoch": 0.1368633198973983,
      "grad_norm": 0.457270415221301,
      "learning_rate": 9.959150120593035e-06,
      "loss": 0.5322,
      "step": 747
    },
    {
      "epoch": 0.13704653719311102,
      "grad_norm": 0.4245718410150897,
      "learning_rate": 9.95874116468971e-06,
      "loss": 0.4952,
      "step": 748
    },
    {
      "epoch": 0.13722975448882374,
      "grad_norm": 0.47589556246220993,
      "learning_rate": 9.958330180382405e-06,
      "loss": 0.5124,
      "step": 749
    },
    {
      "epoch": 0.13741297178453646,
      "grad_norm": 0.45440270321018955,
      "learning_rate": 9.957917167839238e-06,
      "loss": 0.5109,
      "step": 750
    },
    {
      "epoch": 0.13759618908024918,
      "grad_norm": 0.5232685566183007,
      "learning_rate": 9.95750212722915e-06,
      "loss": 0.5174,
      "step": 751
    },
    {
      "epoch": 0.1377794063759619,
      "grad_norm": 0.538012343254673,
      "learning_rate": 9.957085058721918e-06,
      "loss": 0.5105,
      "step": 752
    },
    {
      "epoch": 0.1379626236716746,
      "grad_norm": 0.38959097348139504,
      "learning_rate": 9.956665962488148e-06,
      "loss": 0.4533,
      "step": 753
    },
    {
      "epoch": 0.13814584096738733,
      "grad_norm": 0.4404610898092818,
      "learning_rate": 9.956244838699271e-06,
      "loss": 0.4978,
      "step": 754
    },
    {
      "epoch": 0.13832905826310005,
      "grad_norm": 0.47384922621236214,
      "learning_rate": 9.955821687527554e-06,
      "loss": 0.5309,
      "step": 755
    },
    {
      "epoch": 0.13851227555881276,
      "grad_norm": 0.45266354792605684,
      "learning_rate": 9.955396509146084e-06,
      "loss": 0.5151,
      "step": 756
    },
    {
      "epoch": 0.13869549285452545,
      "grad_norm": 0.4098036061670406,
      "learning_rate": 9.95496930372879e-06,
      "loss": 0.5,
      "step": 757
    },
    {
      "epoch": 0.13887871015023817,
      "grad_norm": 0.4520393902857722,
      "learning_rate": 9.954540071450418e-06,
      "loss": 0.4766,
      "step": 758
    },
    {
      "epoch": 0.1390619274459509,
      "grad_norm": 0.41592431105518124,
      "learning_rate": 9.95410881248655e-06,
      "loss": 0.5279,
      "step": 759
    },
    {
      "epoch": 0.1392451447416636,
      "grad_norm": 0.9055926932881602,
      "learning_rate": 9.953675527013594e-06,
      "loss": 0.4544,
      "step": 760
    },
    {
      "epoch": 0.13942836203737632,
      "grad_norm": 0.4230083797544526,
      "learning_rate": 9.953240215208787e-06,
      "loss": 0.5174,
      "step": 761
    },
    {
      "epoch": 0.13961157933308904,
      "grad_norm": 0.46167105853029766,
      "learning_rate": 9.9528028772502e-06,
      "loss": 0.5087,
      "step": 762
    },
    {
      "epoch": 0.13979479662880176,
      "grad_norm": 0.4614988799168283,
      "learning_rate": 9.952363513316727e-06,
      "loss": 0.5035,
      "step": 763
    },
    {
      "epoch": 0.13997801392451448,
      "grad_norm": 0.4599378559524754,
      "learning_rate": 9.951922123588091e-06,
      "loss": 0.4982,
      "step": 764
    },
    {
      "epoch": 0.1401612312202272,
      "grad_norm": 0.4169187563233499,
      "learning_rate": 9.951478708244847e-06,
      "loss": 0.5205,
      "step": 765
    },
    {
      "epoch": 0.1403444485159399,
      "grad_norm": 1.4346541301623688,
      "learning_rate": 9.951033267468375e-06,
      "loss": 0.4704,
      "step": 766
    },
    {
      "epoch": 0.14052766581165263,
      "grad_norm": 0.4477016515011828,
      "learning_rate": 9.950585801440889e-06,
      "loss": 0.489,
      "step": 767
    },
    {
      "epoch": 0.14071088310736535,
      "grad_norm": 0.4178701034776549,
      "learning_rate": 9.950136310345425e-06,
      "loss": 0.5147,
      "step": 768
    },
    {
      "epoch": 0.14089410040307804,
      "grad_norm": 0.4893104037393506,
      "learning_rate": 9.949684794365848e-06,
      "loss": 0.4913,
      "step": 769
    },
    {
      "epoch": 0.14107731769879076,
      "grad_norm": 0.4018838415277579,
      "learning_rate": 9.949231253686857e-06,
      "loss": 0.464,
      "step": 770
    },
    {
      "epoch": 0.14126053499450347,
      "grad_norm": 0.38608379924790914,
      "learning_rate": 9.948775688493974e-06,
      "loss": 0.5173,
      "step": 771
    },
    {
      "epoch": 0.1414437522902162,
      "grad_norm": 0.4162388206396457,
      "learning_rate": 9.948318098973552e-06,
      "loss": 0.4917,
      "step": 772
    },
    {
      "epoch": 0.1416269695859289,
      "grad_norm": 0.3781950328417496,
      "learning_rate": 9.947858485312772e-06,
      "loss": 0.4446,
      "step": 773
    },
    {
      "epoch": 0.14181018688164163,
      "grad_norm": 0.45959051057998834,
      "learning_rate": 9.947396847699638e-06,
      "loss": 0.4918,
      "step": 774
    },
    {
      "epoch": 0.14199340417735434,
      "grad_norm": 0.480317102541819,
      "learning_rate": 9.946933186322988e-06,
      "loss": 0.5369,
      "step": 775
    },
    {
      "epoch": 0.14217662147306706,
      "grad_norm": 0.5008985493720574,
      "learning_rate": 9.946467501372485e-06,
      "loss": 0.4992,
      "step": 776
    },
    {
      "epoch": 0.14235983876877978,
      "grad_norm": 0.4117591563183844,
      "learning_rate": 9.94599979303862e-06,
      "loss": 0.4818,
      "step": 777
    },
    {
      "epoch": 0.1425430560644925,
      "grad_norm": 0.531108956987235,
      "learning_rate": 9.945530061512714e-06,
      "loss": 0.517,
      "step": 778
    },
    {
      "epoch": 0.1427262733602052,
      "grad_norm": 0.43817523582155865,
      "learning_rate": 9.945058306986911e-06,
      "loss": 0.5473,
      "step": 779
    },
    {
      "epoch": 0.14290949065591793,
      "grad_norm": 0.4747080227792923,
      "learning_rate": 9.944584529654187e-06,
      "loss": 0.4819,
      "step": 780
    },
    {
      "epoch": 0.14309270795163062,
      "grad_norm": 0.4249215515551791,
      "learning_rate": 9.944108729708342e-06,
      "loss": 0.4979,
      "step": 781
    },
    {
      "epoch": 0.14327592524734334,
      "grad_norm": 0.43558257752578505,
      "learning_rate": 9.943630907344008e-06,
      "loss": 0.484,
      "step": 782
    },
    {
      "epoch": 0.14345914254305606,
      "grad_norm": 0.4750219883418816,
      "learning_rate": 9.943151062756638e-06,
      "loss": 0.4825,
      "step": 783
    },
    {
      "epoch": 0.14364235983876877,
      "grad_norm": 0.4332476417990451,
      "learning_rate": 9.942669196142516e-06,
      "loss": 0.4929,
      "step": 784
    },
    {
      "epoch": 0.1438255771344815,
      "grad_norm": 0.48307720479460303,
      "learning_rate": 9.942185307698754e-06,
      "loss": 0.5099,
      "step": 785
    },
    {
      "epoch": 0.1440087944301942,
      "grad_norm": 0.43004630110605213,
      "learning_rate": 9.941699397623289e-06,
      "loss": 0.4435,
      "step": 786
    },
    {
      "epoch": 0.14419201172590693,
      "grad_norm": 0.5526090978976514,
      "learning_rate": 9.941211466114883e-06,
      "loss": 0.4831,
      "step": 787
    },
    {
      "epoch": 0.14437522902161964,
      "grad_norm": 0.4299548315161719,
      "learning_rate": 9.94072151337313e-06,
      "loss": 0.5055,
      "step": 788
    },
    {
      "epoch": 0.14455844631733236,
      "grad_norm": 0.46488795405005734,
      "learning_rate": 9.940229539598449e-06,
      "loss": 0.5247,
      "step": 789
    },
    {
      "epoch": 0.14474166361304508,
      "grad_norm": 0.45711196453956754,
      "learning_rate": 9.939735544992084e-06,
      "loss": 0.5155,
      "step": 790
    },
    {
      "epoch": 0.1449248809087578,
      "grad_norm": 0.39377908902960373,
      "learning_rate": 9.939239529756106e-06,
      "loss": 0.4861,
      "step": 791
    },
    {
      "epoch": 0.14510809820447051,
      "grad_norm": 0.3780973854580167,
      "learning_rate": 9.938741494093413e-06,
      "loss": 0.4743,
      "step": 792
    },
    {
      "epoch": 0.1452913155001832,
      "grad_norm": 0.4442667746426281,
      "learning_rate": 9.93824143820773e-06,
      "loss": 0.5124,
      "step": 793
    },
    {
      "epoch": 0.14547453279589592,
      "grad_norm": 0.47139079982459514,
      "learning_rate": 9.93773936230361e-06,
      "loss": 0.5095,
      "step": 794
    },
    {
      "epoch": 0.14565775009160864,
      "grad_norm": 0.47478113201389704,
      "learning_rate": 9.937235266586425e-06,
      "loss": 0.5242,
      "step": 795
    },
    {
      "epoch": 0.14584096738732136,
      "grad_norm": 0.4253566981856297,
      "learning_rate": 9.936729151262383e-06,
      "loss": 0.4816,
      "step": 796
    },
    {
      "epoch": 0.14602418468303408,
      "grad_norm": 0.3995444614750483,
      "learning_rate": 9.936221016538514e-06,
      "loss": 0.5114,
      "step": 797
    },
    {
      "epoch": 0.1462074019787468,
      "grad_norm": 0.3958644449360531,
      "learning_rate": 9.935710862622671e-06,
      "loss": 0.5044,
      "step": 798
    },
    {
      "epoch": 0.1463906192744595,
      "grad_norm": 0.39672728856696343,
      "learning_rate": 9.935198689723537e-06,
      "loss": 0.5079,
      "step": 799
    },
    {
      "epoch": 0.14657383657017223,
      "grad_norm": 0.4401216989939036,
      "learning_rate": 9.934684498050619e-06,
      "loss": 0.5169,
      "step": 800
    },
    {
      "epoch": 0.14675705386588495,
      "grad_norm": 0.42531985045283716,
      "learning_rate": 9.93416828781425e-06,
      "loss": 0.5039,
      "step": 801
    },
    {
      "epoch": 0.14694027116159766,
      "grad_norm": 0.43641771981673,
      "learning_rate": 9.93365005922559e-06,
      "loss": 0.4705,
      "step": 802
    },
    {
      "epoch": 0.14712348845731038,
      "grad_norm": 0.4154214727248315,
      "learning_rate": 9.933129812496623e-06,
      "loss": 0.4801,
      "step": 803
    },
    {
      "epoch": 0.1473067057530231,
      "grad_norm": 0.4668765888893056,
      "learning_rate": 9.932607547840161e-06,
      "loss": 0.5262,
      "step": 804
    },
    {
      "epoch": 0.1474899230487358,
      "grad_norm": 0.45163409733644266,
      "learning_rate": 9.932083265469836e-06,
      "loss": 0.5009,
      "step": 805
    },
    {
      "epoch": 0.1476731403444485,
      "grad_norm": 0.44602336917824525,
      "learning_rate": 9.93155696560011e-06,
      "loss": 0.5071,
      "step": 806
    },
    {
      "epoch": 0.14785635764016122,
      "grad_norm": 0.40571563214546336,
      "learning_rate": 9.931028648446273e-06,
      "loss": 0.4867,
      "step": 807
    },
    {
      "epoch": 0.14803957493587394,
      "grad_norm": 0.412866633306789,
      "learning_rate": 9.930498314224433e-06,
      "loss": 0.5204,
      "step": 808
    },
    {
      "epoch": 0.14822279223158666,
      "grad_norm": 0.4491417286505596,
      "learning_rate": 9.929965963151526e-06,
      "loss": 0.4959,
      "step": 809
    },
    {
      "epoch": 0.14840600952729938,
      "grad_norm": 0.483698524685917,
      "learning_rate": 9.929431595445315e-06,
      "loss": 0.4945,
      "step": 810
    },
    {
      "epoch": 0.1485892268230121,
      "grad_norm": 0.45809625895317874,
      "learning_rate": 9.928895211324387e-06,
      "loss": 0.4795,
      "step": 811
    },
    {
      "epoch": 0.1487724441187248,
      "grad_norm": 0.4671855759216309,
      "learning_rate": 9.928356811008153e-06,
      "loss": 0.502,
      "step": 812
    },
    {
      "epoch": 0.14895566141443753,
      "grad_norm": 0.49991252806247843,
      "learning_rate": 9.927816394716847e-06,
      "loss": 0.4989,
      "step": 813
    },
    {
      "epoch": 0.14913887871015025,
      "grad_norm": 0.4930888660574426,
      "learning_rate": 9.92727396267153e-06,
      "loss": 0.5256,
      "step": 814
    },
    {
      "epoch": 0.14932209600586296,
      "grad_norm": 0.4301245288190715,
      "learning_rate": 9.926729515094092e-06,
      "loss": 0.5062,
      "step": 815
    },
    {
      "epoch": 0.14950531330157568,
      "grad_norm": 0.4254345495226345,
      "learning_rate": 9.926183052207235e-06,
      "loss": 0.4885,
      "step": 816
    },
    {
      "epoch": 0.14968853059728837,
      "grad_norm": 0.40854658960332646,
      "learning_rate": 9.925634574234499e-06,
      "loss": 0.4909,
      "step": 817
    },
    {
      "epoch": 0.1498717478930011,
      "grad_norm": 0.4307966660913799,
      "learning_rate": 9.925084081400241e-06,
      "loss": 0.4974,
      "step": 818
    },
    {
      "epoch": 0.1500549651887138,
      "grad_norm": 0.44710983697734263,
      "learning_rate": 9.924531573929641e-06,
      "loss": 0.5107,
      "step": 819
    },
    {
      "epoch": 0.15023818248442652,
      "grad_norm": 0.47247983608563787,
      "learning_rate": 9.923977052048708e-06,
      "loss": 0.5185,
      "step": 820
    },
    {
      "epoch": 0.15042139978013924,
      "grad_norm": 0.43408666873611645,
      "learning_rate": 9.923420515984272e-06,
      "loss": 0.5256,
      "step": 821
    },
    {
      "epoch": 0.15060461707585196,
      "grad_norm": 0.3618361050633503,
      "learning_rate": 9.922861965963987e-06,
      "loss": 0.4835,
      "step": 822
    },
    {
      "epoch": 0.15078783437156468,
      "grad_norm": 0.478572088704784,
      "learning_rate": 9.922301402216334e-06,
      "loss": 0.4971,
      "step": 823
    },
    {
      "epoch": 0.1509710516672774,
      "grad_norm": 0.4956723559981853,
      "learning_rate": 9.921738824970611e-06,
      "loss": 0.4664,
      "step": 824
    },
    {
      "epoch": 0.1511542689629901,
      "grad_norm": 0.43104833444065105,
      "learning_rate": 9.921174234456947e-06,
      "loss": 0.5056,
      "step": 825
    },
    {
      "epoch": 0.15133748625870283,
      "grad_norm": 0.38636058698792763,
      "learning_rate": 9.920607630906289e-06,
      "loss": 0.509,
      "step": 826
    },
    {
      "epoch": 0.15152070355441555,
      "grad_norm": 0.4638108326048624,
      "learning_rate": 9.920039014550413e-06,
      "loss": 0.4928,
      "step": 827
    },
    {
      "epoch": 0.15170392085012827,
      "grad_norm": 0.38962500919500936,
      "learning_rate": 9.919468385621912e-06,
      "loss": 0.5025,
      "step": 828
    },
    {
      "epoch": 0.15188713814584096,
      "grad_norm": 0.423466480397551,
      "learning_rate": 9.918895744354204e-06,
      "loss": 0.4903,
      "step": 829
    },
    {
      "epoch": 0.15207035544155367,
      "grad_norm": 0.5945183810824295,
      "learning_rate": 9.918321090981537e-06,
      "loss": 0.4961,
      "step": 830
    },
    {
      "epoch": 0.1522535727372664,
      "grad_norm": 0.3857346095853992,
      "learning_rate": 9.917744425738971e-06,
      "loss": 0.4903,
      "step": 831
    },
    {
      "epoch": 0.1524367900329791,
      "grad_norm": 0.4368019305898112,
      "learning_rate": 9.917165748862398e-06,
      "loss": 0.4985,
      "step": 832
    },
    {
      "epoch": 0.15262000732869183,
      "grad_norm": 0.4622792781214721,
      "learning_rate": 9.916585060588526e-06,
      "loss": 0.5298,
      "step": 833
    },
    {
      "epoch": 0.15280322462440454,
      "grad_norm": 0.43464073300377387,
      "learning_rate": 9.916002361154894e-06,
      "loss": 0.4956,
      "step": 834
    },
    {
      "epoch": 0.15298644192011726,
      "grad_norm": 0.37914154472744427,
      "learning_rate": 9.915417650799855e-06,
      "loss": 0.4914,
      "step": 835
    },
    {
      "epoch": 0.15316965921582998,
      "grad_norm": 0.41110768627195465,
      "learning_rate": 9.914830929762588e-06,
      "loss": 0.4933,
      "step": 836
    },
    {
      "epoch": 0.1533528765115427,
      "grad_norm": 0.3922989216974703,
      "learning_rate": 9.914242198283099e-06,
      "loss": 0.5,
      "step": 837
    },
    {
      "epoch": 0.1535360938072554,
      "grad_norm": 0.4310098182316404,
      "learning_rate": 9.91365145660221e-06,
      "loss": 0.4817,
      "step": 838
    },
    {
      "epoch": 0.15371931110296813,
      "grad_norm": 0.40690680117071937,
      "learning_rate": 9.913058704961565e-06,
      "loss": 0.4778,
      "step": 839
    },
    {
      "epoch": 0.15390252839868085,
      "grad_norm": 0.4584415585967629,
      "learning_rate": 9.912463943603635e-06,
      "loss": 0.4585,
      "step": 840
    },
    {
      "epoch": 0.15408574569439354,
      "grad_norm": 0.48734349658323495,
      "learning_rate": 9.911867172771711e-06,
      "loss": 0.5345,
      "step": 841
    },
    {
      "epoch": 0.15426896299010626,
      "grad_norm": 0.442686833828428,
      "learning_rate": 9.911268392709908e-06,
      "loss": 0.5134,
      "step": 842
    },
    {
      "epoch": 0.15445218028581897,
      "grad_norm": 0.44506293028939,
      "learning_rate": 9.910667603663156e-06,
      "loss": 0.5158,
      "step": 843
    },
    {
      "epoch": 0.1546353975815317,
      "grad_norm": 0.4311992637968541,
      "learning_rate": 9.910064805877214e-06,
      "loss": 0.4837,
      "step": 844
    },
    {
      "epoch": 0.1548186148772444,
      "grad_norm": 0.4493099076015729,
      "learning_rate": 9.90945999959866e-06,
      "loss": 0.5108,
      "step": 845
    },
    {
      "epoch": 0.15500183217295713,
      "grad_norm": 0.45478365363321205,
      "learning_rate": 9.908853185074896e-06,
      "loss": 0.4998,
      "step": 846
    },
    {
      "epoch": 0.15518504946866984,
      "grad_norm": 0.46301414888311854,
      "learning_rate": 9.90824436255414e-06,
      "loss": 0.5225,
      "step": 847
    },
    {
      "epoch": 0.15536826676438256,
      "grad_norm": 0.40118617155548936,
      "learning_rate": 9.907633532285435e-06,
      "loss": 0.4927,
      "step": 848
    },
    {
      "epoch": 0.15555148406009528,
      "grad_norm": 0.49481295874709574,
      "learning_rate": 9.907020694518646e-06,
      "loss": 0.4933,
      "step": 849
    },
    {
      "epoch": 0.155734701355808,
      "grad_norm": 0.4102626498510104,
      "learning_rate": 9.90640584950446e-06,
      "loss": 0.5111,
      "step": 850
    },
    {
      "epoch": 0.15591791865152071,
      "grad_norm": 0.46514690031713113,
      "learning_rate": 9.905788997494377e-06,
      "loss": 0.4728,
      "step": 851
    },
    {
      "epoch": 0.15610113594723343,
      "grad_norm": 0.4054441718781369,
      "learning_rate": 9.905170138740732e-06,
      "loss": 0.4828,
      "step": 852
    },
    {
      "epoch": 0.15628435324294612,
      "grad_norm": 0.4400252577977126,
      "learning_rate": 9.904549273496666e-06,
      "loss": 0.501,
      "step": 853
    },
    {
      "epoch": 0.15646757053865884,
      "grad_norm": 0.42167618939817325,
      "learning_rate": 9.903926402016153e-06,
      "loss": 0.4922,
      "step": 854
    },
    {
      "epoch": 0.15665078783437156,
      "grad_norm": 0.4093707700968775,
      "learning_rate": 9.90330152455398e-06,
      "loss": 0.5014,
      "step": 855
    },
    {
      "epoch": 0.15683400513008428,
      "grad_norm": 0.4523522809487888,
      "learning_rate": 9.902674641365756e-06,
      "loss": 0.4829,
      "step": 856
    },
    {
      "epoch": 0.157017222425797,
      "grad_norm": 0.49143897815029397,
      "learning_rate": 9.902045752707916e-06,
      "loss": 0.5074,
      "step": 857
    },
    {
      "epoch": 0.1572004397215097,
      "grad_norm": 0.43621740114283203,
      "learning_rate": 9.901414858837706e-06,
      "loss": 0.5182,
      "step": 858
    },
    {
      "epoch": 0.15738365701722243,
      "grad_norm": 0.40732498942379025,
      "learning_rate": 9.9007819600132e-06,
      "loss": 0.4674,
      "step": 859
    },
    {
      "epoch": 0.15756687431293515,
      "grad_norm": 0.43675476681733244,
      "learning_rate": 9.900147056493285e-06,
      "loss": 0.5374,
      "step": 860
    },
    {
      "epoch": 0.15775009160864786,
      "grad_norm": 0.43060074322101793,
      "learning_rate": 9.89951014853768e-06,
      "loss": 0.4635,
      "step": 861
    },
    {
      "epoch": 0.15793330890436058,
      "grad_norm": 0.41179202097945616,
      "learning_rate": 9.898871236406907e-06,
      "loss": 0.4951,
      "step": 862
    },
    {
      "epoch": 0.1581165262000733,
      "grad_norm": 0.371370018002044,
      "learning_rate": 9.898230320362323e-06,
      "loss": 0.479,
      "step": 863
    },
    {
      "epoch": 0.158299743495786,
      "grad_norm": 0.4300134296591519,
      "learning_rate": 9.897587400666097e-06,
      "loss": 0.4868,
      "step": 864
    },
    {
      "epoch": 0.1584829607914987,
      "grad_norm": 0.43589209027697184,
      "learning_rate": 9.896942477581221e-06,
      "loss": 0.5124,
      "step": 865
    },
    {
      "epoch": 0.15866617808721142,
      "grad_norm": 0.575247632832188,
      "learning_rate": 9.896295551371503e-06,
      "loss": 0.5053,
      "step": 866
    },
    {
      "epoch": 0.15884939538292414,
      "grad_norm": 0.4273872012355264,
      "learning_rate": 9.89564662230157e-06,
      "loss": 0.5138,
      "step": 867
    },
    {
      "epoch": 0.15903261267863686,
      "grad_norm": 0.4543163505662005,
      "learning_rate": 9.894995690636874e-06,
      "loss": 0.5229,
      "step": 868
    },
    {
      "epoch": 0.15921582997434958,
      "grad_norm": 0.4491410597219586,
      "learning_rate": 9.894342756643684e-06,
      "loss": 0.5045,
      "step": 869
    },
    {
      "epoch": 0.1593990472700623,
      "grad_norm": 0.47087708750396473,
      "learning_rate": 9.893687820589081e-06,
      "loss": 0.5087,
      "step": 870
    },
    {
      "epoch": 0.159582264565775,
      "grad_norm": 0.4240236693967998,
      "learning_rate": 9.893030882740977e-06,
      "loss": 0.5052,
      "step": 871
    },
    {
      "epoch": 0.15976548186148773,
      "grad_norm": 0.4287525963131784,
      "learning_rate": 9.892371943368092e-06,
      "loss": 0.527,
      "step": 872
    },
    {
      "epoch": 0.15994869915720045,
      "grad_norm": 0.4131563755872441,
      "learning_rate": 9.891711002739971e-06,
      "loss": 0.5074,
      "step": 873
    },
    {
      "epoch": 0.16013191645291316,
      "grad_norm": 0.39231143806918656,
      "learning_rate": 9.891048061126975e-06,
      "loss": 0.5018,
      "step": 874
    },
    {
      "epoch": 0.16031513374862588,
      "grad_norm": 0.4391493046066672,
      "learning_rate": 9.890383118800287e-06,
      "loss": 0.518,
      "step": 875
    },
    {
      "epoch": 0.16049835104433857,
      "grad_norm": 0.468016398258625,
      "learning_rate": 9.889716176031903e-06,
      "loss": 0.5191,
      "step": 876
    },
    {
      "epoch": 0.1606815683400513,
      "grad_norm": 0.4619425954928644,
      "learning_rate": 9.88904723309464e-06,
      "loss": 0.5114,
      "step": 877
    },
    {
      "epoch": 0.160864785635764,
      "grad_norm": 0.4581492770476195,
      "learning_rate": 9.888376290262134e-06,
      "loss": 0.5349,
      "step": 878
    },
    {
      "epoch": 0.16104800293147672,
      "grad_norm": 0.390022001202663,
      "learning_rate": 9.887703347808838e-06,
      "loss": 0.507,
      "step": 879
    },
    {
      "epoch": 0.16123122022718944,
      "grad_norm": 0.38271876665722593,
      "learning_rate": 9.887028406010026e-06,
      "loss": 0.4987,
      "step": 880
    },
    {
      "epoch": 0.16141443752290216,
      "grad_norm": 0.39991523006604696,
      "learning_rate": 9.886351465141785e-06,
      "loss": 0.4807,
      "step": 881
    },
    {
      "epoch": 0.16159765481861488,
      "grad_norm": 0.3867195675970219,
      "learning_rate": 9.88567252548102e-06,
      "loss": 0.515,
      "step": 882
    },
    {
      "epoch": 0.1617808721143276,
      "grad_norm": 0.47566001886735476,
      "learning_rate": 9.884991587305459e-06,
      "loss": 0.4461,
      "step": 883
    },
    {
      "epoch": 0.1619640894100403,
      "grad_norm": 0.4453653597243864,
      "learning_rate": 9.884308650893642e-06,
      "loss": 0.5493,
      "step": 884
    },
    {
      "epoch": 0.16214730670575303,
      "grad_norm": 0.38787040868110323,
      "learning_rate": 9.883623716524929e-06,
      "loss": 0.4998,
      "step": 885
    },
    {
      "epoch": 0.16233052400146575,
      "grad_norm": 0.4570258592705414,
      "learning_rate": 9.882936784479498e-06,
      "loss": 0.5226,
      "step": 886
    },
    {
      "epoch": 0.16251374129717847,
      "grad_norm": 0.4438200453897428,
      "learning_rate": 9.882247855038339e-06,
      "loss": 0.5073,
      "step": 887
    },
    {
      "epoch": 0.16269695859289116,
      "grad_norm": 0.3887912773501902,
      "learning_rate": 9.881556928483266e-06,
      "loss": 0.5213,
      "step": 888
    },
    {
      "epoch": 0.16288017588860387,
      "grad_norm": 0.44048355093202696,
      "learning_rate": 9.880864005096906e-06,
      "loss": 0.511,
      "step": 889
    },
    {
      "epoch": 0.1630633931843166,
      "grad_norm": 0.44235079469227967,
      "learning_rate": 9.880169085162703e-06,
      "loss": 0.5223,
      "step": 890
    },
    {
      "epoch": 0.1632466104800293,
      "grad_norm": 0.3877832289424,
      "learning_rate": 9.87947216896492e-06,
      "loss": 0.4537,
      "step": 891
    },
    {
      "epoch": 0.16342982777574203,
      "grad_norm": 0.44869938145307137,
      "learning_rate": 9.878773256788635e-06,
      "loss": 0.5013,
      "step": 892
    },
    {
      "epoch": 0.16361304507145474,
      "grad_norm": 0.46375171983760244,
      "learning_rate": 9.878072348919738e-06,
      "loss": 0.489,
      "step": 893
    },
    {
      "epoch": 0.16379626236716746,
      "grad_norm": 0.49608852880970994,
      "learning_rate": 9.877369445644945e-06,
      "loss": 0.5177,
      "step": 894
    },
    {
      "epoch": 0.16397947966288018,
      "grad_norm": 0.4082183549684993,
      "learning_rate": 9.876664547251781e-06,
      "loss": 0.5091,
      "step": 895
    },
    {
      "epoch": 0.1641626969585929,
      "grad_norm": 0.4347968628628821,
      "learning_rate": 9.875957654028588e-06,
      "loss": 0.5242,
      "step": 896
    },
    {
      "epoch": 0.16434591425430561,
      "grad_norm": 0.45545255399303036,
      "learning_rate": 9.875248766264527e-06,
      "loss": 0.5406,
      "step": 897
    },
    {
      "epoch": 0.16452913155001833,
      "grad_norm": 0.41358644063630506,
      "learning_rate": 9.874537884249574e-06,
      "loss": 0.511,
      "step": 898
    },
    {
      "epoch": 0.16471234884573105,
      "grad_norm": 0.4576667619206321,
      "learning_rate": 9.873825008274514e-06,
      "loss": 0.5073,
      "step": 899
    },
    {
      "epoch": 0.16489556614144374,
      "grad_norm": 0.47721781865779744,
      "learning_rate": 9.87311013863096e-06,
      "loss": 0.5044,
      "step": 900
    },
    {
      "epoch": 0.16507878343715646,
      "grad_norm": 0.4622288619794578,
      "learning_rate": 9.872393275611329e-06,
      "loss": 0.5096,
      "step": 901
    },
    {
      "epoch": 0.16526200073286917,
      "grad_norm": 0.41177688817899083,
      "learning_rate": 9.871674419508864e-06,
      "loss": 0.523,
      "step": 902
    },
    {
      "epoch": 0.1654452180285819,
      "grad_norm": 0.371054810469729,
      "learning_rate": 9.87095357061761e-06,
      "loss": 0.4897,
      "step": 903
    },
    {
      "epoch": 0.1656284353242946,
      "grad_norm": 0.41574608839188226,
      "learning_rate": 9.87023072923244e-06,
      "loss": 0.4823,
      "step": 904
    },
    {
      "epoch": 0.16581165262000733,
      "grad_norm": 0.4218915251343208,
      "learning_rate": 9.869505895649036e-06,
      "loss": 0.5079,
      "step": 905
    },
    {
      "epoch": 0.16599486991572004,
      "grad_norm": 0.5549831381708766,
      "learning_rate": 9.868779070163895e-06,
      "loss": 0.484,
      "step": 906
    },
    {
      "epoch": 0.16617808721143276,
      "grad_norm": 0.3891473564827362,
      "learning_rate": 9.868050253074328e-06,
      "loss": 0.4903,
      "step": 907
    },
    {
      "epoch": 0.16636130450714548,
      "grad_norm": 0.4268945705632106,
      "learning_rate": 9.867319444678465e-06,
      "loss": 0.5024,
      "step": 908
    },
    {
      "epoch": 0.1665445218028582,
      "grad_norm": 0.40332206359603545,
      "learning_rate": 9.866586645275247e-06,
      "loss": 0.4477,
      "step": 909
    },
    {
      "epoch": 0.16672773909857092,
      "grad_norm": 0.3720747589583317,
      "learning_rate": 9.86585185516443e-06,
      "loss": 0.469,
      "step": 910
    },
    {
      "epoch": 0.16691095639428363,
      "grad_norm": 0.451113552639029,
      "learning_rate": 9.865115074646583e-06,
      "loss": 0.5221,
      "step": 911
    },
    {
      "epoch": 0.16709417368999632,
      "grad_norm": 0.4454998332020719,
      "learning_rate": 9.864376304023092e-06,
      "loss": 0.495,
      "step": 912
    },
    {
      "epoch": 0.16727739098570904,
      "grad_norm": 0.5363910151131063,
      "learning_rate": 9.863635543596156e-06,
      "loss": 0.5233,
      "step": 913
    },
    {
      "epoch": 0.16746060828142176,
      "grad_norm": 0.4392403856880424,
      "learning_rate": 9.862892793668787e-06,
      "loss": 0.4998,
      "step": 914
    },
    {
      "epoch": 0.16764382557713448,
      "grad_norm": 0.42093143672832734,
      "learning_rate": 9.862148054544812e-06,
      "loss": 0.4977,
      "step": 915
    },
    {
      "epoch": 0.1678270428728472,
      "grad_norm": 0.39275787265485074,
      "learning_rate": 9.86140132652887e-06,
      "loss": 0.4694,
      "step": 916
    },
    {
      "epoch": 0.1680102601685599,
      "grad_norm": 0.4159908861548446,
      "learning_rate": 9.860652609926417e-06,
      "loss": 0.4886,
      "step": 917
    },
    {
      "epoch": 0.16819347746427263,
      "grad_norm": 0.38579180273261776,
      "learning_rate": 9.859901905043718e-06,
      "loss": 0.48,
      "step": 918
    },
    {
      "epoch": 0.16837669475998535,
      "grad_norm": 0.4273621763370794,
      "learning_rate": 9.859149212187855e-06,
      "loss": 0.512,
      "step": 919
    },
    {
      "epoch": 0.16855991205569806,
      "grad_norm": 0.45913559298162226,
      "learning_rate": 9.85839453166672e-06,
      "loss": 0.4877,
      "step": 920
    },
    {
      "epoch": 0.16874312935141078,
      "grad_norm": 0.3847363797846655,
      "learning_rate": 9.85763786378902e-06,
      "loss": 0.4586,
      "step": 921
    },
    {
      "epoch": 0.1689263466471235,
      "grad_norm": 0.4872256895625686,
      "learning_rate": 9.856879208864277e-06,
      "loss": 0.5035,
      "step": 922
    },
    {
      "epoch": 0.16910956394283622,
      "grad_norm": 0.47756329412725707,
      "learning_rate": 9.85611856720282e-06,
      "loss": 0.5031,
      "step": 923
    },
    {
      "epoch": 0.1692927812385489,
      "grad_norm": 0.474431940864528,
      "learning_rate": 9.8553559391158e-06,
      "loss": 0.512,
      "step": 924
    },
    {
      "epoch": 0.16947599853426162,
      "grad_norm": 0.41972238109665655,
      "learning_rate": 9.85459132491517e-06,
      "loss": 0.5386,
      "step": 925
    },
    {
      "epoch": 0.16965921582997434,
      "grad_norm": 0.4369771091332424,
      "learning_rate": 9.8538247249137e-06,
      "loss": 0.4909,
      "step": 926
    },
    {
      "epoch": 0.16984243312568706,
      "grad_norm": 0.4523772832392917,
      "learning_rate": 9.853056139424974e-06,
      "loss": 0.4951,
      "step": 927
    },
    {
      "epoch": 0.17002565042139978,
      "grad_norm": 0.40235086982371876,
      "learning_rate": 9.852285568763387e-06,
      "loss": 0.4828,
      "step": 928
    },
    {
      "epoch": 0.1702088677171125,
      "grad_norm": 0.4150844402567988,
      "learning_rate": 9.851513013244144e-06,
      "loss": 0.4933,
      "step": 929
    },
    {
      "epoch": 0.1703920850128252,
      "grad_norm": 0.3941602147072389,
      "learning_rate": 9.850738473183266e-06,
      "loss": 0.4855,
      "step": 930
    },
    {
      "epoch": 0.17057530230853793,
      "grad_norm": 0.47118394280764797,
      "learning_rate": 9.849961948897582e-06,
      "loss": 0.5024,
      "step": 931
    },
    {
      "epoch": 0.17075851960425065,
      "grad_norm": 0.3912878663413955,
      "learning_rate": 9.849183440704735e-06,
      "loss": 0.5041,
      "step": 932
    },
    {
      "epoch": 0.17094173689996336,
      "grad_norm": 0.4001961484144372,
      "learning_rate": 9.848402948923177e-06,
      "loss": 0.4753,
      "step": 933
    },
    {
      "epoch": 0.17112495419567608,
      "grad_norm": 0.4441137892348275,
      "learning_rate": 9.847620473872172e-06,
      "loss": 0.5102,
      "step": 934
    },
    {
      "epoch": 0.1713081714913888,
      "grad_norm": 0.40101758764349804,
      "learning_rate": 9.846836015871802e-06,
      "loss": 0.4824,
      "step": 935
    },
    {
      "epoch": 0.1714913887871015,
      "grad_norm": 0.41075944565503664,
      "learning_rate": 9.846049575242949e-06,
      "loss": 0.5235,
      "step": 936
    },
    {
      "epoch": 0.1716746060828142,
      "grad_norm": 0.3780837940291423,
      "learning_rate": 9.845261152307312e-06,
      "loss": 0.5211,
      "step": 937
    },
    {
      "epoch": 0.17185782337852692,
      "grad_norm": 0.45070991069476113,
      "learning_rate": 9.844470747387403e-06,
      "loss": 0.4926,
      "step": 938
    },
    {
      "epoch": 0.17204104067423964,
      "grad_norm": 0.40836878453605646,
      "learning_rate": 9.843678360806542e-06,
      "loss": 0.5173,
      "step": 939
    },
    {
      "epoch": 0.17222425796995236,
      "grad_norm": 0.36676941914948374,
      "learning_rate": 9.842883992888855e-06,
      "loss": 0.5101,
      "step": 940
    },
    {
      "epoch": 0.17240747526566508,
      "grad_norm": 0.42894488109200807,
      "learning_rate": 9.842087643959288e-06,
      "loss": 0.5398,
      "step": 941
    },
    {
      "epoch": 0.1725906925613778,
      "grad_norm": 0.4004768449690441,
      "learning_rate": 9.841289314343591e-06,
      "loss": 0.4967,
      "step": 942
    },
    {
      "epoch": 0.1727739098570905,
      "grad_norm": 0.3781076276341992,
      "learning_rate": 9.840489004368325e-06,
      "loss": 0.4731,
      "step": 943
    },
    {
      "epoch": 0.17295712715280323,
      "grad_norm": 0.45307945315267256,
      "learning_rate": 9.839686714360864e-06,
      "loss": 0.4737,
      "step": 944
    },
    {
      "epoch": 0.17314034444851595,
      "grad_norm": 0.4079247794136399,
      "learning_rate": 9.838882444649387e-06,
      "loss": 0.4927,
      "step": 945
    },
    {
      "epoch": 0.17332356174422867,
      "grad_norm": 0.47171018664243136,
      "learning_rate": 9.838076195562886e-06,
      "loss": 0.5009,
      "step": 946
    },
    {
      "epoch": 0.17350677903994138,
      "grad_norm": 0.44572108382734166,
      "learning_rate": 9.837267967431164e-06,
      "loss": 0.4858,
      "step": 947
    },
    {
      "epoch": 0.17368999633565407,
      "grad_norm": 0.40649730883742624,
      "learning_rate": 9.83645776058483e-06,
      "loss": 0.4943,
      "step": 948
    },
    {
      "epoch": 0.1738732136313668,
      "grad_norm": 0.43002759908477306,
      "learning_rate": 9.835645575355304e-06,
      "loss": 0.4872,
      "step": 949
    },
    {
      "epoch": 0.1740564309270795,
      "grad_norm": 0.39064612647120167,
      "learning_rate": 9.834831412074816e-06,
      "loss": 0.4409,
      "step": 950
    },
    {
      "epoch": 0.17423964822279223,
      "grad_norm": 0.3761573603114394,
      "learning_rate": 9.834015271076405e-06,
      "loss": 0.459,
      "step": 951
    },
    {
      "epoch": 0.17442286551850494,
      "grad_norm": 0.42796519720566795,
      "learning_rate": 9.83319715269392e-06,
      "loss": 0.4906,
      "step": 952
    },
    {
      "epoch": 0.17460608281421766,
      "grad_norm": 0.5091807084893948,
      "learning_rate": 9.832377057262015e-06,
      "loss": 0.5516,
      "step": 953
    },
    {
      "epoch": 0.17478930010993038,
      "grad_norm": 0.48065067696295527,
      "learning_rate": 9.831554985116155e-06,
      "loss": 0.5097,
      "step": 954
    },
    {
      "epoch": 0.1749725174056431,
      "grad_norm": 0.4310364985277951,
      "learning_rate": 9.830730936592615e-06,
      "loss": 0.5025,
      "step": 955
    },
    {
      "epoch": 0.17515573470135581,
      "grad_norm": 0.39964086093662415,
      "learning_rate": 9.829904912028477e-06,
      "loss": 0.494,
      "step": 956
    },
    {
      "epoch": 0.17533895199706853,
      "grad_norm": 0.40935380850458175,
      "learning_rate": 9.829076911761631e-06,
      "loss": 0.5182,
      "step": 957
    },
    {
      "epoch": 0.17552216929278125,
      "grad_norm": 0.454766087870644,
      "learning_rate": 9.828246936130777e-06,
      "loss": 0.5113,
      "step": 958
    },
    {
      "epoch": 0.17570538658849397,
      "grad_norm": 0.5087972352515657,
      "learning_rate": 9.827414985475419e-06,
      "loss": 0.4797,
      "step": 959
    },
    {
      "epoch": 0.17588860388420666,
      "grad_norm": 0.4255562501722622,
      "learning_rate": 9.826581060135873e-06,
      "loss": 0.5045,
      "step": 960
    },
    {
      "epoch": 0.17607182117991937,
      "grad_norm": 0.45583787528367603,
      "learning_rate": 9.825745160453264e-06,
      "loss": 0.4755,
      "step": 961
    },
    {
      "epoch": 0.1762550384756321,
      "grad_norm": 0.45648883029913284,
      "learning_rate": 9.824907286769519e-06,
      "loss": 0.4736,
      "step": 962
    },
    {
      "epoch": 0.1764382557713448,
      "grad_norm": 0.436071698669034,
      "learning_rate": 9.824067439427374e-06,
      "loss": 0.5036,
      "step": 963
    },
    {
      "epoch": 0.17662147306705753,
      "grad_norm": 0.41502000197263017,
      "learning_rate": 9.823225618770378e-06,
      "loss": 0.5121,
      "step": 964
    },
    {
      "epoch": 0.17680469036277024,
      "grad_norm": 0.450252173902017,
      "learning_rate": 9.822381825142879e-06,
      "loss": 0.506,
      "step": 965
    },
    {
      "epoch": 0.17698790765848296,
      "grad_norm": 0.43561778776659554,
      "learning_rate": 9.82153605889004e-06,
      "loss": 0.5304,
      "step": 966
    },
    {
      "epoch": 0.17717112495419568,
      "grad_norm": 0.4306520452468297,
      "learning_rate": 9.820688320357823e-06,
      "loss": 0.5112,
      "step": 967
    },
    {
      "epoch": 0.1773543422499084,
      "grad_norm": 0.39680977809710444,
      "learning_rate": 9.819838609893005e-06,
      "loss": 0.4904,
      "step": 968
    },
    {
      "epoch": 0.17753755954562112,
      "grad_norm": 0.4225898008101114,
      "learning_rate": 9.81898692784316e-06,
      "loss": 0.5137,
      "step": 969
    },
    {
      "epoch": 0.17772077684133383,
      "grad_norm": 0.4267062187327317,
      "learning_rate": 9.818133274556679e-06,
      "loss": 0.517,
      "step": 970
    },
    {
      "epoch": 0.17790399413704655,
      "grad_norm": 0.4450452397955529,
      "learning_rate": 9.81727765038275e-06,
      "loss": 0.4949,
      "step": 971
    },
    {
      "epoch": 0.17808721143275924,
      "grad_norm": 0.42730228271929965,
      "learning_rate": 9.816420055671374e-06,
      "loss": 0.5013,
      "step": 972
    },
    {
      "epoch": 0.17827042872847196,
      "grad_norm": 0.4421383626296405,
      "learning_rate": 9.815560490773356e-06,
      "loss": 0.4679,
      "step": 973
    },
    {
      "epoch": 0.17845364602418468,
      "grad_norm": 0.43307265096730885,
      "learning_rate": 9.814698956040305e-06,
      "loss": 0.4976,
      "step": 974
    },
    {
      "epoch": 0.1786368633198974,
      "grad_norm": 0.40376464412932705,
      "learning_rate": 9.813835451824636e-06,
      "loss": 0.5021,
      "step": 975
    },
    {
      "epoch": 0.1788200806156101,
      "grad_norm": 0.44775776339538953,
      "learning_rate": 9.812969978479573e-06,
      "loss": 0.4824,
      "step": 976
    },
    {
      "epoch": 0.17900329791132283,
      "grad_norm": 0.468036097947629,
      "learning_rate": 9.812102536359142e-06,
      "loss": 0.4966,
      "step": 977
    },
    {
      "epoch": 0.17918651520703555,
      "grad_norm": 0.49832149697786493,
      "learning_rate": 9.811233125818176e-06,
      "loss": 0.4853,
      "step": 978
    },
    {
      "epoch": 0.17936973250274826,
      "grad_norm": 0.44754723833693727,
      "learning_rate": 9.810361747212313e-06,
      "loss": 0.4999,
      "step": 979
    },
    {
      "epoch": 0.17955294979846098,
      "grad_norm": 0.4747011696315986,
      "learning_rate": 9.809488400897996e-06,
      "loss": 0.5108,
      "step": 980
    },
    {
      "epoch": 0.1797361670941737,
      "grad_norm": 0.4055842519238836,
      "learning_rate": 9.808613087232473e-06,
      "loss": 0.4786,
      "step": 981
    },
    {
      "epoch": 0.17991938438988642,
      "grad_norm": 0.3971701443501286,
      "learning_rate": 9.807735806573795e-06,
      "loss": 0.4925,
      "step": 982
    },
    {
      "epoch": 0.18010260168559913,
      "grad_norm": 0.3784480421793096,
      "learning_rate": 9.806856559280819e-06,
      "loss": 0.4762,
      "step": 983
    },
    {
      "epoch": 0.18028581898131182,
      "grad_norm": 0.41465183173286063,
      "learning_rate": 9.80597534571321e-06,
      "loss": 0.5037,
      "step": 984
    },
    {
      "epoch": 0.18046903627702454,
      "grad_norm": 0.4713550462839489,
      "learning_rate": 9.80509216623143e-06,
      "loss": 0.4816,
      "step": 985
    },
    {
      "epoch": 0.18065225357273726,
      "grad_norm": 0.34923786309805893,
      "learning_rate": 9.804207021196751e-06,
      "loss": 0.4903,
      "step": 986
    },
    {
      "epoch": 0.18083547086844998,
      "grad_norm": 0.4359203385963395,
      "learning_rate": 9.803319910971248e-06,
      "loss": 0.5252,
      "step": 987
    },
    {
      "epoch": 0.1810186881641627,
      "grad_norm": 0.40286824933567683,
      "learning_rate": 9.802430835917796e-06,
      "loss": 0.5248,
      "step": 988
    },
    {
      "epoch": 0.1812019054598754,
      "grad_norm": 0.44314680454994215,
      "learning_rate": 9.801539796400078e-06,
      "loss": 0.5071,
      "step": 989
    },
    {
      "epoch": 0.18138512275558813,
      "grad_norm": 0.44074612784921385,
      "learning_rate": 9.80064679278258e-06,
      "loss": 0.5057,
      "step": 990
    },
    {
      "epoch": 0.18156834005130085,
      "grad_norm": 0.40968360376086815,
      "learning_rate": 9.799751825430592e-06,
      "loss": 0.4919,
      "step": 991
    },
    {
      "epoch": 0.18175155734701356,
      "grad_norm": 0.4204735967781832,
      "learning_rate": 9.798854894710202e-06,
      "loss": 0.546,
      "step": 992
    },
    {
      "epoch": 0.18193477464272628,
      "grad_norm": 0.36210624498761007,
      "learning_rate": 9.79795600098831e-06,
      "loss": 0.4904,
      "step": 993
    },
    {
      "epoch": 0.182117991938439,
      "grad_norm": 0.45193717949202633,
      "learning_rate": 9.797055144632609e-06,
      "loss": 0.5292,
      "step": 994
    },
    {
      "epoch": 0.18230120923415172,
      "grad_norm": 0.415630496229026,
      "learning_rate": 9.796152326011604e-06,
      "loss": 0.533,
      "step": 995
    },
    {
      "epoch": 0.1824844265298644,
      "grad_norm": 0.39372729416385926,
      "learning_rate": 9.795247545494594e-06,
      "loss": 0.5147,
      "step": 996
    },
    {
      "epoch": 0.18266764382557713,
      "grad_norm": 0.43540226077312505,
      "learning_rate": 9.794340803451692e-06,
      "loss": 0.49,
      "step": 997
    },
    {
      "epoch": 0.18285086112128984,
      "grad_norm": 0.39372791981341043,
      "learning_rate": 9.7934321002538e-06,
      "loss": 0.4776,
      "step": 998
    },
    {
      "epoch": 0.18303407841700256,
      "grad_norm": 0.40381508343023165,
      "learning_rate": 9.792521436272633e-06,
      "loss": 0.4566,
      "step": 999
    },
    {
      "epoch": 0.18321729571271528,
      "grad_norm": 0.47729065322343966,
      "learning_rate": 9.791608811880702e-06,
      "loss": 0.5013,
      "step": 1000
    }
  ],
  "logging_steps": 1,
  "max_steps": 5458,
  "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": false
      },
      "attributes": {}
    }
  },
  "total_flos": 178891845599232.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}