{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.169658177331354,
  "eval_steps": 500,
  "global_step": 1100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00015423470666486725,
      "grad_norm": 4.125617980957031,
      "learning_rate": 0.0,
      "loss": 2.7158,
      "step": 1
    },
    {
      "epoch": 0.0003084694133297345,
      "grad_norm": 4.080927848815918,
      "learning_rate": 5.1282051282051286e-08,
      "loss": 2.7259,
      "step": 2
    },
    {
      "epoch": 0.0004627041199946018,
      "grad_norm": 4.546864986419678,
      "learning_rate": 1.0256410256410257e-07,
      "loss": 2.7844,
      "step": 3
    },
    {
      "epoch": 0.000616938826659469,
      "grad_norm": 3.869814872741699,
      "learning_rate": 1.5384615384615387e-07,
      "loss": 2.7772,
      "step": 4
    },
    {
      "epoch": 0.0007711735333243363,
      "grad_norm": 3.8191957473754883,
      "learning_rate": 2.0512820512820514e-07,
      "loss": 2.669,
      "step": 5
    },
    {
      "epoch": 0.0009254082399892036,
      "grad_norm": 4.232642650604248,
      "learning_rate": 2.564102564102564e-07,
      "loss": 2.7774,
      "step": 6
    },
    {
      "epoch": 0.001079642946654071,
      "grad_norm": 4.562683582305908,
      "learning_rate": 3.0769230769230774e-07,
      "loss": 2.8335,
      "step": 7
    },
    {
      "epoch": 0.001233877653318938,
      "grad_norm": 4.024895668029785,
      "learning_rate": 3.5897435897435896e-07,
      "loss": 2.7137,
      "step": 8
    },
    {
      "epoch": 0.0013881123599838053,
      "grad_norm": 4.051122665405273,
      "learning_rate": 4.102564102564103e-07,
      "loss": 2.6775,
      "step": 9
    },
    {
      "epoch": 0.0015423470666486726,
      "grad_norm": 4.009105682373047,
      "learning_rate": 4.615384615384616e-07,
      "loss": 2.7797,
      "step": 10
    },
    {
      "epoch": 0.00169658177331354,
      "grad_norm": 3.7827749252319336,
      "learning_rate": 5.128205128205128e-07,
      "loss": 2.7706,
      "step": 11
    },
    {
      "epoch": 0.0018508164799784072,
      "grad_norm": 3.986659288406372,
      "learning_rate": 5.641025641025642e-07,
      "loss": 2.7587,
      "step": 12
    },
    {
      "epoch": 0.0020050511866432745,
      "grad_norm": 4.076365947723389,
      "learning_rate": 6.153846153846155e-07,
      "loss": 2.7406,
      "step": 13
    },
    {
      "epoch": 0.002159285893308142,
      "grad_norm": 3.7137603759765625,
      "learning_rate": 6.666666666666667e-07,
      "loss": 2.6706,
      "step": 14
    },
    {
      "epoch": 0.002313520599973009,
      "grad_norm": 3.197049140930176,
      "learning_rate": 7.179487179487179e-07,
      "loss": 2.5892,
      "step": 15
    },
    {
      "epoch": 0.002467755306637876,
      "grad_norm": 3.490769147872925,
      "learning_rate": 7.692307692307694e-07,
      "loss": 2.6735,
      "step": 16
    },
    {
      "epoch": 0.0026219900133027433,
      "grad_norm": 3.234196662902832,
      "learning_rate": 8.205128205128206e-07,
      "loss": 2.5749,
      "step": 17
    },
    {
      "epoch": 0.0027762247199676106,
      "grad_norm": 3.0578436851501465,
      "learning_rate": 8.717948717948718e-07,
      "loss": 2.5744,
      "step": 18
    },
    {
      "epoch": 0.002930459426632478,
      "grad_norm": 3.03320574760437,
      "learning_rate": 9.230769230769232e-07,
      "loss": 2.5946,
      "step": 19
    },
    {
      "epoch": 0.003084694133297345,
      "grad_norm": 1.8710768222808838,
      "learning_rate": 9.743589743589745e-07,
      "loss": 2.4013,
      "step": 20
    },
    {
      "epoch": 0.0032389288399622125,
      "grad_norm": 1.9252475500106812,
      "learning_rate": 1.0256410256410257e-06,
      "loss": 2.3882,
      "step": 21
    },
    {
      "epoch": 0.00339316354662708,
      "grad_norm": 5.356478691101074,
      "learning_rate": 1.076923076923077e-06,
      "loss": 2.4184,
      "step": 22
    },
    {
      "epoch": 0.003547398253291947,
      "grad_norm": 1.6710164546966553,
      "learning_rate": 1.1282051282051283e-06,
      "loss": 2.3395,
      "step": 23
    },
    {
      "epoch": 0.0037016329599568144,
      "grad_norm": 1.7735917568206787,
      "learning_rate": 1.1794871794871795e-06,
      "loss": 2.35,
      "step": 24
    },
    {
      "epoch": 0.0038558676666216817,
      "grad_norm": 1.9577521085739136,
      "learning_rate": 1.230769230769231e-06,
      "loss": 2.3646,
      "step": 25
    },
    {
      "epoch": 0.004010102373286549,
      "grad_norm": 1.5192971229553223,
      "learning_rate": 1.282051282051282e-06,
      "loss": 2.2818,
      "step": 26
    },
    {
      "epoch": 0.004164337079951416,
      "grad_norm": 1.4385325908660889,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 2.2448,
      "step": 27
    },
    {
      "epoch": 0.004318571786616284,
      "grad_norm": 1.439254879951477,
      "learning_rate": 1.3846153846153848e-06,
      "loss": 2.1641,
      "step": 28
    },
    {
      "epoch": 0.0044728064932811505,
      "grad_norm": 1.5258543491363525,
      "learning_rate": 1.4358974358974359e-06,
      "loss": 2.2035,
      "step": 29
    },
    {
      "epoch": 0.004627041199946018,
      "grad_norm": 1.4886516332626343,
      "learning_rate": 1.4871794871794873e-06,
      "loss": 2.1415,
      "step": 30
    },
    {
      "epoch": 0.004781275906610885,
      "grad_norm": 1.3116540908813477,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 2.1337,
      "step": 31
    },
    {
      "epoch": 0.004935510613275752,
      "grad_norm": 1.0612387657165527,
      "learning_rate": 1.5897435897435897e-06,
      "loss": 2.1492,
      "step": 32
    },
    {
      "epoch": 0.00508974531994062,
      "grad_norm": 1.0794328451156616,
      "learning_rate": 1.6410256410256412e-06,
      "loss": 2.1277,
      "step": 33
    },
    {
      "epoch": 0.0052439800266054866,
      "grad_norm": 1.0971394777297974,
      "learning_rate": 1.6923076923076926e-06,
      "loss": 2.0956,
      "step": 34
    },
    {
      "epoch": 0.005398214733270354,
      "grad_norm": 1.0308690071105957,
      "learning_rate": 1.7435897435897436e-06,
      "loss": 2.1688,
      "step": 35
    },
    {
      "epoch": 0.005552449439935221,
      "grad_norm": 0.8738017678260803,
      "learning_rate": 1.794871794871795e-06,
      "loss": 2.1154,
      "step": 36
    },
    {
      "epoch": 0.005706684146600089,
      "grad_norm": 0.8435266613960266,
      "learning_rate": 1.8461538461538465e-06,
      "loss": 2.1031,
      "step": 37
    },
    {
      "epoch": 0.005860918853264956,
      "grad_norm": 0.865908682346344,
      "learning_rate": 1.8974358974358975e-06,
      "loss": 2.0727,
      "step": 38
    },
    {
      "epoch": 0.0060151535599298235,
      "grad_norm": 0.8485544323921204,
      "learning_rate": 1.948717948717949e-06,
      "loss": 2.0476,
      "step": 39
    },
    {
      "epoch": 0.00616938826659469,
      "grad_norm": 0.8207165598869324,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 2.0544,
      "step": 40
    },
    {
      "epoch": 0.006323622973259558,
      "grad_norm": 0.93079674243927,
      "learning_rate": 2.0512820512820513e-06,
      "loss": 2.0403,
      "step": 41
    },
    {
      "epoch": 0.006477857679924425,
      "grad_norm": 0.7712639570236206,
      "learning_rate": 2.1025641025641028e-06,
      "loss": 1.9826,
      "step": 42
    },
    {
      "epoch": 0.006632092386589292,
      "grad_norm": 0.7859727740287781,
      "learning_rate": 2.153846153846154e-06,
      "loss": 2.01,
      "step": 43
    },
    {
      "epoch": 0.00678632709325416,
      "grad_norm": 0.8241223692893982,
      "learning_rate": 2.2051282051282052e-06,
      "loss": 1.996,
      "step": 44
    },
    {
      "epoch": 0.0069405617999190265,
      "grad_norm": 0.6832298636436462,
      "learning_rate": 2.2564102564102566e-06,
      "loss": 2.0134,
      "step": 45
    },
    {
      "epoch": 0.007094796506583894,
      "grad_norm": 0.7749252915382385,
      "learning_rate": 2.307692307692308e-06,
      "loss": 2.0027,
      "step": 46
    },
    {
      "epoch": 0.007249031213248761,
      "grad_norm": 0.7955232262611389,
      "learning_rate": 2.358974358974359e-06,
      "loss": 1.9854,
      "step": 47
    },
    {
      "epoch": 0.007403265919913629,
      "grad_norm": 0.6382924914360046,
      "learning_rate": 2.4102564102564105e-06,
      "loss": 1.9807,
      "step": 48
    },
    {
      "epoch": 0.007557500626578496,
      "grad_norm": 0.6480730772018433,
      "learning_rate": 2.461538461538462e-06,
      "loss": 1.9236,
      "step": 49
    },
    {
      "epoch": 0.007711735333243363,
      "grad_norm": 0.6792906522750854,
      "learning_rate": 2.512820512820513e-06,
      "loss": 1.9633,
      "step": 50
    },
    {
      "epoch": 0.007865970039908231,
      "grad_norm": 0.6324858665466309,
      "learning_rate": 2.564102564102564e-06,
      "loss": 1.9083,
      "step": 51
    },
    {
      "epoch": 0.008020204746573098,
      "grad_norm": 0.6173593997955322,
      "learning_rate": 2.615384615384616e-06,
      "loss": 1.9377,
      "step": 52
    },
    {
      "epoch": 0.008174439453237965,
      "grad_norm": 0.6734508275985718,
      "learning_rate": 2.666666666666667e-06,
      "loss": 1.8874,
      "step": 53
    },
    {
      "epoch": 0.008328674159902832,
      "grad_norm": 0.6378979086875916,
      "learning_rate": 2.717948717948718e-06,
      "loss": 1.9492,
      "step": 54
    },
    {
      "epoch": 0.008482908866567699,
      "grad_norm": 0.6195733547210693,
      "learning_rate": 2.7692307692307697e-06,
      "loss": 1.9294,
      "step": 55
    },
    {
      "epoch": 0.008637143573232567,
      "grad_norm": 0.6716891527175903,
      "learning_rate": 2.8205128205128207e-06,
      "loss": 1.8872,
      "step": 56
    },
    {
      "epoch": 0.008791378279897434,
      "grad_norm": 0.6091171503067017,
      "learning_rate": 2.8717948717948717e-06,
      "loss": 1.9162,
      "step": 57
    },
    {
      "epoch": 0.008945612986562301,
      "grad_norm": 0.6023731231689453,
      "learning_rate": 2.9230769230769236e-06,
      "loss": 1.8666,
      "step": 58
    },
    {
      "epoch": 0.009099847693227168,
      "grad_norm": 0.5978520512580872,
      "learning_rate": 2.9743589743589746e-06,
      "loss": 1.9221,
      "step": 59
    },
    {
      "epoch": 0.009254082399892036,
      "grad_norm": 0.6359299421310425,
      "learning_rate": 3.0256410256410256e-06,
      "loss": 1.8786,
      "step": 60
    },
    {
      "epoch": 0.009408317106556903,
      "grad_norm": 0.5901927947998047,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 1.8436,
      "step": 61
    },
    {
      "epoch": 0.00956255181322177,
      "grad_norm": 0.5627117156982422,
      "learning_rate": 3.1282051282051284e-06,
      "loss": 1.866,
      "step": 62
    },
    {
      "epoch": 0.009716786519886637,
      "grad_norm": 0.6476024389266968,
      "learning_rate": 3.1794871794871795e-06,
      "loss": 1.8644,
      "step": 63
    },
    {
      "epoch": 0.009871021226551504,
      "grad_norm": 0.5732126832008362,
      "learning_rate": 3.2307692307692313e-06,
      "loss": 1.7929,
      "step": 64
    },
    {
      "epoch": 0.010025255933216373,
      "grad_norm": 0.5641695857048035,
      "learning_rate": 3.2820512820512823e-06,
      "loss": 1.827,
      "step": 65
    },
    {
      "epoch": 0.01017949063988124,
      "grad_norm": 0.6193923950195312,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.8669,
      "step": 66
    },
    {
      "epoch": 0.010333725346546106,
      "grad_norm": 0.5760056376457214,
      "learning_rate": 3.384615384615385e-06,
      "loss": 1.847,
      "step": 67
    },
    {
      "epoch": 0.010487960053210973,
      "grad_norm": 0.585292637348175,
      "learning_rate": 3.435897435897436e-06,
      "loss": 1.8246,
      "step": 68
    },
    {
      "epoch": 0.010642194759875842,
      "grad_norm": 0.6549813151359558,
      "learning_rate": 3.487179487179487e-06,
      "loss": 1.8534,
      "step": 69
    },
    {
      "epoch": 0.010796429466540709,
      "grad_norm": 0.5588355660438538,
      "learning_rate": 3.538461538461539e-06,
      "loss": 1.8266,
      "step": 70
    },
    {
      "epoch": 0.010950664173205575,
      "grad_norm": 0.5905765891075134,
      "learning_rate": 3.58974358974359e-06,
      "loss": 1.8236,
      "step": 71
    },
    {
      "epoch": 0.011104898879870442,
      "grad_norm": 0.6020631790161133,
      "learning_rate": 3.641025641025641e-06,
      "loss": 1.8284,
      "step": 72
    },
    {
      "epoch": 0.011259133586535311,
      "grad_norm": 0.5250481367111206,
      "learning_rate": 3.692307692307693e-06,
      "loss": 1.7922,
      "step": 73
    },
    {
      "epoch": 0.011413368293200178,
      "grad_norm": 0.5569522976875305,
      "learning_rate": 3.743589743589744e-06,
      "loss": 1.785,
      "step": 74
    },
    {
      "epoch": 0.011567602999865045,
      "grad_norm": 0.5544557571411133,
      "learning_rate": 3.794871794871795e-06,
      "loss": 1.7892,
      "step": 75
    },
    {
      "epoch": 0.011721837706529912,
      "grad_norm": 0.6039264798164368,
      "learning_rate": 3.846153846153847e-06,
      "loss": 1.7843,
      "step": 76
    },
    {
      "epoch": 0.011876072413194778,
      "grad_norm": 0.5743536949157715,
      "learning_rate": 3.897435897435898e-06,
      "loss": 1.757,
      "step": 77
    },
    {
      "epoch": 0.012030307119859647,
      "grad_norm": 0.5441644787788391,
      "learning_rate": 3.948717948717949e-06,
      "loss": 1.755,
      "step": 78
    },
    {
      "epoch": 0.012184541826524514,
      "grad_norm": 0.5642723441123962,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.7981,
      "step": 79
    },
    {
      "epoch": 0.01233877653318938,
      "grad_norm": 0.563156008720398,
      "learning_rate": 4.051282051282052e-06,
      "loss": 1.7576,
      "step": 80
    },
    {
      "epoch": 0.012493011239854248,
      "grad_norm": 0.5854681730270386,
      "learning_rate": 4.102564102564103e-06,
      "loss": 1.7968,
      "step": 81
    },
    {
      "epoch": 0.012647245946519116,
      "grad_norm": 0.5472329258918762,
      "learning_rate": 4.1538461538461545e-06,
      "loss": 1.7854,
      "step": 82
    },
    {
      "epoch": 0.012801480653183983,
      "grad_norm": 0.5131881833076477,
      "learning_rate": 4.2051282051282055e-06,
      "loss": 1.7551,
      "step": 83
    },
    {
      "epoch": 0.01295571535984885,
      "grad_norm": 0.5052653551101685,
      "learning_rate": 4.2564102564102566e-06,
      "loss": 1.7032,
      "step": 84
    },
    {
      "epoch": 0.013109950066513717,
      "grad_norm": 0.5798134207725525,
      "learning_rate": 4.307692307692308e-06,
      "loss": 1.7943,
      "step": 85
    },
    {
      "epoch": 0.013264184773178584,
      "grad_norm": 0.5180870294570923,
      "learning_rate": 4.358974358974359e-06,
      "loss": 1.687,
      "step": 86
    },
    {
      "epoch": 0.013418419479843452,
      "grad_norm": 0.49474793672561646,
      "learning_rate": 4.4102564102564104e-06,
      "loss": 1.7606,
      "step": 87
    },
    {
      "epoch": 0.01357265418650832,
      "grad_norm": 0.572761595249176,
      "learning_rate": 4.461538461538462e-06,
      "loss": 1.8037,
      "step": 88
    },
    {
      "epoch": 0.013726888893173186,
      "grad_norm": 0.5462682247161865,
      "learning_rate": 4.512820512820513e-06,
      "loss": 1.7548,
      "step": 89
    },
    {
      "epoch": 0.013881123599838053,
      "grad_norm": 0.5709042549133301,
      "learning_rate": 4.564102564102564e-06,
      "loss": 1.7579,
      "step": 90
    },
    {
      "epoch": 0.014035358306502922,
      "grad_norm": 0.553099513053894,
      "learning_rate": 4.615384615384616e-06,
      "loss": 1.7568,
      "step": 91
    },
    {
      "epoch": 0.014189593013167788,
      "grad_norm": 0.5180365443229675,
      "learning_rate": 4.666666666666667e-06,
      "loss": 1.7567,
      "step": 92
    },
    {
      "epoch": 0.014343827719832655,
      "grad_norm": 0.5820419192314148,
      "learning_rate": 4.717948717948718e-06,
      "loss": 1.7418,
      "step": 93
    },
    {
      "epoch": 0.014498062426497522,
      "grad_norm": 0.5648301243782043,
      "learning_rate": 4.76923076923077e-06,
      "loss": 1.6861,
      "step": 94
    },
    {
      "epoch": 0.01465229713316239,
      "grad_norm": 0.49814391136169434,
      "learning_rate": 4.820512820512821e-06,
      "loss": 1.7415,
      "step": 95
    },
    {
      "epoch": 0.014806531839827258,
      "grad_norm": 0.5138998031616211,
      "learning_rate": 4.871794871794872e-06,
      "loss": 1.7127,
      "step": 96
    },
    {
      "epoch": 0.014960766546492124,
      "grad_norm": 0.6080782413482666,
      "learning_rate": 4.923076923076924e-06,
      "loss": 1.7281,
      "step": 97
    },
    {
      "epoch": 0.015115001253156991,
      "grad_norm": 0.5199087858200073,
      "learning_rate": 4.974358974358975e-06,
      "loss": 1.7136,
      "step": 98
    },
    {
      "epoch": 0.015269235959821858,
      "grad_norm": 0.5357125997543335,
      "learning_rate": 5.025641025641026e-06,
      "loss": 1.7344,
      "step": 99
    },
    {
      "epoch": 0.015423470666486727,
      "grad_norm": 0.547616183757782,
      "learning_rate": 5.076923076923077e-06,
      "loss": 1.7265,
      "step": 100
    },
    {
      "epoch": 0.015577705373151594,
      "grad_norm": 0.5598102807998657,
      "learning_rate": 5.128205128205128e-06,
      "loss": 1.7278,
      "step": 101
    },
    {
      "epoch": 0.015731940079816462,
      "grad_norm": 0.534231960773468,
      "learning_rate": 5.179487179487181e-06,
      "loss": 1.6925,
      "step": 102
    },
    {
      "epoch": 0.015886174786481327,
      "grad_norm": 0.5571257472038269,
      "learning_rate": 5.230769230769232e-06,
      "loss": 1.6954,
      "step": 103
    },
    {
      "epoch": 0.016040409493146196,
      "grad_norm": 0.5430976152420044,
      "learning_rate": 5.282051282051283e-06,
      "loss": 1.7232,
      "step": 104
    },
    {
      "epoch": 0.01619464419981106,
      "grad_norm": 0.5642494559288025,
      "learning_rate": 5.333333333333334e-06,
      "loss": 1.688,
      "step": 105
    },
    {
      "epoch": 0.01634887890647593,
      "grad_norm": 0.5086437463760376,
      "learning_rate": 5.384615384615385e-06,
      "loss": 1.6922,
      "step": 106
    },
    {
      "epoch": 0.0165031136131408,
      "grad_norm": 0.6055141687393188,
      "learning_rate": 5.435897435897436e-06,
      "loss": 1.7557,
      "step": 107
    },
    {
      "epoch": 0.016657348319805663,
      "grad_norm": 0.5747716426849365,
      "learning_rate": 5.487179487179488e-06,
      "loss": 1.6945,
      "step": 108
    },
    {
      "epoch": 0.016811583026470532,
      "grad_norm": 0.5839638113975525,
      "learning_rate": 5.538461538461539e-06,
      "loss": 1.6987,
      "step": 109
    },
    {
      "epoch": 0.016965817733135397,
      "grad_norm": 0.6069226861000061,
      "learning_rate": 5.58974358974359e-06,
      "loss": 1.6843,
      "step": 110
    },
    {
      "epoch": 0.017120052439800266,
      "grad_norm": 0.5975008010864258,
      "learning_rate": 5.641025641025641e-06,
      "loss": 1.6918,
      "step": 111
    },
    {
      "epoch": 0.017274287146465134,
      "grad_norm": 0.5783462524414062,
      "learning_rate": 5.692307692307692e-06,
      "loss": 1.6752,
      "step": 112
    },
    {
      "epoch": 0.01742852185313,
      "grad_norm": 0.5778764486312866,
      "learning_rate": 5.743589743589743e-06,
      "loss": 1.7039,
      "step": 113
    },
    {
      "epoch": 0.017582756559794868,
      "grad_norm": 0.5791458487510681,
      "learning_rate": 5.794871794871796e-06,
      "loss": 1.6676,
      "step": 114
    },
    {
      "epoch": 0.017736991266459737,
      "grad_norm": 0.577421247959137,
      "learning_rate": 5.846153846153847e-06,
      "loss": 1.6754,
      "step": 115
    },
    {
      "epoch": 0.017891225973124602,
      "grad_norm": 0.6224790811538696,
      "learning_rate": 5.897435897435898e-06,
      "loss": 1.6874,
      "step": 116
    },
    {
      "epoch": 0.01804546067978947,
      "grad_norm": 0.5089656114578247,
      "learning_rate": 5.948717948717949e-06,
      "loss": 1.6383,
      "step": 117
    },
    {
      "epoch": 0.018199695386454336,
      "grad_norm": 0.5308560729026794,
      "learning_rate": 6e-06,
      "loss": 1.6398,
      "step": 118
    },
    {
      "epoch": 0.018353930093119204,
      "grad_norm": 0.630492627620697,
      "learning_rate": 6.051282051282051e-06,
      "loss": 1.6644,
      "step": 119
    },
    {
      "epoch": 0.018508164799784073,
      "grad_norm": 0.5221322774887085,
      "learning_rate": 6.102564102564104e-06,
      "loss": 1.6541,
      "step": 120
    },
    {
      "epoch": 0.018662399506448938,
      "grad_norm": 0.5628217458724976,
      "learning_rate": 6.153846153846155e-06,
      "loss": 1.596,
      "step": 121
    },
    {
      "epoch": 0.018816634213113807,
      "grad_norm": 0.5598528385162354,
      "learning_rate": 6.205128205128206e-06,
      "loss": 1.6402,
      "step": 122
    },
    {
      "epoch": 0.018970868919778672,
      "grad_norm": 0.5308383107185364,
      "learning_rate": 6.256410256410257e-06,
      "loss": 1.6523,
      "step": 123
    },
    {
      "epoch": 0.01912510362644354,
      "grad_norm": 0.5692138075828552,
      "learning_rate": 6.307692307692308e-06,
      "loss": 1.6359,
      "step": 124
    },
    {
      "epoch": 0.01927933833310841,
      "grad_norm": 0.6486672163009644,
      "learning_rate": 6.358974358974359e-06,
      "loss": 1.6528,
      "step": 125
    },
    {
      "epoch": 0.019433573039773274,
      "grad_norm": 0.5242540240287781,
      "learning_rate": 6.410256410256412e-06,
      "loss": 1.6354,
      "step": 126
    },
    {
      "epoch": 0.019587807746438143,
      "grad_norm": 0.5515648722648621,
      "learning_rate": 6.461538461538463e-06,
      "loss": 1.6188,
      "step": 127
    },
    {
      "epoch": 0.019742042453103008,
      "grad_norm": 0.5765663981437683,
      "learning_rate": 6.512820512820514e-06,
      "loss": 1.6153,
      "step": 128
    },
    {
      "epoch": 0.019896277159767876,
      "grad_norm": 0.5636539459228516,
      "learning_rate": 6.564102564102565e-06,
      "loss": 1.6604,
      "step": 129
    },
    {
      "epoch": 0.020050511866432745,
      "grad_norm": 0.576707661151886,
      "learning_rate": 6.615384615384616e-06,
      "loss": 1.599,
      "step": 130
    },
    {
      "epoch": 0.02020474657309761,
      "grad_norm": 0.544995129108429,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.5985,
      "step": 131
    },
    {
      "epoch": 0.02035898127976248,
      "grad_norm": 0.545677125453949,
      "learning_rate": 6.717948717948718e-06,
      "loss": 1.6248,
      "step": 132
    },
    {
      "epoch": 0.020513215986427347,
      "grad_norm": 0.5226114988327026,
      "learning_rate": 6.76923076923077e-06,
      "loss": 1.6214,
      "step": 133
    },
    {
      "epoch": 0.020667450693092212,
      "grad_norm": 0.5579204559326172,
      "learning_rate": 6.820512820512821e-06,
      "loss": 1.5938,
      "step": 134
    },
    {
      "epoch": 0.02082168539975708,
      "grad_norm": 0.5821995735168457,
      "learning_rate": 6.871794871794872e-06,
      "loss": 1.6156,
      "step": 135
    },
    {
      "epoch": 0.020975920106421946,
      "grad_norm": 0.5184845924377441,
      "learning_rate": 6.923076923076923e-06,
      "loss": 1.602,
      "step": 136
    },
    {
      "epoch": 0.021130154813086815,
      "grad_norm": 0.5434860587120056,
      "learning_rate": 6.974358974358974e-06,
      "loss": 1.5739,
      "step": 137
    },
    {
      "epoch": 0.021284389519751683,
      "grad_norm": 0.5877484083175659,
      "learning_rate": 7.025641025641025e-06,
      "loss": 1.6584,
      "step": 138
    },
    {
      "epoch": 0.02143862422641655,
      "grad_norm": 0.5218464732170105,
      "learning_rate": 7.076923076923078e-06,
      "loss": 1.5989,
      "step": 139
    },
    {
      "epoch": 0.021592858933081417,
      "grad_norm": 0.5589679479598999,
      "learning_rate": 7.128205128205129e-06,
      "loss": 1.5849,
      "step": 140
    },
    {
      "epoch": 0.021747093639746282,
      "grad_norm": 0.5412963628768921,
      "learning_rate": 7.17948717948718e-06,
      "loss": 1.6026,
      "step": 141
    },
    {
      "epoch": 0.02190132834641115,
      "grad_norm": 0.5312784910202026,
      "learning_rate": 7.230769230769231e-06,
      "loss": 1.6181,
      "step": 142
    },
    {
      "epoch": 0.02205556305307602,
      "grad_norm": 0.5265107750892639,
      "learning_rate": 7.282051282051282e-06,
      "loss": 1.6133,
      "step": 143
    },
    {
      "epoch": 0.022209797759740885,
      "grad_norm": 0.61285799741745,
      "learning_rate": 7.333333333333333e-06,
      "loss": 1.5933,
      "step": 144
    },
    {
      "epoch": 0.022364032466405753,
      "grad_norm": 0.5416054725646973,
      "learning_rate": 7.384615384615386e-06,
      "loss": 1.5919,
      "step": 145
    },
    {
      "epoch": 0.022518267173070622,
      "grad_norm": 0.5560144782066345,
      "learning_rate": 7.435897435897437e-06,
      "loss": 1.5417,
      "step": 146
    },
    {
      "epoch": 0.022672501879735487,
      "grad_norm": 0.5515575408935547,
      "learning_rate": 7.487179487179488e-06,
      "loss": 1.6038,
      "step": 147
    },
    {
      "epoch": 0.022826736586400356,
      "grad_norm": 0.5019623041152954,
      "learning_rate": 7.538461538461539e-06,
      "loss": 1.5825,
      "step": 148
    },
    {
      "epoch": 0.02298097129306522,
      "grad_norm": 0.4966740310192108,
      "learning_rate": 7.58974358974359e-06,
      "loss": 1.5658,
      "step": 149
    },
    {
      "epoch": 0.02313520599973009,
      "grad_norm": 0.4935978353023529,
      "learning_rate": 7.641025641025641e-06,
      "loss": 1.5489,
      "step": 150
    },
    {
      "epoch": 0.023289440706394958,
      "grad_norm": 0.5220585465431213,
      "learning_rate": 7.692307692307694e-06,
      "loss": 1.5868,
      "step": 151
    },
    {
      "epoch": 0.023443675413059823,
      "grad_norm": 0.5111873745918274,
      "learning_rate": 7.743589743589745e-06,
      "loss": 1.5168,
      "step": 152
    },
    {
      "epoch": 0.02359791011972469,
      "grad_norm": 0.5328781604766846,
      "learning_rate": 7.794871794871796e-06,
      "loss": 1.5689,
      "step": 153
    },
    {
      "epoch": 0.023752144826389557,
      "grad_norm": 0.5277527570724487,
      "learning_rate": 7.846153846153847e-06,
      "loss": 1.5745,
      "step": 154
    },
    {
      "epoch": 0.023906379533054425,
      "grad_norm": 0.47298046946525574,
      "learning_rate": 7.897435897435898e-06,
      "loss": 1.5584,
      "step": 155
    },
    {
      "epoch": 0.024060614239719294,
      "grad_norm": 0.5444458723068237,
      "learning_rate": 7.948717948717949e-06,
      "loss": 1.5624,
      "step": 156
    },
    {
      "epoch": 0.02421484894638416,
      "grad_norm": 0.5083262920379639,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.6258,
      "step": 157
    },
    {
      "epoch": 0.024369083653049028,
      "grad_norm": 0.5201572179794312,
      "learning_rate": 8.051282051282052e-06,
      "loss": 1.6051,
      "step": 158
    },
    {
      "epoch": 0.024523318359713893,
      "grad_norm": 0.5222218632698059,
      "learning_rate": 8.102564102564103e-06,
      "loss": 1.5448,
      "step": 159
    },
    {
      "epoch": 0.02467755306637876,
      "grad_norm": 0.4937100410461426,
      "learning_rate": 8.153846153846154e-06,
      "loss": 1.5736,
      "step": 160
    },
    {
      "epoch": 0.02483178777304363,
      "grad_norm": 0.511566162109375,
      "learning_rate": 8.205128205128205e-06,
      "loss": 1.5593,
      "step": 161
    },
    {
      "epoch": 0.024986022479708495,
      "grad_norm": 0.5414931178092957,
      "learning_rate": 8.256410256410256e-06,
      "loss": 1.5372,
      "step": 162
    },
    {
      "epoch": 0.025140257186373364,
      "grad_norm": 0.5102213025093079,
      "learning_rate": 8.307692307692309e-06,
      "loss": 1.5652,
      "step": 163
    },
    {
      "epoch": 0.025294491893038232,
      "grad_norm": 0.49184349179267883,
      "learning_rate": 8.35897435897436e-06,
      "loss": 1.5025,
      "step": 164
    },
    {
      "epoch": 0.025448726599703098,
      "grad_norm": 0.5266595482826233,
      "learning_rate": 8.410256410256411e-06,
      "loss": 1.5522,
      "step": 165
    },
    {
      "epoch": 0.025602961306367966,
      "grad_norm": 0.5281389951705933,
      "learning_rate": 8.461538461538462e-06,
      "loss": 1.5447,
      "step": 166
    },
    {
      "epoch": 0.02575719601303283,
      "grad_norm": 0.5073181390762329,
      "learning_rate": 8.512820512820513e-06,
      "loss": 1.5365,
      "step": 167
    },
    {
      "epoch": 0.0259114307196977,
      "grad_norm": 0.5095680356025696,
      "learning_rate": 8.564102564102564e-06,
      "loss": 1.543,
      "step": 168
    },
    {
      "epoch": 0.02606566542636257,
      "grad_norm": 0.5605200529098511,
      "learning_rate": 8.615384615384617e-06,
      "loss": 1.5694,
      "step": 169
    },
    {
      "epoch": 0.026219900133027434,
      "grad_norm": 0.4923345744609833,
      "learning_rate": 8.666666666666668e-06,
      "loss": 1.5625,
      "step": 170
    },
    {
      "epoch": 0.026374134839692302,
      "grad_norm": 0.520775318145752,
      "learning_rate": 8.717948717948719e-06,
      "loss": 1.549,
      "step": 171
    },
    {
      "epoch": 0.026528369546357167,
      "grad_norm": 0.5525628924369812,
      "learning_rate": 8.76923076923077e-06,
      "loss": 1.5293,
      "step": 172
    },
    {
      "epoch": 0.026682604253022036,
      "grad_norm": 0.4697412848472595,
      "learning_rate": 8.820512820512821e-06,
      "loss": 1.5181,
      "step": 173
    },
    {
      "epoch": 0.026836838959686905,
      "grad_norm": 0.5007003545761108,
      "learning_rate": 8.871794871794872e-06,
      "loss": 1.5086,
      "step": 174
    },
    {
      "epoch": 0.02699107366635177,
      "grad_norm": 0.544914960861206,
      "learning_rate": 8.923076923076925e-06,
      "loss": 1.4978,
      "step": 175
    },
    {
      "epoch": 0.02714530837301664,
      "grad_norm": 0.5071407556533813,
      "learning_rate": 8.974358974358976e-06,
      "loss": 1.5403,
      "step": 176
    },
    {
      "epoch": 0.027299543079681507,
      "grad_norm": 0.49474087357521057,
      "learning_rate": 9.025641025641027e-06,
      "loss": 1.5416,
      "step": 177
    },
    {
      "epoch": 0.027453777786346372,
      "grad_norm": 0.5103272199630737,
      "learning_rate": 9.076923076923078e-06,
      "loss": 1.5159,
      "step": 178
    },
    {
      "epoch": 0.02760801249301124,
      "grad_norm": 0.5085631608963013,
      "learning_rate": 9.128205128205129e-06,
      "loss": 1.5361,
      "step": 179
    },
    {
      "epoch": 0.027762247199676106,
      "grad_norm": 0.5117743015289307,
      "learning_rate": 9.17948717948718e-06,
      "loss": 1.5114,
      "step": 180
    },
    {
      "epoch": 0.027916481906340974,
      "grad_norm": 0.5602802038192749,
      "learning_rate": 9.230769230769232e-06,
      "loss": 1.5534,
      "step": 181
    },
    {
      "epoch": 0.028070716613005843,
      "grad_norm": 0.509979784488678,
      "learning_rate": 9.282051282051283e-06,
      "loss": 1.5526,
      "step": 182
    },
    {
      "epoch": 0.028224951319670708,
      "grad_norm": 0.4802814722061157,
      "learning_rate": 9.333333333333334e-06,
      "loss": 1.5043,
      "step": 183
    },
    {
      "epoch": 0.028379186026335577,
      "grad_norm": 0.511145293712616,
      "learning_rate": 9.384615384615385e-06,
      "loss": 1.5254,
      "step": 184
    },
    {
      "epoch": 0.028533420733000442,
      "grad_norm": 0.5523600578308105,
      "learning_rate": 9.435897435897436e-06,
      "loss": 1.5476,
      "step": 185
    },
    {
      "epoch": 0.02868765543966531,
      "grad_norm": 0.5209853649139404,
      "learning_rate": 9.487179487179487e-06,
      "loss": 1.5132,
      "step": 186
    },
    {
      "epoch": 0.02884189014633018,
      "grad_norm": 0.5149086117744446,
      "learning_rate": 9.53846153846154e-06,
      "loss": 1.5123,
      "step": 187
    },
    {
      "epoch": 0.028996124852995044,
      "grad_norm": 0.5762576460838318,
      "learning_rate": 9.589743589743591e-06,
      "loss": 1.5759,
      "step": 188
    },
    {
      "epoch": 0.029150359559659913,
      "grad_norm": 0.47409066557884216,
      "learning_rate": 9.641025641025642e-06,
      "loss": 1.5326,
      "step": 189
    },
    {
      "epoch": 0.02930459426632478,
      "grad_norm": 0.466664582490921,
      "learning_rate": 9.692307692307693e-06,
      "loss": 1.5393,
      "step": 190
    },
    {
      "epoch": 0.029458828972989647,
      "grad_norm": 0.5279547572135925,
      "learning_rate": 9.743589743589744e-06,
      "loss": 1.5227,
      "step": 191
    },
    {
      "epoch": 0.029613063679654515,
      "grad_norm": 0.486102819442749,
      "learning_rate": 9.794871794871795e-06,
      "loss": 1.5195,
      "step": 192
    },
    {
      "epoch": 0.02976729838631938,
      "grad_norm": 0.4880712926387787,
      "learning_rate": 9.846153846153848e-06,
      "loss": 1.4996,
      "step": 193
    },
    {
      "epoch": 0.02992153309298425,
      "grad_norm": 0.5420333743095398,
      "learning_rate": 9.897435897435899e-06,
      "loss": 1.5643,
      "step": 194
    },
    {
      "epoch": 0.030075767799649118,
      "grad_norm": 0.48395177721977234,
      "learning_rate": 9.94871794871795e-06,
      "loss": 1.5478,
      "step": 195
    },
    {
      "epoch": 0.030230002506313983,
      "grad_norm": 0.5014393329620361,
      "learning_rate": 1e-05,
      "loss": 1.4718,
      "step": 196
    },
    {
      "epoch": 0.03038423721297885,
      "grad_norm": 0.4976564645767212,
      "learning_rate": 9.999999376155205e-06,
      "loss": 1.5321,
      "step": 197
    },
    {
      "epoch": 0.030538471919643716,
      "grad_norm": 0.46849575638771057,
      "learning_rate": 9.999997504620979e-06,
      "loss": 1.5309,
      "step": 198
    },
    {
      "epoch": 0.030692706626308585,
      "grad_norm": 0.5039035081863403,
      "learning_rate": 9.999994385397785e-06,
      "loss": 1.4752,
      "step": 199
    },
    {
      "epoch": 0.030846941332973454,
      "grad_norm": 0.49510613083839417,
      "learning_rate": 9.999990018486403e-06,
      "loss": 1.4921,
      "step": 200
    },
    {
      "epoch": 0.03100117603963832,
      "grad_norm": 0.49737676978111267,
      "learning_rate": 9.999984403887922e-06,
      "loss": 1.501,
      "step": 201
    },
    {
      "epoch": 0.031155410746303187,
      "grad_norm": 0.4791927933692932,
      "learning_rate": 9.999977541603745e-06,
      "loss": 1.4581,
      "step": 202
    },
    {
      "epoch": 0.031309645452968056,
      "grad_norm": 0.46148934960365295,
      "learning_rate": 9.999969431635584e-06,
      "loss": 1.5367,
      "step": 203
    },
    {
      "epoch": 0.031463880159632925,
      "grad_norm": 0.5311307907104492,
      "learning_rate": 9.99996007398546e-06,
      "loss": 1.5426,
      "step": 204
    },
    {
      "epoch": 0.031618114866297786,
      "grad_norm": 0.46260470151901245,
      "learning_rate": 9.999949468655711e-06,
      "loss": 1.5089,
      "step": 205
    },
    {
      "epoch": 0.031772349572962655,
      "grad_norm": 0.46585148572921753,
      "learning_rate": 9.999937615648983e-06,
      "loss": 1.5111,
      "step": 206
    },
    {
      "epoch": 0.03192658427962752,
      "grad_norm": 0.528596043586731,
      "learning_rate": 9.999924514968234e-06,
      "loss": 1.5383,
      "step": 207
    },
    {
      "epoch": 0.03208081898629239,
      "grad_norm": 0.47473981976509094,
      "learning_rate": 9.99991016661673e-06,
      "loss": 1.4716,
      "step": 208
    },
    {
      "epoch": 0.03223505369295726,
      "grad_norm": 0.45880991220474243,
      "learning_rate": 9.999894570598055e-06,
      "loss": 1.49,
      "step": 209
    },
    {
      "epoch": 0.03238928839962212,
      "grad_norm": 0.5201250910758972,
      "learning_rate": 9.9998777269161e-06,
      "loss": 1.4918,
      "step": 210
    },
    {
      "epoch": 0.03254352310628699,
      "grad_norm": 0.492723673582077,
      "learning_rate": 9.999859635575068e-06,
      "loss": 1.5056,
      "step": 211
    },
    {
      "epoch": 0.03269775781295186,
      "grad_norm": 0.45561763644218445,
      "learning_rate": 9.999840296579472e-06,
      "loss": 1.4839,
      "step": 212
    },
    {
      "epoch": 0.03285199251961673,
      "grad_norm": 0.5062990784645081,
      "learning_rate": 9.999819709934142e-06,
      "loss": 1.5299,
      "step": 213
    },
    {
      "epoch": 0.0330062272262816,
      "grad_norm": 0.47736623883247375,
      "learning_rate": 9.99979787564421e-06,
      "loss": 1.4988,
      "step": 214
    },
    {
      "epoch": 0.03316046193294646,
      "grad_norm": 0.4908311069011688,
      "learning_rate": 9.999774793715126e-06,
      "loss": 1.5076,
      "step": 215
    },
    {
      "epoch": 0.03331469663961133,
      "grad_norm": 0.5111482739448547,
      "learning_rate": 9.99975046415265e-06,
      "loss": 1.5338,
      "step": 216
    },
    {
      "epoch": 0.033468931346276196,
      "grad_norm": 0.4972241520881653,
      "learning_rate": 9.999724886962857e-06,
      "loss": 1.506,
      "step": 217
    },
    {
      "epoch": 0.033623166052941064,
      "grad_norm": 0.5056758522987366,
      "learning_rate": 9.999698062152125e-06,
      "loss": 1.4763,
      "step": 218
    },
    {
      "epoch": 0.03377740075960593,
      "grad_norm": 0.5542264580726624,
      "learning_rate": 9.999669989727147e-06,
      "loss": 1.5122,
      "step": 219
    },
    {
      "epoch": 0.033931635466270794,
      "grad_norm": 0.4869178831577301,
      "learning_rate": 9.999640669694932e-06,
      "loss": 1.4991,
      "step": 220
    },
    {
      "epoch": 0.03408587017293566,
      "grad_norm": 0.5192879438400269,
      "learning_rate": 9.999610102062795e-06,
      "loss": 1.5102,
      "step": 221
    },
    {
      "epoch": 0.03424010487960053,
      "grad_norm": 0.5683949589729309,
      "learning_rate": 9.999578286838363e-06,
      "loss": 1.5163,
      "step": 222
    },
    {
      "epoch": 0.0343943395862654,
      "grad_norm": 0.5145759582519531,
      "learning_rate": 9.999545224029574e-06,
      "loss": 1.4913,
      "step": 223
    },
    {
      "epoch": 0.03454857429293027,
      "grad_norm": 0.4716167747974396,
      "learning_rate": 9.99951091364468e-06,
      "loss": 1.4894,
      "step": 224
    },
    {
      "epoch": 0.03470280899959513,
      "grad_norm": 0.5503811836242676,
      "learning_rate": 9.999475355692245e-06,
      "loss": 1.5157,
      "step": 225
    },
    {
      "epoch": 0.03485704370626,
      "grad_norm": 0.4690919816493988,
      "learning_rate": 9.999438550181137e-06,
      "loss": 1.5261,
      "step": 226
    },
    {
      "epoch": 0.03501127841292487,
      "grad_norm": 0.4738644063472748,
      "learning_rate": 9.999400497120545e-06,
      "loss": 1.4896,
      "step": 227
    },
    {
      "epoch": 0.035165513119589736,
      "grad_norm": 0.5106549263000488,
      "learning_rate": 9.999361196519961e-06,
      "loss": 1.4862,
      "step": 228
    },
    {
      "epoch": 0.035319747826254605,
      "grad_norm": 0.46504995226860046,
      "learning_rate": 9.999320648389195e-06,
      "loss": 1.5248,
      "step": 229
    },
    {
      "epoch": 0.035473982532919474,
      "grad_norm": 0.4892650246620178,
      "learning_rate": 9.999278852738364e-06,
      "loss": 1.4917,
      "step": 230
    },
    {
      "epoch": 0.035628217239584335,
      "grad_norm": 0.4626985192298889,
      "learning_rate": 9.999235809577897e-06,
      "loss": 1.489,
      "step": 231
    },
    {
      "epoch": 0.035782451946249204,
      "grad_norm": 0.552362859249115,
      "learning_rate": 9.999191518918536e-06,
      "loss": 1.5348,
      "step": 232
    },
    {
      "epoch": 0.03593668665291407,
      "grad_norm": 0.4743778109550476,
      "learning_rate": 9.999145980771334e-06,
      "loss": 1.4837,
      "step": 233
    },
    {
      "epoch": 0.03609092135957894,
      "grad_norm": 0.47292467951774597,
      "learning_rate": 9.99909919514765e-06,
      "loss": 1.489,
      "step": 234
    },
    {
      "epoch": 0.03624515606624381,
      "grad_norm": 0.4847182333469391,
      "learning_rate": 9.999051162059165e-06,
      "loss": 1.5122,
      "step": 235
    },
    {
      "epoch": 0.03639939077290867,
      "grad_norm": 0.45157289505004883,
      "learning_rate": 9.99900188151786e-06,
      "loss": 1.4347,
      "step": 236
    },
    {
      "epoch": 0.03655362547957354,
      "grad_norm": 0.4606632590293884,
      "learning_rate": 9.998951353536036e-06,
      "loss": 1.4722,
      "step": 237
    },
    {
      "epoch": 0.03670786018623841,
      "grad_norm": 0.5129456520080566,
      "learning_rate": 9.9988995781263e-06,
      "loss": 1.506,
      "step": 238
    },
    {
      "epoch": 0.03686209489290328,
      "grad_norm": 0.49908360838890076,
      "learning_rate": 9.998846555301572e-06,
      "loss": 1.4697,
      "step": 239
    },
    {
      "epoch": 0.037016329599568146,
      "grad_norm": 0.47629302740097046,
      "learning_rate": 9.99879228507508e-06,
      "loss": 1.4859,
      "step": 240
    },
    {
      "epoch": 0.03717056430623301,
      "grad_norm": 0.55729079246521,
      "learning_rate": 9.998736767460373e-06,
      "loss": 1.505,
      "step": 241
    },
    {
      "epoch": 0.037324799012897876,
      "grad_norm": 0.4755311906337738,
      "learning_rate": 9.998680002471302e-06,
      "loss": 1.4776,
      "step": 242
    },
    {
      "epoch": 0.037479033719562745,
      "grad_norm": 0.48049280047416687,
      "learning_rate": 9.99862199012203e-06,
      "loss": 1.4838,
      "step": 243
    },
    {
      "epoch": 0.03763326842622761,
      "grad_norm": 0.5337971448898315,
      "learning_rate": 9.998562730427035e-06,
      "loss": 1.5136,
      "step": 244
    },
    {
      "epoch": 0.03778750313289248,
      "grad_norm": 0.4617859721183777,
      "learning_rate": 9.998502223401104e-06,
      "loss": 1.4989,
      "step": 245
    },
    {
      "epoch": 0.037941737839557343,
      "grad_norm": 0.4700513780117035,
      "learning_rate": 9.998440469059336e-06,
      "loss": 1.4656,
      "step": 246
    },
    {
      "epoch": 0.03809597254622221,
      "grad_norm": 0.5715804696083069,
      "learning_rate": 9.998377467417142e-06,
      "loss": 1.5091,
      "step": 247
    },
    {
      "epoch": 0.03825020725288708,
      "grad_norm": 0.4865875244140625,
      "learning_rate": 9.998313218490244e-06,
      "loss": 1.5167,
      "step": 248
    },
    {
      "epoch": 0.03840444195955195,
      "grad_norm": 0.4848752021789551,
      "learning_rate": 9.99824772229467e-06,
      "loss": 1.5317,
      "step": 249
    },
    {
      "epoch": 0.03855867666621682,
      "grad_norm": 0.46630924940109253,
      "learning_rate": 9.998180978846768e-06,
      "loss": 1.486,
      "step": 250
    },
    {
      "epoch": 0.03871291137288168,
      "grad_norm": 0.4797709584236145,
      "learning_rate": 9.998112988163191e-06,
      "loss": 1.4303,
      "step": 251
    },
    {
      "epoch": 0.03886714607954655,
      "grad_norm": 0.5287182331085205,
      "learning_rate": 9.998043750260906e-06,
      "loss": 1.4937,
      "step": 252
    },
    {
      "epoch": 0.03902138078621142,
      "grad_norm": 0.4925422966480255,
      "learning_rate": 9.997973265157192e-06,
      "loss": 1.4897,
      "step": 253
    },
    {
      "epoch": 0.039175615492876285,
      "grad_norm": 0.5214465856552124,
      "learning_rate": 9.997901532869636e-06,
      "loss": 1.4669,
      "step": 254
    },
    {
      "epoch": 0.039329850199541154,
      "grad_norm": 0.48910996317863464,
      "learning_rate": 9.997828553416136e-06,
      "loss": 1.4673,
      "step": 255
    },
    {
      "epoch": 0.039484084906206016,
      "grad_norm": 0.5112558603286743,
      "learning_rate": 9.997754326814908e-06,
      "loss": 1.476,
      "step": 256
    },
    {
      "epoch": 0.039638319612870884,
      "grad_norm": 0.5078018307685852,
      "learning_rate": 9.99767885308447e-06,
      "loss": 1.4815,
      "step": 257
    },
    {
      "epoch": 0.03979255431953575,
      "grad_norm": 0.5006856322288513,
      "learning_rate": 9.997602132243659e-06,
      "loss": 1.4811,
      "step": 258
    },
    {
      "epoch": 0.03994678902620062,
      "grad_norm": 0.4795530140399933,
      "learning_rate": 9.997524164311616e-06,
      "loss": 1.4706,
      "step": 259
    },
    {
      "epoch": 0.04010102373286549,
      "grad_norm": 0.5512394309043884,
      "learning_rate": 9.9974449493078e-06,
      "loss": 1.5279,
      "step": 260
    },
    {
      "epoch": 0.04025525843953036,
      "grad_norm": 0.5088499784469604,
      "learning_rate": 9.997364487251976e-06,
      "loss": 1.4688,
      "step": 261
    },
    {
      "epoch": 0.04040949314619522,
      "grad_norm": 0.4722222089767456,
      "learning_rate": 9.997282778164223e-06,
      "loss": 1.4721,
      "step": 262
    },
    {
      "epoch": 0.04056372785286009,
      "grad_norm": 0.5502070188522339,
      "learning_rate": 9.997199822064934e-06,
      "loss": 1.4868,
      "step": 263
    },
    {
      "epoch": 0.04071796255952496,
      "grad_norm": 0.4725727140903473,
      "learning_rate": 9.997115618974804e-06,
      "loss": 1.4851,
      "step": 264
    },
    {
      "epoch": 0.040872197266189826,
      "grad_norm": 0.5010146498680115,
      "learning_rate": 9.997030168914848e-06,
      "loss": 1.4376,
      "step": 265
    },
    {
      "epoch": 0.041026431972854695,
      "grad_norm": 0.5068126320838928,
      "learning_rate": 9.996943471906389e-06,
      "loss": 1.4676,
      "step": 266
    },
    {
      "epoch": 0.041180666679519556,
      "grad_norm": 0.47465837001800537,
      "learning_rate": 9.996855527971058e-06,
      "loss": 1.4522,
      "step": 267
    },
    {
      "epoch": 0.041334901386184425,
      "grad_norm": 0.4365485906600952,
      "learning_rate": 9.996766337130806e-06,
      "loss": 1.4403,
      "step": 268
    },
    {
      "epoch": 0.041489136092849294,
      "grad_norm": 0.5462592840194702,
      "learning_rate": 9.996675899407886e-06,
      "loss": 1.5112,
      "step": 269
    },
    {
      "epoch": 0.04164337079951416,
      "grad_norm": 0.4605362117290497,
      "learning_rate": 9.996584214824867e-06,
      "loss": 1.4659,
      "step": 270
    },
    {
      "epoch": 0.04179760550617903,
      "grad_norm": 0.46796712279319763,
      "learning_rate": 9.996491283404626e-06,
      "loss": 1.4506,
      "step": 271
    },
    {
      "epoch": 0.04195184021284389,
      "grad_norm": 0.5163938403129578,
      "learning_rate": 9.996397105170353e-06,
      "loss": 1.4661,
      "step": 272
    },
    {
      "epoch": 0.04210607491950876,
      "grad_norm": 0.5185596942901611,
      "learning_rate": 9.996301680145551e-06,
      "loss": 1.4609,
      "step": 273
    },
    {
      "epoch": 0.04226030962617363,
      "grad_norm": 0.48026686906814575,
      "learning_rate": 9.996205008354031e-06,
      "loss": 1.4885,
      "step": 274
    },
    {
      "epoch": 0.0424145443328385,
      "grad_norm": 0.5034846663475037,
      "learning_rate": 9.996107089819916e-06,
      "loss": 1.4446,
      "step": 275
    },
    {
      "epoch": 0.04256877903950337,
      "grad_norm": 0.5184863805770874,
      "learning_rate": 9.996007924567641e-06,
      "loss": 1.4707,
      "step": 276
    },
    {
      "epoch": 0.04272301374616823,
      "grad_norm": 0.4674989879131317,
      "learning_rate": 9.995907512621952e-06,
      "loss": 1.4632,
      "step": 277
    },
    {
      "epoch": 0.0428772484528331,
      "grad_norm": 0.4774078130722046,
      "learning_rate": 9.995805854007903e-06,
      "loss": 1.4849,
      "step": 278
    },
    {
      "epoch": 0.043031483159497966,
      "grad_norm": 0.4731788635253906,
      "learning_rate": 9.995702948750864e-06,
      "loss": 1.4733,
      "step": 279
    },
    {
      "epoch": 0.043185717866162834,
      "grad_norm": 0.4455285370349884,
      "learning_rate": 9.995598796876514e-06,
      "loss": 1.4507,
      "step": 280
    },
    {
      "epoch": 0.0433399525728277,
      "grad_norm": 0.46435633301734924,
      "learning_rate": 9.995493398410843e-06,
      "loss": 1.4662,
      "step": 281
    },
    {
      "epoch": 0.043494187279492565,
      "grad_norm": 0.47969239950180054,
      "learning_rate": 9.995386753380149e-06,
      "loss": 1.4943,
      "step": 282
    },
    {
      "epoch": 0.04364842198615743,
      "grad_norm": 0.4887070655822754,
      "learning_rate": 9.995278861811047e-06,
      "loss": 1.5,
      "step": 283
    },
    {
      "epoch": 0.0438026566928223,
      "grad_norm": 0.4667576849460602,
      "learning_rate": 9.995169723730458e-06,
      "loss": 1.4772,
      "step": 284
    },
    {
      "epoch": 0.04395689139948717,
      "grad_norm": 0.5419292449951172,
      "learning_rate": 9.995059339165619e-06,
      "loss": 1.4736,
      "step": 285
    },
    {
      "epoch": 0.04411112610615204,
      "grad_norm": 0.4901038706302643,
      "learning_rate": 9.994947708144072e-06,
      "loss": 1.4694,
      "step": 286
    },
    {
      "epoch": 0.0442653608128169,
      "grad_norm": 0.47999393939971924,
      "learning_rate": 9.994834830693676e-06,
      "loss": 1.4669,
      "step": 287
    },
    {
      "epoch": 0.04441959551948177,
      "grad_norm": 0.5101773738861084,
      "learning_rate": 9.994720706842595e-06,
      "loss": 1.5044,
      "step": 288
    },
    {
      "epoch": 0.04457383022614664,
      "grad_norm": 0.45634952187538147,
      "learning_rate": 9.994605336619309e-06,
      "loss": 1.4656,
      "step": 289
    },
    {
      "epoch": 0.044728064932811507,
      "grad_norm": 0.4852774143218994,
      "learning_rate": 9.994488720052606e-06,
      "loss": 1.4386,
      "step": 290
    },
    {
      "epoch": 0.044882299639476375,
      "grad_norm": 0.5093563199043274,
      "learning_rate": 9.99437085717159e-06,
      "loss": 1.4539,
      "step": 291
    },
    {
      "epoch": 0.045036534346141244,
      "grad_norm": 0.4422704577445984,
      "learning_rate": 9.994251748005668e-06,
      "loss": 1.4716,
      "step": 292
    },
    {
      "epoch": 0.045190769052806105,
      "grad_norm": 0.46956226229667664,
      "learning_rate": 9.994131392584565e-06,
      "loss": 1.4452,
      "step": 293
    },
    {
      "epoch": 0.045345003759470974,
      "grad_norm": 0.5034730434417725,
      "learning_rate": 9.994009790938311e-06,
      "loss": 1.4598,
      "step": 294
    },
    {
      "epoch": 0.04549923846613584,
      "grad_norm": 0.47918686270713806,
      "learning_rate": 9.993886943097254e-06,
      "loss": 1.4674,
      "step": 295
    },
    {
      "epoch": 0.04565347317280071,
      "grad_norm": 0.4500378668308258,
      "learning_rate": 9.993762849092047e-06,
      "loss": 1.4652,
      "step": 296
    },
    {
      "epoch": 0.04580770787946558,
      "grad_norm": 0.507198691368103,
      "learning_rate": 9.993637508953658e-06,
      "loss": 1.4752,
      "step": 297
    },
    {
      "epoch": 0.04596194258613044,
      "grad_norm": 0.42645618319511414,
      "learning_rate": 9.993510922713362e-06,
      "loss": 1.4617,
      "step": 298
    },
    {
      "epoch": 0.04611617729279531,
      "grad_norm": 0.4737209677696228,
      "learning_rate": 9.993383090402746e-06,
      "loss": 1.4178,
      "step": 299
    },
    {
      "epoch": 0.04627041199946018,
      "grad_norm": 0.4638979434967041,
      "learning_rate": 9.993254012053713e-06,
      "loss": 1.4434,
      "step": 300
    },
    {
      "epoch": 0.04642464670612505,
      "grad_norm": 0.4808723032474518,
      "learning_rate": 9.99312368769847e-06,
      "loss": 1.44,
      "step": 301
    },
    {
      "epoch": 0.046578881412789916,
      "grad_norm": 0.4738243818283081,
      "learning_rate": 9.99299211736954e-06,
      "loss": 1.4517,
      "step": 302
    },
    {
      "epoch": 0.04673311611945478,
      "grad_norm": 0.4913071095943451,
      "learning_rate": 9.992859301099752e-06,
      "loss": 1.4143,
      "step": 303
    },
    {
      "epoch": 0.046887350826119646,
      "grad_norm": 0.4924822151660919,
      "learning_rate": 9.99272523892225e-06,
      "loss": 1.4787,
      "step": 304
    },
    {
      "epoch": 0.047041585532784515,
      "grad_norm": 0.48220935463905334,
      "learning_rate": 9.99258993087049e-06,
      "loss": 1.4354,
      "step": 305
    },
    {
      "epoch": 0.04719582023944938,
      "grad_norm": 0.48419612646102905,
      "learning_rate": 9.99245337697823e-06,
      "loss": 1.4448,
      "step": 306
    },
    {
      "epoch": 0.04735005494611425,
      "grad_norm": 0.477363258600235,
      "learning_rate": 9.992315577279552e-06,
      "loss": 1.4507,
      "step": 307
    },
    {
      "epoch": 0.047504289652779114,
      "grad_norm": 0.48646730184555054,
      "learning_rate": 9.992176531808841e-06,
      "loss": 1.4706,
      "step": 308
    },
    {
      "epoch": 0.04765852435944398,
      "grad_norm": 0.5142022967338562,
      "learning_rate": 9.992036240600792e-06,
      "loss": 1.4283,
      "step": 309
    },
    {
      "epoch": 0.04781275906610885,
      "grad_norm": 0.4945804178714752,
      "learning_rate": 9.991894703690414e-06,
      "loss": 1.501,
      "step": 310
    },
    {
      "epoch": 0.04796699377277372,
      "grad_norm": 0.44045937061309814,
      "learning_rate": 9.991751921113027e-06,
      "loss": 1.4747,
      "step": 311
    },
    {
      "epoch": 0.04812122847943859,
      "grad_norm": 0.46978119015693665,
      "learning_rate": 9.991607892904259e-06,
      "loss": 1.4119,
      "step": 312
    },
    {
      "epoch": 0.04827546318610345,
      "grad_norm": 0.5342782139778137,
      "learning_rate": 9.991462619100049e-06,
      "loss": 1.4878,
      "step": 313
    },
    {
      "epoch": 0.04842969789276832,
      "grad_norm": 0.4728383421897888,
      "learning_rate": 9.991316099736651e-06,
      "loss": 1.4876,
      "step": 314
    },
    {
      "epoch": 0.04858393259943319,
      "grad_norm": 0.47804784774780273,
      "learning_rate": 9.991168334850628e-06,
      "loss": 1.4649,
      "step": 315
    },
    {
      "epoch": 0.048738167306098056,
      "grad_norm": 0.5584465265274048,
      "learning_rate": 9.99101932447885e-06,
      "loss": 1.4685,
      "step": 316
    },
    {
      "epoch": 0.048892402012762924,
      "grad_norm": 0.4746883809566498,
      "learning_rate": 9.990869068658502e-06,
      "loss": 1.4271,
      "step": 317
    },
    {
      "epoch": 0.049046636719427786,
      "grad_norm": 0.42429521679878235,
      "learning_rate": 9.990717567427079e-06,
      "loss": 1.4502,
      "step": 318
    },
    {
      "epoch": 0.049200871426092654,
      "grad_norm": 0.5619769096374512,
      "learning_rate": 9.990564820822386e-06,
      "loss": 1.4848,
      "step": 319
    },
    {
      "epoch": 0.04935510613275752,
      "grad_norm": 0.49077802896499634,
      "learning_rate": 9.99041082888254e-06,
      "loss": 1.4661,
      "step": 320
    },
    {
      "epoch": 0.04950934083942239,
      "grad_norm": 0.43036144971847534,
      "learning_rate": 9.990255591645966e-06,
      "loss": 1.3897,
      "step": 321
    },
    {
      "epoch": 0.04966357554608726,
      "grad_norm": 0.522767961025238,
      "learning_rate": 9.990099109151402e-06,
      "loss": 1.4244,
      "step": 322
    },
    {
      "epoch": 0.04981781025275213,
      "grad_norm": 0.45697134733200073,
      "learning_rate": 9.989941381437897e-06,
      "loss": 1.4535,
      "step": 323
    },
    {
      "epoch": 0.04997204495941699,
      "grad_norm": 0.44421565532684326,
      "learning_rate": 9.98978240854481e-06,
      "loss": 1.4096,
      "step": 324
    },
    {
      "epoch": 0.05012627966608186,
      "grad_norm": 0.46733853220939636,
      "learning_rate": 9.989622190511811e-06,
      "loss": 1.4702,
      "step": 325
    },
    {
      "epoch": 0.05028051437274673,
      "grad_norm": 0.4815812110900879,
      "learning_rate": 9.98946072737888e-06,
      "loss": 1.4503,
      "step": 326
    },
    {
      "epoch": 0.050434749079411596,
      "grad_norm": 0.4651709198951721,
      "learning_rate": 9.989298019186307e-06,
      "loss": 1.4296,
      "step": 327
    },
    {
      "epoch": 0.050588983786076465,
      "grad_norm": 0.45224279165267944,
      "learning_rate": 9.989134065974695e-06,
      "loss": 1.4073,
      "step": 328
    },
    {
      "epoch": 0.05074321849274133,
      "grad_norm": 0.45871537923812866,
      "learning_rate": 9.988968867784958e-06,
      "loss": 1.4613,
      "step": 329
    },
    {
      "epoch": 0.050897453199406195,
      "grad_norm": 0.4504376947879791,
      "learning_rate": 9.988802424658317e-06,
      "loss": 1.4464,
      "step": 330
    },
    {
      "epoch": 0.051051687906071064,
      "grad_norm": 0.48538896441459656,
      "learning_rate": 9.988634736636307e-06,
      "loss": 1.4279,
      "step": 331
    },
    {
      "epoch": 0.05120592261273593,
      "grad_norm": 0.4870400130748749,
      "learning_rate": 9.988465803760772e-06,
      "loss": 1.4856,
      "step": 332
    },
    {
      "epoch": 0.0513601573194008,
      "grad_norm": 0.4800507426261902,
      "learning_rate": 9.988295626073866e-06,
      "loss": 1.4326,
      "step": 333
    },
    {
      "epoch": 0.05151439202606566,
      "grad_norm": 0.4934535026550293,
      "learning_rate": 9.988124203618056e-06,
      "loss": 1.4114,
      "step": 334
    },
    {
      "epoch": 0.05166862673273053,
      "grad_norm": 0.512982189655304,
      "learning_rate": 9.98795153643612e-06,
      "loss": 1.4247,
      "step": 335
    },
    {
      "epoch": 0.0518228614393954,
      "grad_norm": 0.4471703767776489,
      "learning_rate": 9.987777624571145e-06,
      "loss": 1.4521,
      "step": 336
    },
    {
      "epoch": 0.05197709614606027,
      "grad_norm": 0.45655134320259094,
      "learning_rate": 9.987602468066527e-06,
      "loss": 1.4566,
      "step": 337
    },
    {
      "epoch": 0.05213133085272514,
      "grad_norm": 0.4856320321559906,
      "learning_rate": 9.987426066965972e-06,
      "loss": 1.4331,
      "step": 338
    },
    {
      "epoch": 0.05228556555939,
      "grad_norm": 0.48707860708236694,
      "learning_rate": 9.987248421313503e-06,
      "loss": 1.4198,
      "step": 339
    },
    {
      "epoch": 0.05243980026605487,
      "grad_norm": 0.482424259185791,
      "learning_rate": 9.987069531153446e-06,
      "loss": 1.4082,
      "step": 340
    },
    {
      "epoch": 0.052594034972719736,
      "grad_norm": 0.5362090468406677,
      "learning_rate": 9.986889396530442e-06,
      "loss": 1.4742,
      "step": 341
    },
    {
      "epoch": 0.052748269679384605,
      "grad_norm": 0.47652480006217957,
      "learning_rate": 9.986708017489442e-06,
      "loss": 1.4893,
      "step": 342
    },
    {
      "epoch": 0.05290250438604947,
      "grad_norm": 0.45364049077033997,
      "learning_rate": 9.986525394075707e-06,
      "loss": 1.4454,
      "step": 343
    },
    {
      "epoch": 0.053056739092714335,
      "grad_norm": 0.5351753234863281,
      "learning_rate": 9.986341526334808e-06,
      "loss": 1.4633,
      "step": 344
    },
    {
      "epoch": 0.0532109737993792,
      "grad_norm": 0.46467578411102295,
      "learning_rate": 9.98615641431263e-06,
      "loss": 1.4628,
      "step": 345
    },
    {
      "epoch": 0.05336520850604407,
      "grad_norm": 0.4417281746864319,
      "learning_rate": 9.985970058055359e-06,
      "loss": 1.4535,
      "step": 346
    },
    {
      "epoch": 0.05351944321270894,
      "grad_norm": 0.5379167795181274,
      "learning_rate": 9.985782457609503e-06,
      "loss": 1.4062,
      "step": 347
    },
    {
      "epoch": 0.05367367791937381,
      "grad_norm": 0.4748031795024872,
      "learning_rate": 9.985593613021873e-06,
      "loss": 1.423,
      "step": 348
    },
    {
      "epoch": 0.05382791262603867,
      "grad_norm": 0.4667428731918335,
      "learning_rate": 9.985403524339594e-06,
      "loss": 1.4063,
      "step": 349
    },
    {
      "epoch": 0.05398214733270354,
      "grad_norm": 0.5456904172897339,
      "learning_rate": 9.985212191610101e-06,
      "loss": 1.4352,
      "step": 350
    },
    {
      "epoch": 0.05413638203936841,
      "grad_norm": 0.438327431678772,
      "learning_rate": 9.985019614881137e-06,
      "loss": 1.443,
      "step": 351
    },
    {
      "epoch": 0.05429061674603328,
      "grad_norm": 0.4759283661842346,
      "learning_rate": 9.98482579420076e-06,
      "loss": 1.425,
      "step": 352
    },
    {
      "epoch": 0.054444851452698145,
      "grad_norm": 0.47276994585990906,
      "learning_rate": 9.984630729617331e-06,
      "loss": 1.4331,
      "step": 353
    },
    {
      "epoch": 0.054599086159363014,
      "grad_norm": 0.4805431663990021,
      "learning_rate": 9.98443442117953e-06,
      "loss": 1.4726,
      "step": 354
    },
    {
      "epoch": 0.054753320866027876,
      "grad_norm": 0.4648710787296295,
      "learning_rate": 9.984236868936344e-06,
      "loss": 1.4078,
      "step": 355
    },
    {
      "epoch": 0.054907555572692744,
      "grad_norm": 0.4526321291923523,
      "learning_rate": 9.984038072937065e-06,
      "loss": 1.4609,
      "step": 356
    },
    {
      "epoch": 0.05506179027935761,
      "grad_norm": 0.4688374996185303,
      "learning_rate": 9.983838033231306e-06,
      "loss": 1.4013,
      "step": 357
    },
    {
      "epoch": 0.05521602498602248,
      "grad_norm": 0.4460720121860504,
      "learning_rate": 9.983636749868979e-06,
      "loss": 1.4452,
      "step": 358
    },
    {
      "epoch": 0.05537025969268735,
      "grad_norm": 0.5248163342475891,
      "learning_rate": 9.983434222900316e-06,
      "loss": 1.4115,
      "step": 359
    },
    {
      "epoch": 0.05552449439935221,
      "grad_norm": 0.5059958100318909,
      "learning_rate": 9.983230452375853e-06,
      "loss": 1.4302,
      "step": 360
    },
    {
      "epoch": 0.05567872910601708,
      "grad_norm": 0.4585413336753845,
      "learning_rate": 9.983025438346439e-06,
      "loss": 1.4372,
      "step": 361
    },
    {
      "epoch": 0.05583296381268195,
      "grad_norm": 0.5210665464401245,
      "learning_rate": 9.982819180863233e-06,
      "loss": 1.4344,
      "step": 362
    },
    {
      "epoch": 0.05598719851934682,
      "grad_norm": 0.5004061460494995,
      "learning_rate": 9.982611679977703e-06,
      "loss": 1.4463,
      "step": 363
    },
    {
      "epoch": 0.056141433226011686,
      "grad_norm": 0.4296887218952179,
      "learning_rate": 9.98240293574163e-06,
      "loss": 1.421,
      "step": 364
    },
    {
      "epoch": 0.05629566793267655,
      "grad_norm": 0.4660370945930481,
      "learning_rate": 9.982192948207103e-06,
      "loss": 1.4076,
      "step": 365
    },
    {
      "epoch": 0.056449902639341416,
      "grad_norm": 0.5059395432472229,
      "learning_rate": 9.981981717426522e-06,
      "loss": 1.4589,
      "step": 366
    },
    {
      "epoch": 0.056604137346006285,
      "grad_norm": 0.48390713334083557,
      "learning_rate": 9.981769243452595e-06,
      "loss": 1.4529,
      "step": 367
    },
    {
      "epoch": 0.056758372052671154,
      "grad_norm": 0.46936866641044617,
      "learning_rate": 9.981555526338345e-06,
      "loss": 1.4164,
      "step": 368
    },
    {
      "epoch": 0.05691260675933602,
      "grad_norm": 0.5150275826454163,
      "learning_rate": 9.981340566137103e-06,
      "loss": 1.4518,
      "step": 369
    },
    {
      "epoch": 0.057066841466000884,
      "grad_norm": 0.4665602743625641,
      "learning_rate": 9.981124362902507e-06,
      "loss": 1.39,
      "step": 370
    },
    {
      "epoch": 0.05722107617266575,
      "grad_norm": 0.46382731199264526,
      "learning_rate": 9.980906916688508e-06,
      "loss": 1.4427,
      "step": 371
    },
    {
      "epoch": 0.05737531087933062,
      "grad_norm": 0.49271711707115173,
      "learning_rate": 9.980688227549372e-06,
      "loss": 1.3956,
      "step": 372
    },
    {
      "epoch": 0.05752954558599549,
      "grad_norm": 0.5274916887283325,
      "learning_rate": 9.980468295539663e-06,
      "loss": 1.4818,
      "step": 373
    },
    {
      "epoch": 0.05768378029266036,
      "grad_norm": 0.4875819981098175,
      "learning_rate": 9.980247120714265e-06,
      "loss": 1.3953,
      "step": 374
    },
    {
      "epoch": 0.05783801499932522,
      "grad_norm": 0.4955143630504608,
      "learning_rate": 9.980024703128375e-06,
      "loss": 1.3925,
      "step": 375
    },
    {
      "epoch": 0.05799224970599009,
      "grad_norm": 0.5156822204589844,
      "learning_rate": 9.979801042837487e-06,
      "loss": 1.4886,
      "step": 376
    },
    {
      "epoch": 0.05814648441265496,
      "grad_norm": 0.4711759090423584,
      "learning_rate": 9.979576139897416e-06,
      "loss": 1.4357,
      "step": 377
    },
    {
      "epoch": 0.058300719119319826,
      "grad_norm": 0.4979749321937561,
      "learning_rate": 9.979349994364283e-06,
      "loss": 1.4075,
      "step": 378
    },
    {
      "epoch": 0.058454953825984694,
      "grad_norm": 0.49984684586524963,
      "learning_rate": 9.979122606294523e-06,
      "loss": 1.4653,
      "step": 379
    },
    {
      "epoch": 0.05860918853264956,
      "grad_norm": 0.47537845373153687,
      "learning_rate": 9.978893975744872e-06,
      "loss": 1.439,
      "step": 380
    },
    {
      "epoch": 0.058763423239314425,
      "grad_norm": 0.46360743045806885,
      "learning_rate": 9.978664102772388e-06,
      "loss": 1.4371,
      "step": 381
    },
    {
      "epoch": 0.05891765794597929,
      "grad_norm": 0.5247479677200317,
      "learning_rate": 9.978432987434428e-06,
      "loss": 1.4487,
      "step": 382
    },
    {
      "epoch": 0.05907189265264416,
      "grad_norm": 0.45880356431007385,
      "learning_rate": 9.97820062978867e-06,
      "loss": 1.4783,
      "step": 383
    },
    {
      "epoch": 0.05922612735930903,
      "grad_norm": 0.471962571144104,
      "learning_rate": 9.97796702989309e-06,
      "loss": 1.4386,
      "step": 384
    },
    {
      "epoch": 0.0593803620659739,
      "grad_norm": 0.5721044540405273,
      "learning_rate": 9.977732187805982e-06,
      "loss": 1.4859,
      "step": 385
    },
    {
      "epoch": 0.05953459677263876,
      "grad_norm": 0.5007787942886353,
      "learning_rate": 9.977496103585949e-06,
      "loss": 1.4008,
      "step": 386
    },
    {
      "epoch": 0.05968883147930363,
      "grad_norm": 0.48961400985717773,
      "learning_rate": 9.977258777291902e-06,
      "loss": 1.4447,
      "step": 387
    },
    {
      "epoch": 0.0598430661859685,
      "grad_norm": 0.5735716223716736,
      "learning_rate": 9.977020208983066e-06,
      "loss": 1.4565,
      "step": 388
    },
    {
      "epoch": 0.059997300892633366,
      "grad_norm": 0.477939635515213,
      "learning_rate": 9.976780398718969e-06,
      "loss": 1.4011,
      "step": 389
    },
    {
      "epoch": 0.060151535599298235,
      "grad_norm": 0.4673793613910675,
      "learning_rate": 9.976539346559453e-06,
      "loss": 1.4266,
      "step": 390
    },
    {
      "epoch": 0.0603057703059631,
      "grad_norm": 0.5940758585929871,
      "learning_rate": 9.976297052564672e-06,
      "loss": 1.4022,
      "step": 391
    },
    {
      "epoch": 0.060460005012627965,
      "grad_norm": 0.5857895612716675,
      "learning_rate": 9.976053516795085e-06,
      "loss": 1.3914,
      "step": 392
    },
    {
      "epoch": 0.060614239719292834,
      "grad_norm": 0.5206599235534668,
      "learning_rate": 9.975808739311465e-06,
      "loss": 1.4112,
      "step": 393
    },
    {
      "epoch": 0.0607684744259577,
      "grad_norm": 0.8916546702384949,
      "learning_rate": 9.975562720174892e-06,
      "loss": 1.4316,
      "step": 394
    },
    {
      "epoch": 0.06092270913262257,
      "grad_norm": 0.5995873808860779,
      "learning_rate": 9.97531545944676e-06,
      "loss": 1.4445,
      "step": 395
    },
    {
      "epoch": 0.06107694383928743,
      "grad_norm": 0.6093815565109253,
      "learning_rate": 9.975066957188766e-06,
      "loss": 1.4183,
      "step": 396
    },
    {
      "epoch": 0.0612311785459523,
      "grad_norm": 15.272882461547852,
      "learning_rate": 9.974817213462924e-06,
      "loss": 1.3998,
      "step": 397
    },
    {
      "epoch": 0.06138541325261717,
      "grad_norm": 0.543540358543396,
      "learning_rate": 9.974566228331552e-06,
      "loss": 1.4498,
      "step": 398
    },
    {
      "epoch": 0.06153964795928204,
      "grad_norm": 0.4766438901424408,
      "learning_rate": 9.974314001857281e-06,
      "loss": 1.4315,
      "step": 399
    },
    {
      "epoch": 0.06169388266594691,
      "grad_norm": 0.5161774754524231,
      "learning_rate": 9.974060534103054e-06,
      "loss": 1.3846,
      "step": 400
    },
    {
      "epoch": 0.06184811737261177,
      "grad_norm": 0.44010746479034424,
      "learning_rate": 9.973805825132116e-06,
      "loss": 1.4421,
      "step": 401
    },
    {
      "epoch": 0.06200235207927664,
      "grad_norm": 0.48070842027664185,
      "learning_rate": 9.973549875008029e-06,
      "loss": 1.4278,
      "step": 402
    },
    {
      "epoch": 0.062156586785941506,
      "grad_norm": 0.47066718339920044,
      "learning_rate": 9.973292683794665e-06,
      "loss": 1.4244,
      "step": 403
    },
    {
      "epoch": 0.062310821492606375,
      "grad_norm": 0.4821130335330963,
      "learning_rate": 9.973034251556197e-06,
      "loss": 1.492,
      "step": 404
    },
    {
      "epoch": 0.06246505619927124,
      "grad_norm": 0.47957906126976013,
      "learning_rate": 9.972774578357118e-06,
      "loss": 1.449,
      "step": 405
    },
    {
      "epoch": 0.06261929090593611,
      "grad_norm": 0.42835289239883423,
      "learning_rate": 9.972513664262225e-06,
      "loss": 1.4245,
      "step": 406
    },
    {
      "epoch": 0.06277352561260098,
      "grad_norm": 0.4677746891975403,
      "learning_rate": 9.972251509336626e-06,
      "loss": 1.4087,
      "step": 407
    },
    {
      "epoch": 0.06292776031926585,
      "grad_norm": 0.43734127283096313,
      "learning_rate": 9.971988113645737e-06,
      "loss": 1.4076,
      "step": 408
    },
    {
      "epoch": 0.0630819950259307,
      "grad_norm": 0.5043666362762451,
      "learning_rate": 9.971723477255289e-06,
      "loss": 1.4518,
      "step": 409
    },
    {
      "epoch": 0.06323622973259557,
      "grad_norm": 0.4832194447517395,
      "learning_rate": 9.971457600231316e-06,
      "loss": 1.4511,
      "step": 410
    },
    {
      "epoch": 0.06339046443926044,
      "grad_norm": 0.4825354516506195,
      "learning_rate": 9.971190482640166e-06,
      "loss": 1.4213,
      "step": 411
    },
    {
      "epoch": 0.06354469914592531,
      "grad_norm": 0.5067551136016846,
      "learning_rate": 9.970922124548492e-06,
      "loss": 1.4389,
      "step": 412
    },
    {
      "epoch": 0.06369893385259018,
      "grad_norm": 0.4832996428012848,
      "learning_rate": 9.970652526023262e-06,
      "loss": 1.4497,
      "step": 413
    },
    {
      "epoch": 0.06385316855925505,
      "grad_norm": 0.4582710862159729,
      "learning_rate": 9.970381687131751e-06,
      "loss": 1.4547,
      "step": 414
    },
    {
      "epoch": 0.06400740326591992,
      "grad_norm": 0.45857349038124084,
      "learning_rate": 9.970109607941544e-06,
      "loss": 1.4107,
      "step": 415
    },
    {
      "epoch": 0.06416163797258478,
      "grad_norm": 0.4977855086326599,
      "learning_rate": 9.969836288520534e-06,
      "loss": 1.4477,
      "step": 416
    },
    {
      "epoch": 0.06431587267924965,
      "grad_norm": 0.4536735713481903,
      "learning_rate": 9.969561728936922e-06,
      "loss": 1.3812,
      "step": 417
    },
    {
      "epoch": 0.06447010738591452,
      "grad_norm": 0.4105980098247528,
      "learning_rate": 9.969285929259226e-06,
      "loss": 1.399,
      "step": 418
    },
    {
      "epoch": 0.06462434209257938,
      "grad_norm": 0.479280948638916,
      "learning_rate": 9.969008889556269e-06,
      "loss": 1.4202,
      "step": 419
    },
    {
      "epoch": 0.06477857679924424,
      "grad_norm": 0.44376733899116516,
      "learning_rate": 9.968730609897177e-06,
      "loss": 1.4232,
      "step": 420
    },
    {
      "epoch": 0.06493281150590911,
      "grad_norm": 0.41896483302116394,
      "learning_rate": 9.968451090351394e-06,
      "loss": 1.4072,
      "step": 421
    },
    {
      "epoch": 0.06508704621257398,
      "grad_norm": 0.5083034634590149,
      "learning_rate": 9.968170330988673e-06,
      "loss": 1.4272,
      "step": 422
    },
    {
      "epoch": 0.06524128091923885,
      "grad_norm": 0.44754496216773987,
      "learning_rate": 9.967888331879073e-06,
      "loss": 1.4526,
      "step": 423
    },
    {
      "epoch": 0.06539551562590372,
      "grad_norm": 0.4666605293750763,
      "learning_rate": 9.96760509309296e-06,
      "loss": 1.4244,
      "step": 424
    },
    {
      "epoch": 0.06554975033256859,
      "grad_norm": 0.4701414108276367,
      "learning_rate": 9.967320614701019e-06,
      "loss": 1.4179,
      "step": 425
    },
    {
      "epoch": 0.06570398503923346,
      "grad_norm": 0.4674292206764221,
      "learning_rate": 9.967034896774233e-06,
      "loss": 1.4335,
      "step": 426
    },
    {
      "epoch": 0.06585821974589832,
      "grad_norm": 0.44576194882392883,
      "learning_rate": 9.9667479393839e-06,
      "loss": 1.3998,
      "step": 427
    },
    {
      "epoch": 0.0660124544525632,
      "grad_norm": 0.4495324194431305,
      "learning_rate": 9.966459742601631e-06,
      "loss": 1.4125,
      "step": 428
    },
    {
      "epoch": 0.06616668915922806,
      "grad_norm": 0.5072752833366394,
      "learning_rate": 9.966170306499337e-06,
      "loss": 1.4233,
      "step": 429
    },
    {
      "epoch": 0.06632092386589292,
      "grad_norm": 0.4577788710594177,
      "learning_rate": 9.965879631149245e-06,
      "loss": 1.3595,
      "step": 430
    },
    {
      "epoch": 0.06647515857255779,
      "grad_norm": 0.5432767271995544,
      "learning_rate": 9.96558771662389e-06,
      "loss": 1.4081,
      "step": 431
    },
    {
      "epoch": 0.06662939327922265,
      "grad_norm": 0.5130487084388733,
      "learning_rate": 9.965294562996118e-06,
      "loss": 1.4438,
      "step": 432
    },
    {
      "epoch": 0.06678362798588752,
      "grad_norm": 0.4710625410079956,
      "learning_rate": 9.965000170339076e-06,
      "loss": 1.3916,
      "step": 433
    },
    {
      "epoch": 0.06693786269255239,
      "grad_norm": 0.4908548891544342,
      "learning_rate": 9.964704538726231e-06,
      "loss": 1.3885,
      "step": 434
    },
    {
      "epoch": 0.06709209739921726,
      "grad_norm": 0.5099604725837708,
      "learning_rate": 9.964407668231354e-06,
      "loss": 1.4404,
      "step": 435
    },
    {
      "epoch": 0.06724633210588213,
      "grad_norm": 0.4541493356227875,
      "learning_rate": 9.964109558928524e-06,
      "loss": 1.4706,
      "step": 436
    },
    {
      "epoch": 0.067400566812547,
      "grad_norm": 0.5716336965560913,
      "learning_rate": 9.96381021089213e-06,
      "loss": 1.395,
      "step": 437
    },
    {
      "epoch": 0.06755480151921187,
      "grad_norm": 0.5119542479515076,
      "learning_rate": 9.963509624196871e-06,
      "loss": 1.4332,
      "step": 438
    },
    {
      "epoch": 0.06770903622587673,
      "grad_norm": 0.4324071407318115,
      "learning_rate": 9.963207798917758e-06,
      "loss": 1.4372,
      "step": 439
    },
    {
      "epoch": 0.06786327093254159,
      "grad_norm": 0.46256715059280396,
      "learning_rate": 9.962904735130105e-06,
      "loss": 1.425,
      "step": 440
    },
    {
      "epoch": 0.06801750563920646,
      "grad_norm": 0.5053929686546326,
      "learning_rate": 9.962600432909537e-06,
      "loss": 1.4286,
      "step": 441
    },
    {
      "epoch": 0.06817174034587133,
      "grad_norm": 0.46784093976020813,
      "learning_rate": 9.96229489233199e-06,
      "loss": 1.4165,
      "step": 442
    },
    {
      "epoch": 0.0683259750525362,
      "grad_norm": 0.4608568549156189,
      "learning_rate": 9.961988113473708e-06,
      "loss": 1.4143,
      "step": 443
    },
    {
      "epoch": 0.06848020975920106,
      "grad_norm": 0.5083256363868713,
      "learning_rate": 9.961680096411245e-06,
      "loss": 1.4202,
      "step": 444
    },
    {
      "epoch": 0.06863444446586593,
      "grad_norm": 0.4442773163318634,
      "learning_rate": 9.961370841221461e-06,
      "loss": 1.3899,
      "step": 445
    },
    {
      "epoch": 0.0687886791725308,
      "grad_norm": 0.42745813727378845,
      "learning_rate": 9.961060347981529e-06,
      "loss": 1.4039,
      "step": 446
    },
    {
      "epoch": 0.06894291387919567,
      "grad_norm": 0.5610799193382263,
      "learning_rate": 9.960748616768927e-06,
      "loss": 1.4298,
      "step": 447
    },
    {
      "epoch": 0.06909714858586054,
      "grad_norm": 0.47480154037475586,
      "learning_rate": 9.960435647661444e-06,
      "loss": 1.4292,
      "step": 448
    },
    {
      "epoch": 0.0692513832925254,
      "grad_norm": 0.46681931614875793,
      "learning_rate": 9.960121440737179e-06,
      "loss": 1.3831,
      "step": 449
    },
    {
      "epoch": 0.06940561799919026,
      "grad_norm": 0.47058233618736267,
      "learning_rate": 9.959805996074538e-06,
      "loss": 1.4064,
      "step": 450
    },
    {
      "epoch": 0.06955985270585513,
      "grad_norm": 0.42082738876342773,
      "learning_rate": 9.959489313752235e-06,
      "loss": 1.4319,
      "step": 451
    },
    {
      "epoch": 0.06971408741252,
      "grad_norm": 0.4539491534233093,
      "learning_rate": 9.959171393849296e-06,
      "loss": 1.4184,
      "step": 452
    },
    {
      "epoch": 0.06986832211918487,
      "grad_norm": 0.45387235283851624,
      "learning_rate": 9.958852236445054e-06,
      "loss": 1.4268,
      "step": 453
    },
    {
      "epoch": 0.07002255682584974,
      "grad_norm": 0.4910185933113098,
      "learning_rate": 9.958531841619149e-06,
      "loss": 1.4459,
      "step": 454
    },
    {
      "epoch": 0.0701767915325146,
      "grad_norm": 0.46503302454948425,
      "learning_rate": 9.958210209451534e-06,
      "loss": 1.3797,
      "step": 455
    },
    {
      "epoch": 0.07033102623917947,
      "grad_norm": 0.4434162974357605,
      "learning_rate": 9.957887340022467e-06,
      "loss": 1.3723,
      "step": 456
    },
    {
      "epoch": 0.07048526094584434,
      "grad_norm": 0.48650768399238586,
      "learning_rate": 9.957563233412515e-06,
      "loss": 1.4484,
      "step": 457
    },
    {
      "epoch": 0.07063949565250921,
      "grad_norm": 0.48305729031562805,
      "learning_rate": 9.957237889702559e-06,
      "loss": 1.4251,
      "step": 458
    },
    {
      "epoch": 0.07079373035917408,
      "grad_norm": 0.4370194971561432,
      "learning_rate": 9.95691130897378e-06,
      "loss": 1.3741,
      "step": 459
    },
    {
      "epoch": 0.07094796506583895,
      "grad_norm": 0.47357454895973206,
      "learning_rate": 9.956583491307674e-06,
      "loss": 1.4555,
      "step": 460
    },
    {
      "epoch": 0.0711021997725038,
      "grad_norm": 0.4653128981590271,
      "learning_rate": 9.956254436786045e-06,
      "loss": 1.4068,
      "step": 461
    },
    {
      "epoch": 0.07125643447916867,
      "grad_norm": 0.4576094448566437,
      "learning_rate": 9.955924145491005e-06,
      "loss": 1.4189,
      "step": 462
    },
    {
      "epoch": 0.07141066918583354,
      "grad_norm": 0.5161656141281128,
      "learning_rate": 9.955592617504972e-06,
      "loss": 1.4005,
      "step": 463
    },
    {
      "epoch": 0.07156490389249841,
      "grad_norm": 0.44188177585601807,
      "learning_rate": 9.955259852910675e-06,
      "loss": 1.4089,
      "step": 464
    },
    {
      "epoch": 0.07171913859916328,
      "grad_norm": 0.4291110038757324,
      "learning_rate": 9.954925851791153e-06,
      "loss": 1.3606,
      "step": 465
    },
    {
      "epoch": 0.07187337330582814,
      "grad_norm": 0.46663251519203186,
      "learning_rate": 9.954590614229752e-06,
      "loss": 1.4353,
      "step": 466
    },
    {
      "epoch": 0.07202760801249301,
      "grad_norm": 0.4870257079601288,
      "learning_rate": 9.954254140310125e-06,
      "loss": 1.414,
      "step": 467
    },
    {
      "epoch": 0.07218184271915788,
      "grad_norm": 0.43861398100852966,
      "learning_rate": 9.953916430116234e-06,
      "loss": 1.3677,
      "step": 468
    },
    {
      "epoch": 0.07233607742582275,
      "grad_norm": 0.47025996446609497,
      "learning_rate": 9.953577483732355e-06,
      "loss": 1.3961,
      "step": 469
    },
    {
      "epoch": 0.07249031213248762,
      "grad_norm": 0.450141966342926,
      "learning_rate": 9.953237301243064e-06,
      "loss": 1.4271,
      "step": 470
    },
    {
      "epoch": 0.07264454683915247,
      "grad_norm": 0.46584373712539673,
      "learning_rate": 9.952895882733249e-06,
      "loss": 1.3757,
      "step": 471
    },
    {
      "epoch": 0.07279878154581734,
      "grad_norm": 0.5051293969154358,
      "learning_rate": 9.95255322828811e-06,
      "loss": 1.4145,
      "step": 472
    },
    {
      "epoch": 0.07295301625248221,
      "grad_norm": 0.4585671126842499,
      "learning_rate": 9.95220933799315e-06,
      "loss": 1.4621,
      "step": 473
    },
    {
      "epoch": 0.07310725095914708,
      "grad_norm": 0.458593487739563,
      "learning_rate": 9.951864211934183e-06,
      "loss": 1.3547,
      "step": 474
    },
    {
      "epoch": 0.07326148566581195,
      "grad_norm": 0.47871628403663635,
      "learning_rate": 9.951517850197331e-06,
      "loss": 1.4152,
      "step": 475
    },
    {
      "epoch": 0.07341572037247682,
      "grad_norm": 0.44403016567230225,
      "learning_rate": 9.951170252869027e-06,
      "loss": 1.4207,
      "step": 476
    },
    {
      "epoch": 0.07356995507914169,
      "grad_norm": 0.4658779203891754,
      "learning_rate": 9.950821420036006e-06,
      "loss": 1.4242,
      "step": 477
    },
    {
      "epoch": 0.07372418978580655,
      "grad_norm": 0.4666624069213867,
      "learning_rate": 9.950471351785317e-06,
      "loss": 1.3814,
      "step": 478
    },
    {
      "epoch": 0.07387842449247142,
      "grad_norm": 0.4759904444217682,
      "learning_rate": 9.950120048204314e-06,
      "loss": 1.4177,
      "step": 479
    },
    {
      "epoch": 0.07403265919913629,
      "grad_norm": 0.4552132785320282,
      "learning_rate": 9.949767509380661e-06,
      "loss": 1.3858,
      "step": 480
    },
    {
      "epoch": 0.07418689390580115,
      "grad_norm": 0.4564070999622345,
      "learning_rate": 9.949413735402332e-06,
      "loss": 1.3558,
      "step": 481
    },
    {
      "epoch": 0.07434112861246601,
      "grad_norm": 0.5005853772163391,
      "learning_rate": 9.949058726357603e-06,
      "loss": 1.4229,
      "step": 482
    },
    {
      "epoch": 0.07449536331913088,
      "grad_norm": 0.44161200523376465,
      "learning_rate": 9.948702482335067e-06,
      "loss": 1.3854,
      "step": 483
    },
    {
      "epoch": 0.07464959802579575,
      "grad_norm": 0.43942132592201233,
      "learning_rate": 9.948345003423615e-06,
      "loss": 1.3854,
      "step": 484
    },
    {
      "epoch": 0.07480383273246062,
      "grad_norm": 0.5212045907974243,
      "learning_rate": 9.947986289712456e-06,
      "loss": 1.4402,
      "step": 485
    },
    {
      "epoch": 0.07495806743912549,
      "grad_norm": 0.44679194688796997,
      "learning_rate": 9.9476263412911e-06,
      "loss": 1.4045,
      "step": 486
    },
    {
      "epoch": 0.07511230214579036,
      "grad_norm": 0.44435495138168335,
      "learning_rate": 9.94726515824937e-06,
      "loss": 1.4144,
      "step": 487
    },
    {
      "epoch": 0.07526653685245523,
      "grad_norm": 0.4690037667751312,
      "learning_rate": 9.946902740677392e-06,
      "loss": 1.4128,
      "step": 488
    },
    {
      "epoch": 0.0754207715591201,
      "grad_norm": 0.47753167152404785,
      "learning_rate": 9.946539088665605e-06,
      "loss": 1.4132,
      "step": 489
    },
    {
      "epoch": 0.07557500626578496,
      "grad_norm": 0.4679515063762665,
      "learning_rate": 9.946174202304755e-06,
      "loss": 1.4144,
      "step": 490
    },
    {
      "epoch": 0.07572924097244983,
      "grad_norm": 0.5085904598236084,
      "learning_rate": 9.945808081685893e-06,
      "loss": 1.4096,
      "step": 491
    },
    {
      "epoch": 0.07588347567911469,
      "grad_norm": 0.4496731162071228,
      "learning_rate": 9.94544072690038e-06,
      "loss": 1.3907,
      "step": 492
    },
    {
      "epoch": 0.07603771038577956,
      "grad_norm": 0.4731052815914154,
      "learning_rate": 9.945072138039884e-06,
      "loss": 1.3817,
      "step": 493
    },
    {
      "epoch": 0.07619194509244442,
      "grad_norm": 0.49058064818382263,
      "learning_rate": 9.944702315196386e-06,
      "loss": 1.4072,
      "step": 494
    },
    {
      "epoch": 0.07634617979910929,
      "grad_norm": 0.3971802890300751,
      "learning_rate": 9.944331258462167e-06,
      "loss": 1.4191,
      "step": 495
    },
    {
      "epoch": 0.07650041450577416,
      "grad_norm": 0.44060471653938293,
      "learning_rate": 9.943958967929821e-06,
      "loss": 1.3959,
      "step": 496
    },
    {
      "epoch": 0.07665464921243903,
      "grad_norm": 0.5203728675842285,
      "learning_rate": 9.943585443692248e-06,
      "loss": 1.4548,
      "step": 497
    },
    {
      "epoch": 0.0768088839191039,
      "grad_norm": 0.475394606590271,
      "learning_rate": 9.943210685842658e-06,
      "loss": 1.4231,
      "step": 498
    },
    {
      "epoch": 0.07696311862576877,
      "grad_norm": 0.4179852604866028,
      "learning_rate": 9.942834694474565e-06,
      "loss": 1.4058,
      "step": 499
    },
    {
      "epoch": 0.07711735333243364,
      "grad_norm": 0.4127453863620758,
      "learning_rate": 9.942457469681794e-06,
      "loss": 1.3713,
      "step": 500
    },
    {
      "epoch": 0.0772715880390985,
      "grad_norm": 0.42606067657470703,
      "learning_rate": 9.942079011558478e-06,
      "loss": 1.4487,
      "step": 501
    },
    {
      "epoch": 0.07742582274576336,
      "grad_norm": 0.4229513108730316,
      "learning_rate": 9.941699320199056e-06,
      "loss": 1.3565,
      "step": 502
    },
    {
      "epoch": 0.07758005745242823,
      "grad_norm": 0.44844040274620056,
      "learning_rate": 9.941318395698277e-06,
      "loss": 1.4019,
      "step": 503
    },
    {
      "epoch": 0.0777342921590931,
      "grad_norm": 0.48133325576782227,
      "learning_rate": 9.940936238151192e-06,
      "loss": 1.4171,
      "step": 504
    },
    {
      "epoch": 0.07788852686575796,
      "grad_norm": 0.47930583357810974,
      "learning_rate": 9.940552847653166e-06,
      "loss": 1.373,
      "step": 505
    },
    {
      "epoch": 0.07804276157242283,
      "grad_norm": 0.48454520106315613,
      "learning_rate": 9.940168224299872e-06,
      "loss": 1.3805,
      "step": 506
    },
    {
      "epoch": 0.0781969962790877,
      "grad_norm": 0.5197977423667908,
      "learning_rate": 9.939782368187284e-06,
      "loss": 1.4257,
      "step": 507
    },
    {
      "epoch": 0.07835123098575257,
      "grad_norm": 0.4360477328300476,
      "learning_rate": 9.93939527941169e-06,
      "loss": 1.4139,
      "step": 508
    },
    {
      "epoch": 0.07850546569241744,
      "grad_norm": 0.45959925651550293,
      "learning_rate": 9.939006958069682e-06,
      "loss": 1.4148,
      "step": 509
    },
    {
      "epoch": 0.07865970039908231,
      "grad_norm": 0.5091875195503235,
      "learning_rate": 9.938617404258163e-06,
      "loss": 1.4721,
      "step": 510
    },
    {
      "epoch": 0.07881393510574718,
      "grad_norm": 0.4871944785118103,
      "learning_rate": 9.93822661807434e-06,
      "loss": 1.4325,
      "step": 511
    },
    {
      "epoch": 0.07896816981241203,
      "grad_norm": 0.4516337513923645,
      "learning_rate": 9.937834599615728e-06,
      "loss": 1.3699,
      "step": 512
    },
    {
      "epoch": 0.0791224045190769,
      "grad_norm": 0.4808272421360016,
      "learning_rate": 9.937441348980152e-06,
      "loss": 1.3957,
      "step": 513
    },
    {
      "epoch": 0.07927663922574177,
      "grad_norm": 0.43793177604675293,
      "learning_rate": 9.937046866265742e-06,
      "loss": 1.3617,
      "step": 514
    },
    {
      "epoch": 0.07943087393240664,
      "grad_norm": 0.44577157497406006,
      "learning_rate": 9.93665115157094e-06,
      "loss": 1.3803,
      "step": 515
    },
    {
      "epoch": 0.0795851086390715,
      "grad_norm": 0.4638996720314026,
      "learning_rate": 9.936254204994487e-06,
      "loss": 1.4105,
      "step": 516
    },
    {
      "epoch": 0.07973934334573637,
      "grad_norm": 0.46314525604248047,
      "learning_rate": 9.935856026635436e-06,
      "loss": 1.4045,
      "step": 517
    },
    {
      "epoch": 0.07989357805240124,
      "grad_norm": 0.4436893165111542,
      "learning_rate": 9.935456616593151e-06,
      "loss": 1.3591,
      "step": 518
    },
    {
      "epoch": 0.08004781275906611,
      "grad_norm": 0.5167062282562256,
      "learning_rate": 9.935055974967299e-06,
      "loss": 1.4185,
      "step": 519
    },
    {
      "epoch": 0.08020204746573098,
      "grad_norm": 0.4903646409511566,
      "learning_rate": 9.934654101857854e-06,
      "loss": 1.4204,
      "step": 520
    },
    {
      "epoch": 0.08035628217239585,
      "grad_norm": 0.41959303617477417,
      "learning_rate": 9.9342509973651e-06,
      "loss": 1.4077,
      "step": 521
    },
    {
      "epoch": 0.08051051687906072,
      "grad_norm": 0.47519707679748535,
      "learning_rate": 9.933846661589628e-06,
      "loss": 1.3868,
      "step": 522
    },
    {
      "epoch": 0.08066475158572557,
      "grad_norm": 0.42335045337677,
      "learning_rate": 9.93344109463233e-06,
      "loss": 1.392,
      "step": 523
    },
    {
      "epoch": 0.08081898629239044,
      "grad_norm": 0.4691239595413208,
      "learning_rate": 9.933034296594415e-06,
      "loss": 1.4053,
      "step": 524
    },
    {
      "epoch": 0.08097322099905531,
      "grad_norm": 0.49240291118621826,
      "learning_rate": 9.932626267577394e-06,
      "loss": 1.4047,
      "step": 525
    },
    {
      "epoch": 0.08112745570572018,
      "grad_norm": 0.4474901556968689,
      "learning_rate": 9.932217007683085e-06,
      "loss": 1.3939,
      "step": 526
    },
    {
      "epoch": 0.08128169041238505,
      "grad_norm": 0.45516321063041687,
      "learning_rate": 9.931806517013612e-06,
      "loss": 1.3773,
      "step": 527
    },
    {
      "epoch": 0.08143592511904992,
      "grad_norm": 0.4424881339073181,
      "learning_rate": 9.931394795671412e-06,
      "loss": 1.3964,
      "step": 528
    },
    {
      "epoch": 0.08159015982571478,
      "grad_norm": 0.46286776661872864,
      "learning_rate": 9.930981843759222e-06,
      "loss": 1.4642,
      "step": 529
    },
    {
      "epoch": 0.08174439453237965,
      "grad_norm": 0.49491822719573975,
      "learning_rate": 9.930567661380089e-06,
      "loss": 1.446,
      "step": 530
    },
    {
      "epoch": 0.08189862923904452,
      "grad_norm": 0.4691407084465027,
      "learning_rate": 9.930152248637371e-06,
      "loss": 1.386,
      "step": 531
    },
    {
      "epoch": 0.08205286394570939,
      "grad_norm": 0.48742979764938354,
      "learning_rate": 9.929735605634722e-06,
      "loss": 1.4453,
      "step": 532
    },
    {
      "epoch": 0.08220709865237424,
      "grad_norm": 0.4329613745212555,
      "learning_rate": 9.929317732476118e-06,
      "loss": 1.402,
      "step": 533
    },
    {
      "epoch": 0.08236133335903911,
      "grad_norm": 0.4595591425895691,
      "learning_rate": 9.92889862926583e-06,
      "loss": 1.3863,
      "step": 534
    },
    {
      "epoch": 0.08251556806570398,
      "grad_norm": 0.46652889251708984,
      "learning_rate": 9.928478296108442e-06,
      "loss": 1.4018,
      "step": 535
    },
    {
      "epoch": 0.08266980277236885,
      "grad_norm": 0.45996397733688354,
      "learning_rate": 9.928056733108842e-06,
      "loss": 1.3791,
      "step": 536
    },
    {
      "epoch": 0.08282403747903372,
      "grad_norm": 0.4811365008354187,
      "learning_rate": 9.927633940372226e-06,
      "loss": 1.4108,
      "step": 537
    },
    {
      "epoch": 0.08297827218569859,
      "grad_norm": 0.5085799694061279,
      "learning_rate": 9.927209918004095e-06,
      "loss": 1.432,
      "step": 538
    },
    {
      "epoch": 0.08313250689236346,
      "grad_norm": 0.4167533218860626,
      "learning_rate": 9.926784666110263e-06,
      "loss": 1.379,
      "step": 539
    },
    {
      "epoch": 0.08328674159902832,
      "grad_norm": 0.4424698054790497,
      "learning_rate": 9.926358184796843e-06,
      "loss": 1.3841,
      "step": 540
    },
    {
      "epoch": 0.08344097630569319,
      "grad_norm": 0.527334451675415,
      "learning_rate": 9.925930474170258e-06,
      "loss": 1.4124,
      "step": 541
    },
    {
      "epoch": 0.08359521101235806,
      "grad_norm": 0.458374947309494,
      "learning_rate": 9.925501534337241e-06,
      "loss": 1.3867,
      "step": 542
    },
    {
      "epoch": 0.08374944571902292,
      "grad_norm": 0.45460638403892517,
      "learning_rate": 9.925071365404826e-06,
      "loss": 1.3858,
      "step": 543
    },
    {
      "epoch": 0.08390368042568778,
      "grad_norm": 0.5177557468414307,
      "learning_rate": 9.924639967480358e-06,
      "loss": 1.3962,
      "step": 544
    },
    {
      "epoch": 0.08405791513235265,
      "grad_norm": 0.47196143865585327,
      "learning_rate": 9.924207340671486e-06,
      "loss": 1.3993,
      "step": 545
    },
    {
      "epoch": 0.08421214983901752,
      "grad_norm": 0.4587273895740509,
      "learning_rate": 9.92377348508617e-06,
      "loss": 1.3671,
      "step": 546
    },
    {
      "epoch": 0.08436638454568239,
      "grad_norm": 0.5247725248336792,
      "learning_rate": 9.923338400832668e-06,
      "loss": 1.4056,
      "step": 547
    },
    {
      "epoch": 0.08452061925234726,
      "grad_norm": 0.436396062374115,
      "learning_rate": 9.922902088019552e-06,
      "loss": 1.412,
      "step": 548
    },
    {
      "epoch": 0.08467485395901213,
      "grad_norm": 0.5055376887321472,
      "learning_rate": 9.922464546755702e-06,
      "loss": 1.431,
      "step": 549
    },
    {
      "epoch": 0.084829088665677,
      "grad_norm": 0.4799085259437561,
      "learning_rate": 9.922025777150299e-06,
      "loss": 1.402,
      "step": 550
    },
    {
      "epoch": 0.08498332337234187,
      "grad_norm": 0.46313726902008057,
      "learning_rate": 9.92158577931283e-06,
      "loss": 1.4204,
      "step": 551
    },
    {
      "epoch": 0.08513755807900673,
      "grad_norm": 0.4481837749481201,
      "learning_rate": 9.921144553353095e-06,
      "loss": 1.3986,
      "step": 552
    },
    {
      "epoch": 0.0852917927856716,
      "grad_norm": 0.4738650619983673,
      "learning_rate": 9.920702099381193e-06,
      "loss": 1.4068,
      "step": 553
    },
    {
      "epoch": 0.08544602749233646,
      "grad_norm": 0.44733938574790955,
      "learning_rate": 9.920258417507538e-06,
      "loss": 1.4144,
      "step": 554
    },
    {
      "epoch": 0.08560026219900133,
      "grad_norm": 0.4281189739704132,
      "learning_rate": 9.919813507842842e-06,
      "loss": 1.3828,
      "step": 555
    },
    {
      "epoch": 0.0857544969056662,
      "grad_norm": 0.4548930823802948,
      "learning_rate": 9.919367370498125e-06,
      "loss": 1.4117,
      "step": 556
    },
    {
      "epoch": 0.08590873161233106,
      "grad_norm": 0.4846823513507843,
      "learning_rate": 9.91892000558472e-06,
      "loss": 1.4223,
      "step": 557
    },
    {
      "epoch": 0.08606296631899593,
      "grad_norm": 0.42776811122894287,
      "learning_rate": 9.918471413214257e-06,
      "loss": 1.358,
      "step": 558
    },
    {
      "epoch": 0.0862172010256608,
      "grad_norm": 0.4271652102470398,
      "learning_rate": 9.91802159349868e-06,
      "loss": 1.3906,
      "step": 559
    },
    {
      "epoch": 0.08637143573232567,
      "grad_norm": 0.4573461413383484,
      "learning_rate": 9.917570546550235e-06,
      "loss": 1.4093,
      "step": 560
    },
    {
      "epoch": 0.08652567043899054,
      "grad_norm": 0.4535485506057739,
      "learning_rate": 9.917118272481474e-06,
      "loss": 1.3617,
      "step": 561
    },
    {
      "epoch": 0.0866799051456554,
      "grad_norm": 0.42113131284713745,
      "learning_rate": 9.91666477140526e-06,
      "loss": 1.4072,
      "step": 562
    },
    {
      "epoch": 0.08683413985232027,
      "grad_norm": 0.4826679527759552,
      "learning_rate": 9.916210043434753e-06,
      "loss": 1.3735,
      "step": 563
    },
    {
      "epoch": 0.08698837455898513,
      "grad_norm": 0.4485046863555908,
      "learning_rate": 9.915754088683431e-06,
      "loss": 1.4153,
      "step": 564
    },
    {
      "epoch": 0.08714260926565,
      "grad_norm": 0.47439292073249817,
      "learning_rate": 9.915296907265068e-06,
      "loss": 1.3989,
      "step": 565
    },
    {
      "epoch": 0.08729684397231487,
      "grad_norm": 0.43979352712631226,
      "learning_rate": 9.914838499293751e-06,
      "loss": 1.3966,
      "step": 566
    },
    {
      "epoch": 0.08745107867897974,
      "grad_norm": 0.46793246269226074,
      "learning_rate": 9.914378864883866e-06,
      "loss": 1.4114,
      "step": 567
    },
    {
      "epoch": 0.0876053133856446,
      "grad_norm": 0.41661736369132996,
      "learning_rate": 9.913918004150115e-06,
      "loss": 1.3494,
      "step": 568
    },
    {
      "epoch": 0.08775954809230947,
      "grad_norm": 0.4547062814235687,
      "learning_rate": 9.913455917207495e-06,
      "loss": 1.3858,
      "step": 569
    },
    {
      "epoch": 0.08791378279897434,
      "grad_norm": 0.45024025440216064,
      "learning_rate": 9.912992604171318e-06,
      "loss": 1.4006,
      "step": 570
    },
    {
      "epoch": 0.08806801750563921,
      "grad_norm": 0.4168526530265808,
      "learning_rate": 9.912528065157195e-06,
      "loss": 1.3758,
      "step": 571
    },
    {
      "epoch": 0.08822225221230408,
      "grad_norm": 0.5073679089546204,
      "learning_rate": 9.912062300281046e-06,
      "loss": 1.3924,
      "step": 572
    },
    {
      "epoch": 0.08837648691896895,
      "grad_norm": 0.4602091610431671,
      "learning_rate": 9.9115953096591e-06,
      "loss": 1.3909,
      "step": 573
    },
    {
      "epoch": 0.0885307216256338,
      "grad_norm": 0.4300541877746582,
      "learning_rate": 9.91112709340789e-06,
      "loss": 1.3835,
      "step": 574
    },
    {
      "epoch": 0.08868495633229867,
      "grad_norm": 0.4745858609676361,
      "learning_rate": 9.91065765164425e-06,
      "loss": 1.4162,
      "step": 575
    },
    {
      "epoch": 0.08883919103896354,
      "grad_norm": 0.4340323805809021,
      "learning_rate": 9.910186984485321e-06,
      "loss": 1.4063,
      "step": 576
    },
    {
      "epoch": 0.08899342574562841,
      "grad_norm": 0.4576251804828644,
      "learning_rate": 9.90971509204856e-06,
      "loss": 1.4024,
      "step": 577
    },
    {
      "epoch": 0.08914766045229328,
      "grad_norm": 0.4342866539955139,
      "learning_rate": 9.909241974451716e-06,
      "loss": 1.3802,
      "step": 578
    },
    {
      "epoch": 0.08930189515895814,
      "grad_norm": 0.42924508452415466,
      "learning_rate": 9.908767631812852e-06,
      "loss": 1.4092,
      "step": 579
    },
    {
      "epoch": 0.08945612986562301,
      "grad_norm": 0.46538642048835754,
      "learning_rate": 9.908292064250333e-06,
      "loss": 1.4128,
      "step": 580
    },
    {
      "epoch": 0.08961036457228788,
      "grad_norm": 0.4348650574684143,
      "learning_rate": 9.907815271882834e-06,
      "loss": 1.3565,
      "step": 581
    },
    {
      "epoch": 0.08976459927895275,
      "grad_norm": 0.43505600094795227,
      "learning_rate": 9.90733725482933e-06,
      "loss": 1.4258,
      "step": 582
    },
    {
      "epoch": 0.08991883398561762,
      "grad_norm": 0.4212532341480255,
      "learning_rate": 9.906858013209108e-06,
      "loss": 1.3833,
      "step": 583
    },
    {
      "epoch": 0.09007306869228249,
      "grad_norm": 0.43133947253227234,
      "learning_rate": 9.906377547141751e-06,
      "loss": 1.3456,
      "step": 584
    },
    {
      "epoch": 0.09022730339894734,
      "grad_norm": 0.45998647809028625,
      "learning_rate": 9.90589585674716e-06,
      "loss": 1.4233,
      "step": 585
    },
    {
      "epoch": 0.09038153810561221,
      "grad_norm": 0.4433024227619171,
      "learning_rate": 9.905412942145531e-06,
      "loss": 1.3847,
      "step": 586
    },
    {
      "epoch": 0.09053577281227708,
      "grad_norm": 0.45453864336013794,
      "learning_rate": 9.90492880345737e-06,
      "loss": 1.3716,
      "step": 587
    },
    {
      "epoch": 0.09069000751894195,
      "grad_norm": 0.4899587333202362,
      "learning_rate": 9.904443440803488e-06,
      "loss": 1.393,
      "step": 588
    },
    {
      "epoch": 0.09084424222560682,
      "grad_norm": 0.42056283354759216,
      "learning_rate": 9.903956854305003e-06,
      "loss": 1.3851,
      "step": 589
    },
    {
      "epoch": 0.09099847693227169,
      "grad_norm": 0.45995429158210754,
      "learning_rate": 9.903469044083336e-06,
      "loss": 1.3902,
      "step": 590
    },
    {
      "epoch": 0.09115271163893655,
      "grad_norm": 0.4808293581008911,
      "learning_rate": 9.902980010260212e-06,
      "loss": 1.4228,
      "step": 591
    },
    {
      "epoch": 0.09130694634560142,
      "grad_norm": 0.40312135219573975,
      "learning_rate": 9.902489752957667e-06,
      "loss": 1.3703,
      "step": 592
    },
    {
      "epoch": 0.09146118105226629,
      "grad_norm": 0.45251020789146423,
      "learning_rate": 9.901998272298037e-06,
      "loss": 1.3894,
      "step": 593
    },
    {
      "epoch": 0.09161541575893116,
      "grad_norm": 0.48815882205963135,
      "learning_rate": 9.901505568403963e-06,
      "loss": 1.4019,
      "step": 594
    },
    {
      "epoch": 0.09176965046559601,
      "grad_norm": 0.4210294485092163,
      "learning_rate": 9.901011641398398e-06,
      "loss": 1.4528,
      "step": 595
    },
    {
      "epoch": 0.09192388517226088,
      "grad_norm": 0.4283800423145294,
      "learning_rate": 9.900516491404592e-06,
      "loss": 1.3665,
      "step": 596
    },
    {
      "epoch": 0.09207811987892575,
      "grad_norm": 0.5013861060142517,
      "learning_rate": 9.900020118546105e-06,
      "loss": 1.3941,
      "step": 597
    },
    {
      "epoch": 0.09223235458559062,
      "grad_norm": 0.48357877135276794,
      "learning_rate": 9.899522522946799e-06,
      "loss": 1.4081,
      "step": 598
    },
    {
      "epoch": 0.09238658929225549,
      "grad_norm": 0.46260568499565125,
      "learning_rate": 9.899023704730845e-06,
      "loss": 1.3829,
      "step": 599
    },
    {
      "epoch": 0.09254082399892036,
      "grad_norm": 0.4724692702293396,
      "learning_rate": 9.898523664022717e-06,
      "loss": 1.3559,
      "step": 600
    },
    {
      "epoch": 0.09269505870558523,
      "grad_norm": 0.45046794414520264,
      "learning_rate": 9.898022400947194e-06,
      "loss": 1.4507,
      "step": 601
    },
    {
      "epoch": 0.0928492934122501,
      "grad_norm": 0.4409693479537964,
      "learning_rate": 9.897519915629358e-06,
      "loss": 1.3446,
      "step": 602
    },
    {
      "epoch": 0.09300352811891496,
      "grad_norm": 0.4491008222103119,
      "learning_rate": 9.897016208194604e-06,
      "loss": 1.3471,
      "step": 603
    },
    {
      "epoch": 0.09315776282557983,
      "grad_norm": 0.430292546749115,
      "learning_rate": 9.896511278768619e-06,
      "loss": 1.4139,
      "step": 604
    },
    {
      "epoch": 0.09331199753224469,
      "grad_norm": 0.4214726984500885,
      "learning_rate": 9.896005127477403e-06,
      "loss": 1.3987,
      "step": 605
    },
    {
      "epoch": 0.09346623223890956,
      "grad_norm": 0.4535372853279114,
      "learning_rate": 9.895497754447266e-06,
      "loss": 1.3829,
      "step": 606
    },
    {
      "epoch": 0.09362046694557442,
      "grad_norm": 0.4810298979282379,
      "learning_rate": 9.894989159804811e-06,
      "loss": 1.4361,
      "step": 607
    },
    {
      "epoch": 0.09377470165223929,
      "grad_norm": 0.4526076912879944,
      "learning_rate": 9.894479343676953e-06,
      "loss": 1.3716,
      "step": 608
    },
    {
      "epoch": 0.09392893635890416,
      "grad_norm": 0.47583112120628357,
      "learning_rate": 9.89396830619091e-06,
      "loss": 1.4015,
      "step": 609
    },
    {
      "epoch": 0.09408317106556903,
      "grad_norm": 0.4426989257335663,
      "learning_rate": 9.893456047474209e-06,
      "loss": 1.4258,
      "step": 610
    },
    {
      "epoch": 0.0942374057722339,
      "grad_norm": 0.44269248843193054,
      "learning_rate": 9.892942567654672e-06,
      "loss": 1.4032,
      "step": 611
    },
    {
      "epoch": 0.09439164047889877,
      "grad_norm": 0.4720871150493622,
      "learning_rate": 9.892427866860436e-06,
      "loss": 1.349,
      "step": 612
    },
    {
      "epoch": 0.09454587518556364,
      "grad_norm": 0.4475248456001282,
      "learning_rate": 9.891911945219939e-06,
      "loss": 1.4319,
      "step": 613
    },
    {
      "epoch": 0.0947001098922285,
      "grad_norm": 0.42031311988830566,
      "learning_rate": 9.89139480286192e-06,
      "loss": 1.3978,
      "step": 614
    },
    {
      "epoch": 0.09485434459889337,
      "grad_norm": 0.4473288953304291,
      "learning_rate": 9.890876439915426e-06,
      "loss": 1.372,
      "step": 615
    },
    {
      "epoch": 0.09500857930555823,
      "grad_norm": 0.47153615951538086,
      "learning_rate": 9.89035685650981e-06,
      "loss": 1.4072,
      "step": 616
    },
    {
      "epoch": 0.0951628140122231,
      "grad_norm": 0.43978315591812134,
      "learning_rate": 9.889836052774726e-06,
      "loss": 1.373,
      "step": 617
    },
    {
      "epoch": 0.09531704871888796,
      "grad_norm": 0.47577616572380066,
      "learning_rate": 9.889314028840136e-06,
      "loss": 1.3831,
      "step": 618
    },
    {
      "epoch": 0.09547128342555283,
      "grad_norm": 0.49060899019241333,
      "learning_rate": 9.888790784836302e-06,
      "loss": 1.379,
      "step": 619
    },
    {
      "epoch": 0.0956255181322177,
      "grad_norm": 0.45542556047439575,
      "learning_rate": 9.888266320893797e-06,
      "loss": 1.3848,
      "step": 620
    },
    {
      "epoch": 0.09577975283888257,
      "grad_norm": 0.42567890882492065,
      "learning_rate": 9.887740637143492e-06,
      "loss": 1.3765,
      "step": 621
    },
    {
      "epoch": 0.09593398754554744,
      "grad_norm": 0.46832770109176636,
      "learning_rate": 9.887213733716566e-06,
      "loss": 1.3881,
      "step": 622
    },
    {
      "epoch": 0.09608822225221231,
      "grad_norm": 0.4309346675872803,
      "learning_rate": 9.8866856107445e-06,
      "loss": 1.4073,
      "step": 623
    },
    {
      "epoch": 0.09624245695887718,
      "grad_norm": 0.46827831864356995,
      "learning_rate": 9.886156268359082e-06,
      "loss": 1.407,
      "step": 624
    },
    {
      "epoch": 0.09639669166554204,
      "grad_norm": 0.4217469394207001,
      "learning_rate": 9.885625706692403e-06,
      "loss": 1.3838,
      "step": 625
    },
    {
      "epoch": 0.0965509263722069,
      "grad_norm": 0.48413199186325073,
      "learning_rate": 9.885093925876858e-06,
      "loss": 1.4183,
      "step": 626
    },
    {
      "epoch": 0.09670516107887177,
      "grad_norm": 0.464229553937912,
      "learning_rate": 9.884560926045149e-06,
      "loss": 1.3891,
      "step": 627
    },
    {
      "epoch": 0.09685939578553664,
      "grad_norm": 0.4711248576641083,
      "learning_rate": 9.884026707330274e-06,
      "loss": 1.4003,
      "step": 628
    },
    {
      "epoch": 0.0970136304922015,
      "grad_norm": 0.4417979121208191,
      "learning_rate": 9.883491269865544e-06,
      "loss": 1.3903,
      "step": 629
    },
    {
      "epoch": 0.09716786519886637,
      "grad_norm": 0.424405038356781,
      "learning_rate": 9.882954613784572e-06,
      "loss": 1.3815,
      "step": 630
    },
    {
      "epoch": 0.09732209990553124,
      "grad_norm": 0.43959248065948486,
      "learning_rate": 9.882416739221273e-06,
      "loss": 1.3575,
      "step": 631
    },
    {
      "epoch": 0.09747633461219611,
      "grad_norm": 0.45256170630455017,
      "learning_rate": 9.881877646309867e-06,
      "loss": 1.4527,
      "step": 632
    },
    {
      "epoch": 0.09763056931886098,
      "grad_norm": 0.4756705164909363,
      "learning_rate": 9.881337335184879e-06,
      "loss": 1.3538,
      "step": 633
    },
    {
      "epoch": 0.09778480402552585,
      "grad_norm": 0.4799821376800537,
      "learning_rate": 9.880795805981136e-06,
      "loss": 1.3892,
      "step": 634
    },
    {
      "epoch": 0.09793903873219072,
      "grad_norm": 0.4668935537338257,
      "learning_rate": 9.880253058833769e-06,
      "loss": 1.3748,
      "step": 635
    },
    {
      "epoch": 0.09809327343885557,
      "grad_norm": 0.41657018661499023,
      "learning_rate": 9.879709093878218e-06,
      "loss": 1.3714,
      "step": 636
    },
    {
      "epoch": 0.09824750814552044,
      "grad_norm": 0.4139534533023834,
      "learning_rate": 9.879163911250217e-06,
      "loss": 1.348,
      "step": 637
    },
    {
      "epoch": 0.09840174285218531,
      "grad_norm": 0.4281579852104187,
      "learning_rate": 9.878617511085815e-06,
      "loss": 1.3785,
      "step": 638
    },
    {
      "epoch": 0.09855597755885018,
      "grad_norm": 0.4374670684337616,
      "learning_rate": 9.878069893521358e-06,
      "loss": 1.4011,
      "step": 639
    },
    {
      "epoch": 0.09871021226551505,
      "grad_norm": 0.45784109830856323,
      "learning_rate": 9.877521058693495e-06,
      "loss": 1.4022,
      "step": 640
    },
    {
      "epoch": 0.09886444697217991,
      "grad_norm": 0.4866529107093811,
      "learning_rate": 9.876971006739185e-06,
      "loss": 1.3893,
      "step": 641
    },
    {
      "epoch": 0.09901868167884478,
      "grad_norm": 0.42958325147628784,
      "learning_rate": 9.876419737795683e-06,
      "loss": 1.4028,
      "step": 642
    },
    {
      "epoch": 0.09917291638550965,
      "grad_norm": 0.45041510462760925,
      "learning_rate": 9.875867252000555e-06,
      "loss": 1.3597,
      "step": 643
    },
    {
      "epoch": 0.09932715109217452,
      "grad_norm": 0.4576846659183502,
      "learning_rate": 9.875313549491665e-06,
      "loss": 1.3534,
      "step": 644
    },
    {
      "epoch": 0.09948138579883939,
      "grad_norm": 0.44920751452445984,
      "learning_rate": 9.874758630407184e-06,
      "loss": 1.4016,
      "step": 645
    },
    {
      "epoch": 0.09963562050550426,
      "grad_norm": 0.4399208724498749,
      "learning_rate": 9.874202494885582e-06,
      "loss": 1.3635,
      "step": 646
    },
    {
      "epoch": 0.09978985521216911,
      "grad_norm": 0.44910728931427,
      "learning_rate": 9.873645143065641e-06,
      "loss": 1.4061,
      "step": 647
    },
    {
      "epoch": 0.09994408991883398,
      "grad_norm": 0.44466012716293335,
      "learning_rate": 9.873086575086439e-06,
      "loss": 1.3774,
      "step": 648
    },
    {
      "epoch": 0.10009832462549885,
      "grad_norm": 0.46906405687332153,
      "learning_rate": 9.87252679108736e-06,
      "loss": 1.3687,
      "step": 649
    },
    {
      "epoch": 0.10025255933216372,
      "grad_norm": 0.49215084314346313,
      "learning_rate": 9.87196579120809e-06,
      "loss": 1.4054,
      "step": 650
    },
    {
      "epoch": 0.10040679403882859,
      "grad_norm": 0.4284855127334595,
      "learning_rate": 9.87140357558862e-06,
      "loss": 1.4053,
      "step": 651
    },
    {
      "epoch": 0.10056102874549346,
      "grad_norm": 0.4497528672218323,
      "learning_rate": 9.870840144369247e-06,
      "loss": 1.346,
      "step": 652
    },
    {
      "epoch": 0.10071526345215832,
      "grad_norm": 0.4088262915611267,
      "learning_rate": 9.870275497690565e-06,
      "loss": 1.3576,
      "step": 653
    },
    {
      "epoch": 0.10086949815882319,
      "grad_norm": 0.44848471879959106,
      "learning_rate": 9.869709635693478e-06,
      "loss": 1.4066,
      "step": 654
    },
    {
      "epoch": 0.10102373286548806,
      "grad_norm": 0.4360213577747345,
      "learning_rate": 9.869142558519187e-06,
      "loss": 1.3374,
      "step": 655
    },
    {
      "epoch": 0.10117796757215293,
      "grad_norm": 0.427133709192276,
      "learning_rate": 9.868574266309201e-06,
      "loss": 1.3747,
      "step": 656
    },
    {
      "epoch": 0.10133220227881778,
      "grad_norm": 0.45959198474884033,
      "learning_rate": 9.86800475920533e-06,
      "loss": 1.368,
      "step": 657
    },
    {
      "epoch": 0.10148643698548265,
      "grad_norm": 0.4175659120082855,
      "learning_rate": 9.867434037349685e-06,
      "loss": 1.4017,
      "step": 658
    },
    {
      "epoch": 0.10164067169214752,
      "grad_norm": 0.44435247778892517,
      "learning_rate": 9.866862100884688e-06,
      "loss": 1.3568,
      "step": 659
    },
    {
      "epoch": 0.10179490639881239,
      "grad_norm": 0.44337543845176697,
      "learning_rate": 9.866288949953057e-06,
      "loss": 1.4203,
      "step": 660
    },
    {
      "epoch": 0.10194914110547726,
      "grad_norm": 0.4260760247707367,
      "learning_rate": 9.865714584697812e-06,
      "loss": 1.3508,
      "step": 661
    },
    {
      "epoch": 0.10210337581214213,
      "grad_norm": 0.4754478931427002,
      "learning_rate": 9.865139005262283e-06,
      "loss": 1.3922,
      "step": 662
    },
    {
      "epoch": 0.102257610518807,
      "grad_norm": 0.45311057567596436,
      "learning_rate": 9.864562211790095e-06,
      "loss": 1.395,
      "step": 663
    },
    {
      "epoch": 0.10241184522547186,
      "grad_norm": 0.42213669419288635,
      "learning_rate": 9.863984204425183e-06,
      "loss": 1.3314,
      "step": 664
    },
    {
      "epoch": 0.10256607993213673,
      "grad_norm": 0.4258612096309662,
      "learning_rate": 9.863404983311779e-06,
      "loss": 1.3581,
      "step": 665
    },
    {
      "epoch": 0.1027203146388016,
      "grad_norm": 0.45202726125717163,
      "learning_rate": 9.862824548594423e-06,
      "loss": 1.3909,
      "step": 666
    },
    {
      "epoch": 0.10287454934546646,
      "grad_norm": 0.42718157172203064,
      "learning_rate": 9.862242900417954e-06,
      "loss": 1.3761,
      "step": 667
    },
    {
      "epoch": 0.10302878405213133,
      "grad_norm": 0.42724889516830444,
      "learning_rate": 9.861660038927516e-06,
      "loss": 1.3607,
      "step": 668
    },
    {
      "epoch": 0.1031830187587962,
      "grad_norm": 0.4834701716899872,
      "learning_rate": 9.861075964268556e-06,
      "loss": 1.382,
      "step": 669
    },
    {
      "epoch": 0.10333725346546106,
      "grad_norm": 0.43723833560943604,
      "learning_rate": 9.86049067658682e-06,
      "loss": 1.3925,
      "step": 670
    },
    {
      "epoch": 0.10349148817212593,
      "grad_norm": 0.43868571519851685,
      "learning_rate": 9.859904176028364e-06,
      "loss": 1.3673,
      "step": 671
    },
    {
      "epoch": 0.1036457228787908,
      "grad_norm": 0.5013939142227173,
      "learning_rate": 9.859316462739536e-06,
      "loss": 1.4017,
      "step": 672
    },
    {
      "epoch": 0.10379995758545567,
      "grad_norm": 0.4373146891593933,
      "learning_rate": 9.858727536866997e-06,
      "loss": 1.3865,
      "step": 673
    },
    {
      "epoch": 0.10395419229212054,
      "grad_norm": 0.45148810744285583,
      "learning_rate": 9.858137398557705e-06,
      "loss": 1.3455,
      "step": 674
    },
    {
      "epoch": 0.1041084269987854,
      "grad_norm": 0.47480836510658264,
      "learning_rate": 9.857546047958923e-06,
      "loss": 1.4015,
      "step": 675
    },
    {
      "epoch": 0.10426266170545027,
      "grad_norm": 0.42332684993743896,
      "learning_rate": 9.856953485218215e-06,
      "loss": 1.3788,
      "step": 676
    },
    {
      "epoch": 0.10441689641211514,
      "grad_norm": 0.4758417308330536,
      "learning_rate": 9.856359710483446e-06,
      "loss": 1.3571,
      "step": 677
    },
    {
      "epoch": 0.10457113111878,
      "grad_norm": 0.4168352484703064,
      "learning_rate": 9.855764723902788e-06,
      "loss": 1.3588,
      "step": 678
    },
    {
      "epoch": 0.10472536582544487,
      "grad_norm": 0.4867233633995056,
      "learning_rate": 9.85516852562471e-06,
      "loss": 1.4212,
      "step": 679
    },
    {
      "epoch": 0.10487960053210973,
      "grad_norm": 0.43304023146629333,
      "learning_rate": 9.854571115797989e-06,
      "loss": 1.4159,
      "step": 680
    },
    {
      "epoch": 0.1050338352387746,
      "grad_norm": 0.4678906202316284,
      "learning_rate": 9.8539724945717e-06,
      "loss": 1.381,
      "step": 681
    },
    {
      "epoch": 0.10518806994543947,
      "grad_norm": 0.4683140516281128,
      "learning_rate": 9.853372662095222e-06,
      "loss": 1.394,
      "step": 682
    },
    {
      "epoch": 0.10534230465210434,
      "grad_norm": 0.4631720185279846,
      "learning_rate": 9.852771618518234e-06,
      "loss": 1.3296,
      "step": 683
    },
    {
      "epoch": 0.10549653935876921,
      "grad_norm": 0.43896958231925964,
      "learning_rate": 9.852169363990722e-06,
      "loss": 1.3613,
      "step": 684
    },
    {
      "epoch": 0.10565077406543408,
      "grad_norm": 0.4659942090511322,
      "learning_rate": 9.85156589866297e-06,
      "loss": 1.3918,
      "step": 685
    },
    {
      "epoch": 0.10580500877209895,
      "grad_norm": 0.44510629773139954,
      "learning_rate": 9.850961222685566e-06,
      "loss": 1.3554,
      "step": 686
    },
    {
      "epoch": 0.10595924347876381,
      "grad_norm": 0.43709951639175415,
      "learning_rate": 9.850355336209398e-06,
      "loss": 1.3751,
      "step": 687
    },
    {
      "epoch": 0.10611347818542867,
      "grad_norm": 0.4969240725040436,
      "learning_rate": 9.849748239385661e-06,
      "loss": 1.4108,
      "step": 688
    },
    {
      "epoch": 0.10626771289209354,
      "grad_norm": 0.44518589973449707,
      "learning_rate": 9.849139932365844e-06,
      "loss": 1.356,
      "step": 689
    },
    {
      "epoch": 0.1064219475987584,
      "grad_norm": 0.4284765422344208,
      "learning_rate": 9.848530415301748e-06,
      "loss": 1.3381,
      "step": 690
    },
    {
      "epoch": 0.10657618230542328,
      "grad_norm": 0.48101645708084106,
      "learning_rate": 9.847919688345465e-06,
      "loss": 1.3908,
      "step": 691
    },
    {
      "epoch": 0.10673041701208814,
      "grad_norm": 0.41678643226623535,
      "learning_rate": 9.847307751649398e-06,
      "loss": 1.3842,
      "step": 692
    },
    {
      "epoch": 0.10688465171875301,
      "grad_norm": 0.5147459506988525,
      "learning_rate": 9.846694605366248e-06,
      "loss": 1.3385,
      "step": 693
    },
    {
      "epoch": 0.10703888642541788,
      "grad_norm": 0.5198816657066345,
      "learning_rate": 9.846080249649018e-06,
      "loss": 1.3991,
      "step": 694
    },
    {
      "epoch": 0.10719312113208275,
      "grad_norm": 0.42182374000549316,
      "learning_rate": 9.845464684651015e-06,
      "loss": 1.3687,
      "step": 695
    },
    {
      "epoch": 0.10734735583874762,
      "grad_norm": 0.46364015340805054,
      "learning_rate": 9.84484791052584e-06,
      "loss": 1.3715,
      "step": 696
    },
    {
      "epoch": 0.10750159054541249,
      "grad_norm": 0.47631120681762695,
      "learning_rate": 9.844229927427408e-06,
      "loss": 1.4032,
      "step": 697
    },
    {
      "epoch": 0.10765582525207734,
      "grad_norm": 0.41958925127983093,
      "learning_rate": 9.843610735509927e-06,
      "loss": 1.3897,
      "step": 698
    },
    {
      "epoch": 0.10781005995874221,
      "grad_norm": 0.44546833634376526,
      "learning_rate": 9.842990334927906e-06,
      "loss": 1.3419,
      "step": 699
    },
    {
      "epoch": 0.10796429466540708,
      "grad_norm": 0.4467076063156128,
      "learning_rate": 9.842368725836163e-06,
      "loss": 1.3475,
      "step": 700
    },
    {
      "epoch": 0.10811852937207195,
      "grad_norm": 0.43269142508506775,
      "learning_rate": 9.84174590838981e-06,
      "loss": 1.3879,
      "step": 701
    },
    {
      "epoch": 0.10827276407873682,
      "grad_norm": 0.4104573130607605,
      "learning_rate": 9.841121882744267e-06,
      "loss": 1.3569,
      "step": 702
    },
    {
      "epoch": 0.10842699878540168,
      "grad_norm": 0.44374406337738037,
      "learning_rate": 9.840496649055246e-06,
      "loss": 1.3612,
      "step": 703
    },
    {
      "epoch": 0.10858123349206655,
      "grad_norm": 0.44906085729599,
      "learning_rate": 9.839870207478772e-06,
      "loss": 1.3744,
      "step": 704
    },
    {
      "epoch": 0.10873546819873142,
      "grad_norm": 0.43347328901290894,
      "learning_rate": 9.839242558171165e-06,
      "loss": 1.3777,
      "step": 705
    },
    {
      "epoch": 0.10888970290539629,
      "grad_norm": 0.3885936737060547,
      "learning_rate": 9.838613701289043e-06,
      "loss": 1.3612,
      "step": 706
    },
    {
      "epoch": 0.10904393761206116,
      "grad_norm": 0.43079331517219543,
      "learning_rate": 9.837983636989337e-06,
      "loss": 1.429,
      "step": 707
    },
    {
      "epoch": 0.10919817231872603,
      "grad_norm": 0.4597599506378174,
      "learning_rate": 9.837352365429265e-06,
      "loss": 1.3698,
      "step": 708
    },
    {
      "epoch": 0.10935240702539088,
      "grad_norm": 0.4553662836551666,
      "learning_rate": 9.836719886766357e-06,
      "loss": 1.3528,
      "step": 709
    },
    {
      "epoch": 0.10950664173205575,
      "grad_norm": 0.5038666129112244,
      "learning_rate": 9.83608620115844e-06,
      "loss": 1.3793,
      "step": 710
    },
    {
      "epoch": 0.10966087643872062,
      "grad_norm": 0.44157353043556213,
      "learning_rate": 9.835451308763642e-06,
      "loss": 1.3579,
      "step": 711
    },
    {
      "epoch": 0.10981511114538549,
      "grad_norm": 0.43951690196990967,
      "learning_rate": 9.834815209740393e-06,
      "loss": 1.3518,
      "step": 712
    },
    {
      "epoch": 0.10996934585205036,
      "grad_norm": 0.4976160526275635,
      "learning_rate": 9.834177904247423e-06,
      "loss": 1.3759,
      "step": 713
    },
    {
      "epoch": 0.11012358055871523,
      "grad_norm": 0.4879795014858246,
      "learning_rate": 9.833539392443764e-06,
      "loss": 1.3789,
      "step": 714
    },
    {
      "epoch": 0.1102778152653801,
      "grad_norm": 0.40307706594467163,
      "learning_rate": 9.83289967448875e-06,
      "loss": 1.3235,
      "step": 715
    },
    {
      "epoch": 0.11043204997204496,
      "grad_norm": 0.4686153531074524,
      "learning_rate": 9.832258750542016e-06,
      "loss": 1.3707,
      "step": 716
    },
    {
      "epoch": 0.11058628467870983,
      "grad_norm": 0.4298979938030243,
      "learning_rate": 9.831616620763494e-06,
      "loss": 1.3485,
      "step": 717
    },
    {
      "epoch": 0.1107405193853747,
      "grad_norm": 0.43293291330337524,
      "learning_rate": 9.830973285313418e-06,
      "loss": 1.3612,
      "step": 718
    },
    {
      "epoch": 0.11089475409203955,
      "grad_norm": 0.5468214154243469,
      "learning_rate": 9.830328744352332e-06,
      "loss": 1.4148,
      "step": 719
    },
    {
      "epoch": 0.11104898879870442,
      "grad_norm": 0.46644720435142517,
      "learning_rate": 9.829682998041065e-06,
      "loss": 1.3908,
      "step": 720
    },
    {
      "epoch": 0.11120322350536929,
      "grad_norm": 0.43414947390556335,
      "learning_rate": 9.829036046540763e-06,
      "loss": 1.387,
      "step": 721
    },
    {
      "epoch": 0.11135745821203416,
      "grad_norm": 0.5359362959861755,
      "learning_rate": 9.828387890012858e-06,
      "loss": 1.3777,
      "step": 722
    },
    {
      "epoch": 0.11151169291869903,
      "grad_norm": 0.43617022037506104,
      "learning_rate": 9.827738528619095e-06,
      "loss": 1.3629,
      "step": 723
    },
    {
      "epoch": 0.1116659276253639,
      "grad_norm": 0.42857152223587036,
      "learning_rate": 9.82708796252151e-06,
      "loss": 1.3716,
      "step": 724
    },
    {
      "epoch": 0.11182016233202877,
      "grad_norm": 0.45435619354248047,
      "learning_rate": 9.826436191882446e-06,
      "loss": 1.3584,
      "step": 725
    },
    {
      "epoch": 0.11197439703869363,
      "grad_norm": 0.42542126774787903,
      "learning_rate": 9.825783216864545e-06,
      "loss": 1.3684,
      "step": 726
    },
    {
      "epoch": 0.1121286317453585,
      "grad_norm": 0.4516334533691406,
      "learning_rate": 9.825129037630748e-06,
      "loss": 1.3659,
      "step": 727
    },
    {
      "epoch": 0.11228286645202337,
      "grad_norm": 0.4535215497016907,
      "learning_rate": 9.824473654344297e-06,
      "loss": 1.3876,
      "step": 728
    },
    {
      "epoch": 0.11243710115868824,
      "grad_norm": 0.4218862056732178,
      "learning_rate": 9.823817067168737e-06,
      "loss": 1.4005,
      "step": 729
    },
    {
      "epoch": 0.1125913358653531,
      "grad_norm": 0.45613449811935425,
      "learning_rate": 9.823159276267911e-06,
      "loss": 1.3582,
      "step": 730
    },
    {
      "epoch": 0.11274557057201796,
      "grad_norm": 0.446209579706192,
      "learning_rate": 9.82250028180596e-06,
      "loss": 1.3736,
      "step": 731
    },
    {
      "epoch": 0.11289980527868283,
      "grad_norm": 0.4332846701145172,
      "learning_rate": 9.821840083947332e-06,
      "loss": 1.3837,
      "step": 732
    },
    {
      "epoch": 0.1130540399853477,
      "grad_norm": 0.4187641739845276,
      "learning_rate": 9.82117868285677e-06,
      "loss": 1.3782,
      "step": 733
    },
    {
      "epoch": 0.11320827469201257,
      "grad_norm": 0.4384898543357849,
      "learning_rate": 9.820516078699316e-06,
      "loss": 1.343,
      "step": 734
    },
    {
      "epoch": 0.11336250939867744,
      "grad_norm": 0.4686693549156189,
      "learning_rate": 9.819852271640318e-06,
      "loss": 1.4175,
      "step": 735
    },
    {
      "epoch": 0.11351674410534231,
      "grad_norm": 0.4504725933074951,
      "learning_rate": 9.81918726184542e-06,
      "loss": 1.3828,
      "step": 736
    },
    {
      "epoch": 0.11367097881200718,
      "grad_norm": 0.43755751848220825,
      "learning_rate": 9.818521049480566e-06,
      "loss": 1.3831,
      "step": 737
    },
    {
      "epoch": 0.11382521351867204,
      "grad_norm": 0.4349918067455292,
      "learning_rate": 9.817853634712004e-06,
      "loss": 1.3678,
      "step": 738
    },
    {
      "epoch": 0.11397944822533691,
      "grad_norm": 0.4427676796913147,
      "learning_rate": 9.817185017706277e-06,
      "loss": 1.3417,
      "step": 739
    },
    {
      "epoch": 0.11413368293200177,
      "grad_norm": 0.4279189109802246,
      "learning_rate": 9.816515198630232e-06,
      "loss": 1.3511,
      "step": 740
    },
    {
      "epoch": 0.11428791763866664,
      "grad_norm": 0.48394179344177246,
      "learning_rate": 9.815844177651012e-06,
      "loss": 1.3928,
      "step": 741
    },
    {
      "epoch": 0.1144421523453315,
      "grad_norm": 0.42550933361053467,
      "learning_rate": 9.815171954936064e-06,
      "loss": 1.3732,
      "step": 742
    },
    {
      "epoch": 0.11459638705199637,
      "grad_norm": 0.4324493408203125,
      "learning_rate": 9.814498530653135e-06,
      "loss": 1.379,
      "step": 743
    },
    {
      "epoch": 0.11475062175866124,
      "grad_norm": 0.4873847961425781,
      "learning_rate": 9.813823904970264e-06,
      "loss": 1.3833,
      "step": 744
    },
    {
      "epoch": 0.11490485646532611,
      "grad_norm": 0.4243428111076355,
      "learning_rate": 9.813148078055801e-06,
      "loss": 1.3913,
      "step": 745
    },
    {
      "epoch": 0.11505909117199098,
      "grad_norm": 0.42793238162994385,
      "learning_rate": 9.812471050078389e-06,
      "loss": 1.3644,
      "step": 746
    },
    {
      "epoch": 0.11521332587865585,
      "grad_norm": 0.46352872252464294,
      "learning_rate": 9.81179282120697e-06,
      "loss": 1.3711,
      "step": 747
    },
    {
      "epoch": 0.11536756058532072,
      "grad_norm": 0.45879653096199036,
      "learning_rate": 9.81111339161079e-06,
      "loss": 1.4004,
      "step": 748
    },
    {
      "epoch": 0.11552179529198559,
      "grad_norm": 0.4373062551021576,
      "learning_rate": 9.810432761459392e-06,
      "loss": 1.3753,
      "step": 749
    },
    {
      "epoch": 0.11567602999865044,
      "grad_norm": 0.4547845423221588,
      "learning_rate": 9.80975093092262e-06,
      "loss": 1.3864,
      "step": 750
    },
    {
      "epoch": 0.11583026470531531,
      "grad_norm": 0.4024827480316162,
      "learning_rate": 9.809067900170614e-06,
      "loss": 1.3696,
      "step": 751
    },
    {
      "epoch": 0.11598449941198018,
      "grad_norm": 0.4349729120731354,
      "learning_rate": 9.80838366937382e-06,
      "loss": 1.4157,
      "step": 752
    },
    {
      "epoch": 0.11613873411864505,
      "grad_norm": 0.4341493844985962,
      "learning_rate": 9.807698238702975e-06,
      "loss": 1.3466,
      "step": 753
    },
    {
      "epoch": 0.11629296882530991,
      "grad_norm": 0.44768255949020386,
      "learning_rate": 9.807011608329121e-06,
      "loss": 1.3646,
      "step": 754
    },
    {
      "epoch": 0.11644720353197478,
      "grad_norm": 0.4326797127723694,
      "learning_rate": 9.806323778423603e-06,
      "loss": 1.3306,
      "step": 755
    },
    {
      "epoch": 0.11660143823863965,
      "grad_norm": 0.4152265191078186,
      "learning_rate": 9.805634749158056e-06,
      "loss": 1.3257,
      "step": 756
    },
    {
      "epoch": 0.11675567294530452,
      "grad_norm": 0.4453123211860657,
      "learning_rate": 9.804944520704421e-06,
      "loss": 1.4102,
      "step": 757
    },
    {
      "epoch": 0.11690990765196939,
      "grad_norm": 0.4206942319869995,
      "learning_rate": 9.804253093234933e-06,
      "loss": 1.3729,
      "step": 758
    },
    {
      "epoch": 0.11706414235863426,
      "grad_norm": 0.41999831795692444,
      "learning_rate": 9.803560466922131e-06,
      "loss": 1.3468,
      "step": 759
    },
    {
      "epoch": 0.11721837706529913,
      "grad_norm": 0.48270198702812195,
      "learning_rate": 9.802866641938853e-06,
      "loss": 1.3515,
      "step": 760
    },
    {
      "epoch": 0.11737261177196398,
      "grad_norm": 0.46071720123291016,
      "learning_rate": 9.802171618458235e-06,
      "loss": 1.3699,
      "step": 761
    },
    {
      "epoch": 0.11752684647862885,
      "grad_norm": 0.43003150820732117,
      "learning_rate": 9.80147539665371e-06,
      "loss": 1.336,
      "step": 762
    },
    {
      "epoch": 0.11768108118529372,
      "grad_norm": 0.43177399039268494,
      "learning_rate": 9.800777976699012e-06,
      "loss": 1.3794,
      "step": 763
    },
    {
      "epoch": 0.11783531589195859,
      "grad_norm": 0.42428192496299744,
      "learning_rate": 9.800079358768173e-06,
      "loss": 1.3696,
      "step": 764
    },
    {
      "epoch": 0.11798955059862345,
      "grad_norm": 0.42885127663612366,
      "learning_rate": 9.799379543035527e-06,
      "loss": 1.3453,
      "step": 765
    },
    {
      "epoch": 0.11814378530528832,
      "grad_norm": 0.4755268394947052,
      "learning_rate": 9.798678529675702e-06,
      "loss": 1.3988,
      "step": 766
    },
    {
      "epoch": 0.11829802001195319,
      "grad_norm": 0.4355374872684479,
      "learning_rate": 9.79797631886363e-06,
      "loss": 1.3498,
      "step": 767
    },
    {
      "epoch": 0.11845225471861806,
      "grad_norm": 0.4248248338699341,
      "learning_rate": 9.797272910774535e-06,
      "loss": 1.3342,
      "step": 768
    },
    {
      "epoch": 0.11860648942528293,
      "grad_norm": 0.46175631880760193,
      "learning_rate": 9.796568305583949e-06,
      "loss": 1.3667,
      "step": 769
    },
    {
      "epoch": 0.1187607241319478,
      "grad_norm": 0.4631972908973694,
      "learning_rate": 9.795862503467695e-06,
      "loss": 1.3815,
      "step": 770
    },
    {
      "epoch": 0.11891495883861265,
      "grad_norm": 0.4286893904209137,
      "learning_rate": 9.795155504601897e-06,
      "loss": 1.3581,
      "step": 771
    },
    {
      "epoch": 0.11906919354527752,
      "grad_norm": 0.5162618160247803,
      "learning_rate": 9.79444730916298e-06,
      "loss": 1.3318,
      "step": 772
    },
    {
      "epoch": 0.11922342825194239,
      "grad_norm": 0.433258056640625,
      "learning_rate": 9.793737917327665e-06,
      "loss": 1.4014,
      "step": 773
    },
    {
      "epoch": 0.11937766295860726,
      "grad_norm": 0.448296457529068,
      "learning_rate": 9.79302732927297e-06,
      "loss": 1.3916,
      "step": 774
    },
    {
      "epoch": 0.11953189766527213,
      "grad_norm": 0.40876999497413635,
      "learning_rate": 9.792315545176216e-06,
      "loss": 1.3495,
      "step": 775
    },
    {
      "epoch": 0.119686132371937,
      "grad_norm": 0.4315214157104492,
      "learning_rate": 9.79160256521502e-06,
      "loss": 1.4107,
      "step": 776
    },
    {
      "epoch": 0.11984036707860186,
      "grad_norm": 0.46493902802467346,
      "learning_rate": 9.790888389567297e-06,
      "loss": 1.35,
      "step": 777
    },
    {
      "epoch": 0.11999460178526673,
      "grad_norm": 0.4051251709461212,
      "learning_rate": 9.79017301841126e-06,
      "loss": 1.3896,
      "step": 778
    },
    {
      "epoch": 0.1201488364919316,
      "grad_norm": 0.4369288980960846,
      "learning_rate": 9.789456451925423e-06,
      "loss": 1.3605,
      "step": 779
    },
    {
      "epoch": 0.12030307119859647,
      "grad_norm": 0.4111925959587097,
      "learning_rate": 9.788738690288595e-06,
      "loss": 1.3671,
      "step": 780
    },
    {
      "epoch": 0.12045730590526132,
      "grad_norm": 0.44237005710601807,
      "learning_rate": 9.788019733679885e-06,
      "loss": 1.365,
      "step": 781
    },
    {
      "epoch": 0.1206115406119262,
      "grad_norm": 0.43273478746414185,
      "learning_rate": 9.787299582278702e-06,
      "loss": 1.38,
      "step": 782
    },
    {
      "epoch": 0.12076577531859106,
      "grad_norm": 0.42054733633995056,
      "learning_rate": 9.786578236264748e-06,
      "loss": 1.3513,
      "step": 783
    },
    {
      "epoch": 0.12092001002525593,
      "grad_norm": 0.41294705867767334,
      "learning_rate": 9.785855695818028e-06,
      "loss": 1.3609,
      "step": 784
    },
    {
      "epoch": 0.1210742447319208,
      "grad_norm": 0.4511352479457855,
      "learning_rate": 9.785131961118843e-06,
      "loss": 1.3624,
      "step": 785
    },
    {
      "epoch": 0.12122847943858567,
      "grad_norm": 0.4249388873577118,
      "learning_rate": 9.784407032347792e-06,
      "loss": 1.3466,
      "step": 786
    },
    {
      "epoch": 0.12138271414525054,
      "grad_norm": 0.4437543749809265,
      "learning_rate": 9.783680909685772e-06,
      "loss": 1.3322,
      "step": 787
    },
    {
      "epoch": 0.1215369488519154,
      "grad_norm": 0.44621041417121887,
      "learning_rate": 9.782953593313978e-06,
      "loss": 1.393,
      "step": 788
    },
    {
      "epoch": 0.12169118355858027,
      "grad_norm": 0.43378946185112,
      "learning_rate": 9.782225083413905e-06,
      "loss": 1.3732,
      "step": 789
    },
    {
      "epoch": 0.12184541826524514,
      "grad_norm": 0.4374745190143585,
      "learning_rate": 9.78149538016734e-06,
      "loss": 1.3789,
      "step": 790
    },
    {
      "epoch": 0.12199965297191001,
      "grad_norm": 0.48876145482063293,
      "learning_rate": 9.780764483756375e-06,
      "loss": 1.3626,
      "step": 791
    },
    {
      "epoch": 0.12215388767857487,
      "grad_norm": 0.44628390669822693,
      "learning_rate": 9.780032394363397e-06,
      "loss": 1.3866,
      "step": 792
    },
    {
      "epoch": 0.12230812238523973,
      "grad_norm": 0.43893060088157654,
      "learning_rate": 9.779299112171086e-06,
      "loss": 1.3582,
      "step": 793
    },
    {
      "epoch": 0.1224623570919046,
      "grad_norm": 0.6715783476829529,
      "learning_rate": 9.778564637362426e-06,
      "loss": 1.3684,
      "step": 794
    },
    {
      "epoch": 0.12261659179856947,
      "grad_norm": 0.3864818215370178,
      "learning_rate": 9.777828970120697e-06,
      "loss": 1.3793,
      "step": 795
    },
    {
      "epoch": 0.12277082650523434,
      "grad_norm": 0.39914947748184204,
      "learning_rate": 9.777092110629475e-06,
      "loss": 1.3372,
      "step": 796
    },
    {
      "epoch": 0.12292506121189921,
      "grad_norm": 0.4848926067352295,
      "learning_rate": 9.776354059072638e-06,
      "loss": 1.3822,
      "step": 797
    },
    {
      "epoch": 0.12307929591856408,
      "grad_norm": 0.42694732546806335,
      "learning_rate": 9.77561481563435e-06,
      "loss": 1.3719,
      "step": 798
    },
    {
      "epoch": 0.12323353062522895,
      "grad_norm": 0.4372820258140564,
      "learning_rate": 9.774874380499086e-06,
      "loss": 1.3837,
      "step": 799
    },
    {
      "epoch": 0.12338776533189381,
      "grad_norm": 0.45094388723373413,
      "learning_rate": 9.774132753851611e-06,
      "loss": 1.3531,
      "step": 800
    },
    {
      "epoch": 0.12354200003855868,
      "grad_norm": 0.42827868461608887,
      "learning_rate": 9.773389935876992e-06,
      "loss": 1.391,
      "step": 801
    },
    {
      "epoch": 0.12369623474522354,
      "grad_norm": 0.4116842448711395,
      "learning_rate": 9.772645926760584e-06,
      "loss": 1.3232,
      "step": 802
    },
    {
      "epoch": 0.1238504694518884,
      "grad_norm": 0.4244012236595154,
      "learning_rate": 9.77190072668805e-06,
      "loss": 1.3327,
      "step": 803
    },
    {
      "epoch": 0.12400470415855327,
      "grad_norm": 0.45345398783683777,
      "learning_rate": 9.771154335845345e-06,
      "loss": 1.371,
      "step": 804
    },
    {
      "epoch": 0.12415893886521814,
      "grad_norm": 0.4498525559902191,
      "learning_rate": 9.770406754418723e-06,
      "loss": 1.3164,
      "step": 805
    },
    {
      "epoch": 0.12431317357188301,
      "grad_norm": 0.4071389138698578,
      "learning_rate": 9.76965798259473e-06,
      "loss": 1.3267,
      "step": 806
    },
    {
      "epoch": 0.12446740827854788,
      "grad_norm": 0.4534646272659302,
      "learning_rate": 9.768908020560218e-06,
      "loss": 1.3653,
      "step": 807
    },
    {
      "epoch": 0.12462164298521275,
      "grad_norm": 0.41573333740234375,
      "learning_rate": 9.768156868502328e-06,
      "loss": 1.3693,
      "step": 808
    },
    {
      "epoch": 0.12477587769187762,
      "grad_norm": 0.42766252160072327,
      "learning_rate": 9.7674045266085e-06,
      "loss": 1.3123,
      "step": 809
    },
    {
      "epoch": 0.12493011239854249,
      "grad_norm": 0.4314340054988861,
      "learning_rate": 9.766650995066474e-06,
      "loss": 1.346,
      "step": 810
    },
    {
      "epoch": 0.12508434710520736,
      "grad_norm": 0.4385635554790497,
      "learning_rate": 9.765896274064283e-06,
      "loss": 1.3588,
      "step": 811
    },
    {
      "epoch": 0.12523858181187222,
      "grad_norm": 0.4253045320510864,
      "learning_rate": 9.76514036379026e-06,
      "loss": 1.3636,
      "step": 812
    },
    {
      "epoch": 0.1253928165185371,
      "grad_norm": 0.44756290316581726,
      "learning_rate": 9.764383264433033e-06,
      "loss": 1.3638,
      "step": 813
    },
    {
      "epoch": 0.12554705122520196,
      "grad_norm": 0.4206821024417877,
      "learning_rate": 9.763624976181524e-06,
      "loss": 1.3488,
      "step": 814
    },
    {
      "epoch": 0.12570128593186683,
      "grad_norm": 0.4024866223335266,
      "learning_rate": 9.76286549922496e-06,
      "loss": 1.3053,
      "step": 815
    },
    {
      "epoch": 0.1258555206385317,
      "grad_norm": 0.4793635904788971,
      "learning_rate": 9.762104833752857e-06,
      "loss": 1.3593,
      "step": 816
    },
    {
      "epoch": 0.12600975534519654,
      "grad_norm": 0.4584435820579529,
      "learning_rate": 9.761342979955028e-06,
      "loss": 1.383,
      "step": 817
    },
    {
      "epoch": 0.1261639900518614,
      "grad_norm": 0.41160374879837036,
      "learning_rate": 9.760579938021586e-06,
      "loss": 1.3495,
      "step": 818
    },
    {
      "epoch": 0.12631822475852628,
      "grad_norm": 0.45896100997924805,
      "learning_rate": 9.75981570814294e-06,
      "loss": 1.3939,
      "step": 819
    },
    {
      "epoch": 0.12647245946519114,
      "grad_norm": 0.446318656206131,
      "learning_rate": 9.759050290509793e-06,
      "loss": 1.3978,
      "step": 820
    },
    {
      "epoch": 0.126626694171856,
      "grad_norm": 0.4176998436450958,
      "learning_rate": 9.758283685313144e-06,
      "loss": 1.3333,
      "step": 821
    },
    {
      "epoch": 0.12678092887852088,
      "grad_norm": 0.43633905053138733,
      "learning_rate": 9.757515892744294e-06,
      "loss": 1.4195,
      "step": 822
    },
    {
      "epoch": 0.12693516358518575,
      "grad_norm": 0.41222596168518066,
      "learning_rate": 9.756746912994832e-06,
      "loss": 1.384,
      "step": 823
    },
    {
      "epoch": 0.12708939829185062,
      "grad_norm": 0.46195337176322937,
      "learning_rate": 9.755976746256652e-06,
      "loss": 1.3395,
      "step": 824
    },
    {
      "epoch": 0.1272436329985155,
      "grad_norm": 0.44091737270355225,
      "learning_rate": 9.755205392721937e-06,
      "loss": 1.361,
      "step": 825
    },
    {
      "epoch": 0.12739786770518036,
      "grad_norm": 0.4277057945728302,
      "learning_rate": 9.754432852583168e-06,
      "loss": 1.3699,
      "step": 826
    },
    {
      "epoch": 0.12755210241184523,
      "grad_norm": 0.40494897961616516,
      "learning_rate": 9.753659126033126e-06,
      "loss": 1.33,
      "step": 827
    },
    {
      "epoch": 0.1277063371185101,
      "grad_norm": 0.45431292057037354,
      "learning_rate": 9.752884213264885e-06,
      "loss": 1.3496,
      "step": 828
    },
    {
      "epoch": 0.12786057182517496,
      "grad_norm": 0.4500824213027954,
      "learning_rate": 9.752108114471812e-06,
      "loss": 1.3996,
      "step": 829
    },
    {
      "epoch": 0.12801480653183983,
      "grad_norm": 0.4346963167190552,
      "learning_rate": 9.751330829847577e-06,
      "loss": 1.359,
      "step": 830
    },
    {
      "epoch": 0.1281690412385047,
      "grad_norm": 0.40969425439834595,
      "learning_rate": 9.750552359586138e-06,
      "loss": 1.3537,
      "step": 831
    },
    {
      "epoch": 0.12832327594516957,
      "grad_norm": 0.4441986382007599,
      "learning_rate": 9.749772703881756e-06,
      "loss": 1.3715,
      "step": 832
    },
    {
      "epoch": 0.12847751065183444,
      "grad_norm": 0.43944767117500305,
      "learning_rate": 9.748991862928982e-06,
      "loss": 1.3516,
      "step": 833
    },
    {
      "epoch": 0.1286317453584993,
      "grad_norm": 0.4336840808391571,
      "learning_rate": 9.74820983692267e-06,
      "loss": 1.3596,
      "step": 834
    },
    {
      "epoch": 0.12878598006516417,
      "grad_norm": 0.46026572585105896,
      "learning_rate": 9.747426626057959e-06,
      "loss": 1.3847,
      "step": 835
    },
    {
      "epoch": 0.12894021477182904,
      "grad_norm": 0.4183986783027649,
      "learning_rate": 9.746642230530294e-06,
      "loss": 1.3972,
      "step": 836
    },
    {
      "epoch": 0.1290944494784939,
      "grad_norm": 0.3883669376373291,
      "learning_rate": 9.74585665053541e-06,
      "loss": 1.3173,
      "step": 837
    },
    {
      "epoch": 0.12924868418515875,
      "grad_norm": 0.4601610004901886,
      "learning_rate": 9.74506988626934e-06,
      "loss": 1.4002,
      "step": 838
    },
    {
      "epoch": 0.12940291889182362,
      "grad_norm": 0.44732755422592163,
      "learning_rate": 9.74428193792841e-06,
      "loss": 1.3914,
      "step": 839
    },
    {
      "epoch": 0.1295571535984885,
      "grad_norm": 0.45681890845298767,
      "learning_rate": 9.743492805709244e-06,
      "loss": 1.3837,
      "step": 840
    },
    {
      "epoch": 0.12971138830515336,
      "grad_norm": 0.471767783164978,
      "learning_rate": 9.742702489808761e-06,
      "loss": 1.3944,
      "step": 841
    },
    {
      "epoch": 0.12986562301181823,
      "grad_norm": 0.40301236510276794,
      "learning_rate": 9.741910990424173e-06,
      "loss": 1.3561,
      "step": 842
    },
    {
      "epoch": 0.1300198577184831,
      "grad_norm": 0.4440667927265167,
      "learning_rate": 9.74111830775299e-06,
      "loss": 1.368,
      "step": 843
    },
    {
      "epoch": 0.13017409242514796,
      "grad_norm": 0.5251238346099854,
      "learning_rate": 9.740324441993018e-06,
      "loss": 1.3995,
      "step": 844
    },
    {
      "epoch": 0.13032832713181283,
      "grad_norm": 0.41470107436180115,
      "learning_rate": 9.739529393342355e-06,
      "loss": 1.3369,
      "step": 845
    },
    {
      "epoch": 0.1304825618384777,
      "grad_norm": 0.4555199146270752,
      "learning_rate": 9.738733161999394e-06,
      "loss": 1.3424,
      "step": 846
    },
    {
      "epoch": 0.13063679654514257,
      "grad_norm": 0.4859868586063385,
      "learning_rate": 9.737935748162828e-06,
      "loss": 1.4185,
      "step": 847
    },
    {
      "epoch": 0.13079103125180744,
      "grad_norm": 0.4408034682273865,
      "learning_rate": 9.73713715203164e-06,
      "loss": 1.4042,
      "step": 848
    },
    {
      "epoch": 0.1309452659584723,
      "grad_norm": 0.4171469509601593,
      "learning_rate": 9.736337373805111e-06,
      "loss": 1.3344,
      "step": 849
    },
    {
      "epoch": 0.13109950066513718,
      "grad_norm": 0.3878125846385956,
      "learning_rate": 9.735536413682816e-06,
      "loss": 1.3237,
      "step": 850
    },
    {
      "epoch": 0.13125373537180204,
      "grad_norm": 0.43099191784858704,
      "learning_rate": 9.734734271864625e-06,
      "loss": 1.3605,
      "step": 851
    },
    {
      "epoch": 0.1314079700784669,
      "grad_norm": 0.4446086585521698,
      "learning_rate": 9.733930948550702e-06,
      "loss": 1.3158,
      "step": 852
    },
    {
      "epoch": 0.13156220478513178,
      "grad_norm": 0.474112331867218,
      "learning_rate": 9.733126443941507e-06,
      "loss": 1.3609,
      "step": 853
    },
    {
      "epoch": 0.13171643949179665,
      "grad_norm": 0.43937546014785767,
      "learning_rate": 9.732320758237794e-06,
      "loss": 1.3735,
      "step": 854
    },
    {
      "epoch": 0.13187067419846152,
      "grad_norm": 0.41289564967155457,
      "learning_rate": 9.731513891640613e-06,
      "loss": 1.3828,
      "step": 855
    },
    {
      "epoch": 0.1320249089051264,
      "grad_norm": 0.4423099458217621,
      "learning_rate": 9.730705844351309e-06,
      "loss": 1.3681,
      "step": 856
    },
    {
      "epoch": 0.13217914361179126,
      "grad_norm": 0.4533740282058716,
      "learning_rate": 9.729896616571519e-06,
      "loss": 1.367,
      "step": 857
    },
    {
      "epoch": 0.13233337831845612,
      "grad_norm": 0.46049922704696655,
      "learning_rate": 9.729086208503174e-06,
      "loss": 1.3652,
      "step": 858
    },
    {
      "epoch": 0.13248761302512096,
      "grad_norm": 0.46381819248199463,
      "learning_rate": 9.728274620348504e-06,
      "loss": 1.3368,
      "step": 859
    },
    {
      "epoch": 0.13264184773178583,
      "grad_norm": 0.4899432957172394,
      "learning_rate": 9.727461852310032e-06,
      "loss": 1.3698,
      "step": 860
    },
    {
      "epoch": 0.1327960824384507,
      "grad_norm": 0.3975609242916107,
      "learning_rate": 9.726647904590572e-06,
      "loss": 1.3373,
      "step": 861
    },
    {
      "epoch": 0.13295031714511557,
      "grad_norm": 0.42786306142807007,
      "learning_rate": 9.725832777393234e-06,
      "loss": 1.3073,
      "step": 862
    },
    {
      "epoch": 0.13310455185178044,
      "grad_norm": 0.4436379373073578,
      "learning_rate": 9.725016470921426e-06,
      "loss": 1.3573,
      "step": 863
    },
    {
      "epoch": 0.1332587865584453,
      "grad_norm": 0.45234617590904236,
      "learning_rate": 9.724198985378847e-06,
      "loss": 1.3429,
      "step": 864
    },
    {
      "epoch": 0.13341302126511018,
      "grad_norm": 0.41882964968681335,
      "learning_rate": 9.72338032096949e-06,
      "loss": 1.3558,
      "step": 865
    },
    {
      "epoch": 0.13356725597177505,
      "grad_norm": 0.48642703890800476,
      "learning_rate": 9.722560477897642e-06,
      "loss": 1.3553,
      "step": 866
    },
    {
      "epoch": 0.1337214906784399,
      "grad_norm": 0.4716489017009735,
      "learning_rate": 9.721739456367886e-06,
      "loss": 1.3769,
      "step": 867
    },
    {
      "epoch": 0.13387572538510478,
      "grad_norm": 0.4115413427352905,
      "learning_rate": 9.720917256585098e-06,
      "loss": 1.3463,
      "step": 868
    },
    {
      "epoch": 0.13402996009176965,
      "grad_norm": 0.47043079137802124,
      "learning_rate": 9.72009387875445e-06,
      "loss": 1.3902,
      "step": 869
    },
    {
      "epoch": 0.13418419479843452,
      "grad_norm": 0.4174489974975586,
      "learning_rate": 9.719269323081399e-06,
      "loss": 1.374,
      "step": 870
    },
    {
      "epoch": 0.1343384295050994,
      "grad_norm": 0.40322747826576233,
      "learning_rate": 9.71844358977171e-06,
      "loss": 1.3286,
      "step": 871
    },
    {
      "epoch": 0.13449266421176426,
      "grad_norm": 0.4627361595630646,
      "learning_rate": 9.717616679031432e-06,
      "loss": 1.3526,
      "step": 872
    },
    {
      "epoch": 0.13464689891842913,
      "grad_norm": 0.42119118571281433,
      "learning_rate": 9.716788591066911e-06,
      "loss": 1.3801,
      "step": 873
    },
    {
      "epoch": 0.134801133625094,
      "grad_norm": 0.44734987616539,
      "learning_rate": 9.715959326084787e-06,
      "loss": 1.3503,
      "step": 874
    },
    {
      "epoch": 0.13495536833175886,
      "grad_norm": 0.425028532743454,
      "learning_rate": 9.715128884291991e-06,
      "loss": 1.3584,
      "step": 875
    },
    {
      "epoch": 0.13510960303842373,
      "grad_norm": 0.4234400689601898,
      "learning_rate": 9.714297265895754e-06,
      "loss": 1.3766,
      "step": 876
    },
    {
      "epoch": 0.1352638377450886,
      "grad_norm": 0.41229814291000366,
      "learning_rate": 9.71346447110359e-06,
      "loss": 1.3476,
      "step": 877
    },
    {
      "epoch": 0.13541807245175347,
      "grad_norm": 0.4352407455444336,
      "learning_rate": 9.712630500123317e-06,
      "loss": 1.3671,
      "step": 878
    },
    {
      "epoch": 0.1355723071584183,
      "grad_norm": 0.44079896807670593,
      "learning_rate": 9.711795353163042e-06,
      "loss": 1.4354,
      "step": 879
    },
    {
      "epoch": 0.13572654186508318,
      "grad_norm": 0.41312628984451294,
      "learning_rate": 9.710959030431167e-06,
      "loss": 1.3838,
      "step": 880
    },
    {
      "epoch": 0.13588077657174805,
      "grad_norm": 0.44633620977401733,
      "learning_rate": 9.710121532136383e-06,
      "loss": 1.3159,
      "step": 881
    },
    {
      "epoch": 0.13603501127841292,
      "grad_norm": 0.43944814801216125,
      "learning_rate": 9.709282858487682e-06,
      "loss": 1.374,
      "step": 882
    },
    {
      "epoch": 0.13618924598507778,
      "grad_norm": 0.4563775062561035,
      "learning_rate": 9.708443009694339e-06,
      "loss": 1.3817,
      "step": 883
    },
    {
      "epoch": 0.13634348069174265,
      "grad_norm": 0.4323616921901703,
      "learning_rate": 9.707601985965932e-06,
      "loss": 1.3369,
      "step": 884
    },
    {
      "epoch": 0.13649771539840752,
      "grad_norm": 0.45572516322135925,
      "learning_rate": 9.706759787512328e-06,
      "loss": 1.392,
      "step": 885
    },
    {
      "epoch": 0.1366519501050724,
      "grad_norm": 0.41872602701187134,
      "learning_rate": 9.705916414543688e-06,
      "loss": 1.3704,
      "step": 886
    },
    {
      "epoch": 0.13680618481173726,
      "grad_norm": 0.42820021510124207,
      "learning_rate": 9.705071867270464e-06,
      "loss": 1.3701,
      "step": 887
    },
    {
      "epoch": 0.13696041951840213,
      "grad_norm": 0.4520045220851898,
      "learning_rate": 9.704226145903401e-06,
      "loss": 1.3537,
      "step": 888
    },
    {
      "epoch": 0.137114654225067,
      "grad_norm": 0.42270752787590027,
      "learning_rate": 9.703379250653544e-06,
      "loss": 1.3596,
      "step": 889
    },
    {
      "epoch": 0.13726888893173186,
      "grad_norm": 0.4259921610355377,
      "learning_rate": 9.70253118173222e-06,
      "loss": 1.3489,
      "step": 890
    },
    {
      "epoch": 0.13742312363839673,
      "grad_norm": 0.48967087268829346,
      "learning_rate": 9.701681939351058e-06,
      "loss": 1.3759,
      "step": 891
    },
    {
      "epoch": 0.1375773583450616,
      "grad_norm": 0.43577277660369873,
      "learning_rate": 9.700831523721974e-06,
      "loss": 1.3622,
      "step": 892
    },
    {
      "epoch": 0.13773159305172647,
      "grad_norm": 0.4252191483974457,
      "learning_rate": 9.69997993505718e-06,
      "loss": 1.3046,
      "step": 893
    },
    {
      "epoch": 0.13788582775839134,
      "grad_norm": 0.48695123195648193,
      "learning_rate": 9.69912717356918e-06,
      "loss": 1.3962,
      "step": 894
    },
    {
      "epoch": 0.1380400624650562,
      "grad_norm": 0.43830999732017517,
      "learning_rate": 9.698273239470767e-06,
      "loss": 1.3528,
      "step": 895
    },
    {
      "epoch": 0.13819429717172108,
      "grad_norm": 0.4534652829170227,
      "learning_rate": 9.697418132975036e-06,
      "loss": 1.3556,
      "step": 896
    },
    {
      "epoch": 0.13834853187838594,
      "grad_norm": 0.4268292188644409,
      "learning_rate": 9.696561854295363e-06,
      "loss": 1.3599,
      "step": 897
    },
    {
      "epoch": 0.1385027665850508,
      "grad_norm": 0.45288994908332825,
      "learning_rate": 9.695704403645425e-06,
      "loss": 1.3747,
      "step": 898
    },
    {
      "epoch": 0.13865700129171568,
      "grad_norm": 0.39907610416412354,
      "learning_rate": 9.694845781239188e-06,
      "loss": 1.3374,
      "step": 899
    },
    {
      "epoch": 0.13881123599838052,
      "grad_norm": 0.43562206625938416,
      "learning_rate": 9.69398598729091e-06,
      "loss": 1.3245,
      "step": 900
    },
    {
      "epoch": 0.1389654707050454,
      "grad_norm": 0.45644381642341614,
      "learning_rate": 9.693125022015142e-06,
      "loss": 1.3876,
      "step": 901
    },
    {
      "epoch": 0.13911970541171026,
      "grad_norm": 0.43593016266822815,
      "learning_rate": 9.692262885626728e-06,
      "loss": 1.3541,
      "step": 902
    },
    {
      "epoch": 0.13927394011837513,
      "grad_norm": 0.45160678029060364,
      "learning_rate": 9.691399578340804e-06,
      "loss": 1.3582,
      "step": 903
    },
    {
      "epoch": 0.13942817482504,
      "grad_norm": 0.4698358178138733,
      "learning_rate": 9.690535100372798e-06,
      "loss": 1.3571,
      "step": 904
    },
    {
      "epoch": 0.13958240953170487,
      "grad_norm": 0.44391974806785583,
      "learning_rate": 9.689669451938429e-06,
      "loss": 1.3267,
      "step": 905
    },
    {
      "epoch": 0.13973664423836973,
      "grad_norm": 0.40989452600479126,
      "learning_rate": 9.68880263325371e-06,
      "loss": 1.3324,
      "step": 906
    },
    {
      "epoch": 0.1398908789450346,
      "grad_norm": 0.43320152163505554,
      "learning_rate": 9.687934644534946e-06,
      "loss": 1.3727,
      "step": 907
    },
    {
      "epoch": 0.14004511365169947,
      "grad_norm": 0.4489857852458954,
      "learning_rate": 9.68706548599873e-06,
      "loss": 1.3658,
      "step": 908
    },
    {
      "epoch": 0.14019934835836434,
      "grad_norm": 0.426380455493927,
      "learning_rate": 9.686195157861954e-06,
      "loss": 1.3786,
      "step": 909
    },
    {
      "epoch": 0.1403535830650292,
      "grad_norm": 0.46585339307785034,
      "learning_rate": 9.685323660341795e-06,
      "loss": 1.3756,
      "step": 910
    },
    {
      "epoch": 0.14050781777169408,
      "grad_norm": 0.41312724351882935,
      "learning_rate": 9.684450993655726e-06,
      "loss": 1.3106,
      "step": 911
    },
    {
      "epoch": 0.14066205247835895,
      "grad_norm": 0.41777506470680237,
      "learning_rate": 9.68357715802151e-06,
      "loss": 1.3505,
      "step": 912
    },
    {
      "epoch": 0.14081628718502381,
      "grad_norm": 0.45272010564804077,
      "learning_rate": 9.682702153657201e-06,
      "loss": 1.3374,
      "step": 913
    },
    {
      "epoch": 0.14097052189168868,
      "grad_norm": 0.44729048013687134,
      "learning_rate": 9.68182598078115e-06,
      "loss": 1.3296,
      "step": 914
    },
    {
      "epoch": 0.14112475659835355,
      "grad_norm": 0.4224020540714264,
      "learning_rate": 9.680948639611989e-06,
      "loss": 1.3445,
      "step": 915
    },
    {
      "epoch": 0.14127899130501842,
      "grad_norm": 0.47426870465278625,
      "learning_rate": 9.680070130368652e-06,
      "loss": 1.3767,
      "step": 916
    },
    {
      "epoch": 0.1414332260116833,
      "grad_norm": 0.39650842547416687,
      "learning_rate": 9.679190453270362e-06,
      "loss": 1.369,
      "step": 917
    },
    {
      "epoch": 0.14158746071834816,
      "grad_norm": 0.44931554794311523,
      "learning_rate": 9.678309608536626e-06,
      "loss": 1.3383,
      "step": 918
    },
    {
      "epoch": 0.14174169542501303,
      "grad_norm": 0.45185673236846924,
      "learning_rate": 9.677427596387254e-06,
      "loss": 1.3198,
      "step": 919
    },
    {
      "epoch": 0.1418959301316779,
      "grad_norm": 0.4276964068412781,
      "learning_rate": 9.676544417042337e-06,
      "loss": 1.3496,
      "step": 920
    },
    {
      "epoch": 0.14205016483834274,
      "grad_norm": 0.4189877212047577,
      "learning_rate": 9.675660070722266e-06,
      "loss": 1.3889,
      "step": 921
    },
    {
      "epoch": 0.1422043995450076,
      "grad_norm": 0.47979092597961426,
      "learning_rate": 9.674774557647715e-06,
      "loss": 1.352,
      "step": 922
    },
    {
      "epoch": 0.14235863425167247,
      "grad_norm": 0.4569590091705322,
      "learning_rate": 9.673887878039654e-06,
      "loss": 1.4002,
      "step": 923
    },
    {
      "epoch": 0.14251286895833734,
      "grad_norm": 0.43906304240226746,
      "learning_rate": 9.673000032119346e-06,
      "loss": 1.3192,
      "step": 924
    },
    {
      "epoch": 0.1426671036650022,
      "grad_norm": 0.4342762231826782,
      "learning_rate": 9.672111020108339e-06,
      "loss": 1.3421,
      "step": 925
    },
    {
      "epoch": 0.14282133837166708,
      "grad_norm": 0.4287731647491455,
      "learning_rate": 9.671220842228476e-06,
      "loss": 1.38,
      "step": 926
    },
    {
      "epoch": 0.14297557307833195,
      "grad_norm": 0.44181495904922485,
      "learning_rate": 9.670329498701892e-06,
      "loss": 1.3506,
      "step": 927
    },
    {
      "epoch": 0.14312980778499682,
      "grad_norm": 0.4163863956928253,
      "learning_rate": 9.669436989751007e-06,
      "loss": 1.3436,
      "step": 928
    },
    {
      "epoch": 0.14328404249166168,
      "grad_norm": 0.46653324365615845,
      "learning_rate": 9.66854331559854e-06,
      "loss": 1.3784,
      "step": 929
    },
    {
      "epoch": 0.14343827719832655,
      "grad_norm": 0.4513910710811615,
      "learning_rate": 9.667648476467495e-06,
      "loss": 1.3189,
      "step": 930
    },
    {
      "epoch": 0.14359251190499142,
      "grad_norm": 0.39425528049468994,
      "learning_rate": 9.666752472581166e-06,
      "loss": 1.318,
      "step": 931
    },
    {
      "epoch": 0.1437467466116563,
      "grad_norm": 0.4691428542137146,
      "learning_rate": 9.665855304163145e-06,
      "loss": 1.3478,
      "step": 932
    },
    {
      "epoch": 0.14390098131832116,
      "grad_norm": 0.4166172444820404,
      "learning_rate": 9.664956971437306e-06,
      "loss": 1.3009,
      "step": 933
    },
    {
      "epoch": 0.14405521602498603,
      "grad_norm": 0.4548834264278412,
      "learning_rate": 9.664057474627817e-06,
      "loss": 1.3604,
      "step": 934
    },
    {
      "epoch": 0.1442094507316509,
      "grad_norm": 0.4332239031791687,
      "learning_rate": 9.663156813959139e-06,
      "loss": 1.3781,
      "step": 935
    },
    {
      "epoch": 0.14436368543831576,
      "grad_norm": 0.4403667449951172,
      "learning_rate": 9.662254989656018e-06,
      "loss": 1.3606,
      "step": 936
    },
    {
      "epoch": 0.14451792014498063,
      "grad_norm": 0.40217745304107666,
      "learning_rate": 9.661352001943494e-06,
      "loss": 1.3691,
      "step": 937
    },
    {
      "epoch": 0.1446721548516455,
      "grad_norm": 0.44711363315582275,
      "learning_rate": 9.660447851046898e-06,
      "loss": 1.3521,
      "step": 938
    },
    {
      "epoch": 0.14482638955831037,
      "grad_norm": 0.4292323589324951,
      "learning_rate": 9.659542537191849e-06,
      "loss": 1.3577,
      "step": 939
    },
    {
      "epoch": 0.14498062426497524,
      "grad_norm": 0.4293443560600281,
      "learning_rate": 9.658636060604258e-06,
      "loss": 1.3742,
      "step": 940
    },
    {
      "epoch": 0.1451348589716401,
      "grad_norm": 0.5029200315475464,
      "learning_rate": 9.657728421510324e-06,
      "loss": 1.3586,
      "step": 941
    },
    {
      "epoch": 0.14528909367830495,
      "grad_norm": 0.444880872964859,
      "learning_rate": 9.656819620136538e-06,
      "loss": 1.3642,
      "step": 942
    },
    {
      "epoch": 0.14544332838496982,
      "grad_norm": 0.4514634311199188,
      "learning_rate": 9.65590965670968e-06,
      "loss": 1.3464,
      "step": 943
    },
    {
      "epoch": 0.14559756309163469,
      "grad_norm": 0.43868181109428406,
      "learning_rate": 9.654998531456822e-06,
      "loss": 1.3661,
      "step": 944
    },
    {
      "epoch": 0.14575179779829955,
      "grad_norm": 0.4196695387363434,
      "learning_rate": 9.654086244605324e-06,
      "loss": 1.349,
      "step": 945
    },
    {
      "epoch": 0.14590603250496442,
      "grad_norm": 0.419352650642395,
      "learning_rate": 9.653172796382834e-06,
      "loss": 1.3434,
      "step": 946
    },
    {
      "epoch": 0.1460602672116293,
      "grad_norm": 0.45958200097084045,
      "learning_rate": 9.652258187017293e-06,
      "loss": 1.341,
      "step": 947
    },
    {
      "epoch": 0.14621450191829416,
      "grad_norm": 0.40580084919929504,
      "learning_rate": 9.651342416736932e-06,
      "loss": 1.3546,
      "step": 948
    },
    {
      "epoch": 0.14636873662495903,
      "grad_norm": 0.418164998292923,
      "learning_rate": 9.650425485770268e-06,
      "loss": 1.3513,
      "step": 949
    },
    {
      "epoch": 0.1465229713316239,
      "grad_norm": 0.43666157126426697,
      "learning_rate": 9.649507394346113e-06,
      "loss": 1.3664,
      "step": 950
    },
    {
      "epoch": 0.14667720603828877,
      "grad_norm": 0.4278264343738556,
      "learning_rate": 9.648588142693562e-06,
      "loss": 1.3672,
      "step": 951
    },
    {
      "epoch": 0.14683144074495363,
      "grad_norm": 0.4103301465511322,
      "learning_rate": 9.647667731042008e-06,
      "loss": 1.3623,
      "step": 952
    },
    {
      "epoch": 0.1469856754516185,
      "grad_norm": 0.41230452060699463,
      "learning_rate": 9.646746159621124e-06,
      "loss": 1.3589,
      "step": 953
    },
    {
      "epoch": 0.14713991015828337,
      "grad_norm": 0.4213346838951111,
      "learning_rate": 9.645823428660879e-06,
      "loss": 1.3741,
      "step": 954
    },
    {
      "epoch": 0.14729414486494824,
      "grad_norm": 0.3737544119358063,
      "learning_rate": 9.64489953839153e-06,
      "loss": 1.3636,
      "step": 955
    },
    {
      "epoch": 0.1474483795716131,
      "grad_norm": 0.4100255072116852,
      "learning_rate": 9.64397448904362e-06,
      "loss": 1.3759,
      "step": 956
    },
    {
      "epoch": 0.14760261427827798,
      "grad_norm": 0.4222797751426697,
      "learning_rate": 9.643048280847987e-06,
      "loss": 1.3542,
      "step": 957
    },
    {
      "epoch": 0.14775684898494285,
      "grad_norm": 0.41794565320014954,
      "learning_rate": 9.642120914035752e-06,
      "loss": 1.3968,
      "step": 958
    },
    {
      "epoch": 0.14791108369160771,
      "grad_norm": 0.4112548232078552,
      "learning_rate": 9.641192388838333e-06,
      "loss": 1.3563,
      "step": 959
    },
    {
      "epoch": 0.14806531839827258,
      "grad_norm": 0.4756266176700592,
      "learning_rate": 9.640262705487429e-06,
      "loss": 1.3644,
      "step": 960
    },
    {
      "epoch": 0.14821955310493745,
      "grad_norm": 0.421749085187912,
      "learning_rate": 9.639331864215028e-06,
      "loss": 1.3707,
      "step": 961
    },
    {
      "epoch": 0.1483737878116023,
      "grad_norm": 0.43855324387550354,
      "learning_rate": 9.638399865253417e-06,
      "loss": 1.3365,
      "step": 962
    },
    {
      "epoch": 0.14852802251826716,
      "grad_norm": 0.45007380843162537,
      "learning_rate": 9.637466708835162e-06,
      "loss": 1.3759,
      "step": 963
    },
    {
      "epoch": 0.14868225722493203,
      "grad_norm": 0.38929274678230286,
      "learning_rate": 9.636532395193119e-06,
      "loss": 1.3174,
      "step": 964
    },
    {
      "epoch": 0.1488364919315969,
      "grad_norm": 0.4295295476913452,
      "learning_rate": 9.635596924560435e-06,
      "loss": 1.3696,
      "step": 965
    },
    {
      "epoch": 0.14899072663826177,
      "grad_norm": 0.46214181184768677,
      "learning_rate": 9.634660297170549e-06,
      "loss": 1.3849,
      "step": 966
    },
    {
      "epoch": 0.14914496134492664,
      "grad_norm": 0.4029429852962494,
      "learning_rate": 9.633722513257183e-06,
      "loss": 1.3407,
      "step": 967
    },
    {
      "epoch": 0.1492991960515915,
      "grad_norm": 0.445356547832489,
      "learning_rate": 9.632783573054347e-06,
      "loss": 1.3544,
      "step": 968
    },
    {
      "epoch": 0.14945343075825637,
      "grad_norm": 0.49758583307266235,
      "learning_rate": 9.631843476796348e-06,
      "loss": 1.3534,
      "step": 969
    },
    {
      "epoch": 0.14960766546492124,
      "grad_norm": 0.40837979316711426,
      "learning_rate": 9.63090222471777e-06,
      "loss": 1.3528,
      "step": 970
    },
    {
      "epoch": 0.1497619001715861,
      "grad_norm": 0.42469340562820435,
      "learning_rate": 9.629959817053494e-06,
      "loss": 1.3276,
      "step": 971
    },
    {
      "epoch": 0.14991613487825098,
      "grad_norm": 0.44042232632637024,
      "learning_rate": 9.629016254038685e-06,
      "loss": 1.3828,
      "step": 972
    },
    {
      "epoch": 0.15007036958491585,
      "grad_norm": 0.3965208828449249,
      "learning_rate": 9.628071535908799e-06,
      "loss": 1.3825,
      "step": 973
    },
    {
      "epoch": 0.15022460429158072,
      "grad_norm": 0.420837938785553,
      "learning_rate": 9.627125662899578e-06,
      "loss": 1.3114,
      "step": 974
    },
    {
      "epoch": 0.15037883899824558,
      "grad_norm": 0.40281036496162415,
      "learning_rate": 9.626178635247054e-06,
      "loss": 1.3175,
      "step": 975
    },
    {
      "epoch": 0.15053307370491045,
      "grad_norm": 0.4451516270637512,
      "learning_rate": 9.625230453187548e-06,
      "loss": 1.3864,
      "step": 976
    },
    {
      "epoch": 0.15068730841157532,
      "grad_norm": 0.3975658714771271,
      "learning_rate": 9.624281116957664e-06,
      "loss": 1.3553,
      "step": 977
    },
    {
      "epoch": 0.1508415431182402,
      "grad_norm": 0.41001689434051514,
      "learning_rate": 9.623330626794298e-06,
      "loss": 1.3426,
      "step": 978
    },
    {
      "epoch": 0.15099577782490506,
      "grad_norm": 0.44088777899742126,
      "learning_rate": 9.622378982934635e-06,
      "loss": 1.3996,
      "step": 979
    },
    {
      "epoch": 0.15115001253156993,
      "grad_norm": 0.4407082796096802,
      "learning_rate": 9.621426185616146e-06,
      "loss": 1.3617,
      "step": 980
    },
    {
      "epoch": 0.1513042472382348,
      "grad_norm": 0.41040855646133423,
      "learning_rate": 9.620472235076587e-06,
      "loss": 1.3487,
      "step": 981
    },
    {
      "epoch": 0.15145848194489966,
      "grad_norm": 0.4353833496570587,
      "learning_rate": 9.619517131554008e-06,
      "loss": 1.3804,
      "step": 982
    },
    {
      "epoch": 0.1516127166515645,
      "grad_norm": 0.4234057366847992,
      "learning_rate": 9.618560875286745e-06,
      "loss": 1.3538,
      "step": 983
    },
    {
      "epoch": 0.15176695135822937,
      "grad_norm": 0.428462415933609,
      "learning_rate": 9.617603466513415e-06,
      "loss": 1.3589,
      "step": 984
    },
    {
      "epoch": 0.15192118606489424,
      "grad_norm": 0.45914000272750854,
      "learning_rate": 9.616644905472932e-06,
      "loss": 1.3518,
      "step": 985
    },
    {
      "epoch": 0.1520754207715591,
      "grad_norm": 0.4305824935436249,
      "learning_rate": 9.615685192404494e-06,
      "loss": 1.3552,
      "step": 986
    },
    {
      "epoch": 0.15222965547822398,
      "grad_norm": 0.44539913535118103,
      "learning_rate": 9.614724327547582e-06,
      "loss": 1.3502,
      "step": 987
    },
    {
      "epoch": 0.15238389018488885,
      "grad_norm": 0.4699813723564148,
      "learning_rate": 9.61376231114197e-06,
      "loss": 1.3621,
      "step": 988
    },
    {
      "epoch": 0.15253812489155372,
      "grad_norm": 0.3810335397720337,
      "learning_rate": 9.612799143427717e-06,
      "loss": 1.3402,
      "step": 989
    },
    {
      "epoch": 0.15269235959821859,
      "grad_norm": 0.4232047200202942,
      "learning_rate": 9.611834824645172e-06,
      "loss": 1.3369,
      "step": 990
    },
    {
      "epoch": 0.15284659430488345,
      "grad_norm": 0.4489132761955261,
      "learning_rate": 9.610869355034968e-06,
      "loss": 1.3648,
      "step": 991
    },
    {
      "epoch": 0.15300082901154832,
      "grad_norm": 0.4090892970561981,
      "learning_rate": 9.609902734838024e-06,
      "loss": 1.3336,
      "step": 992
    },
    {
      "epoch": 0.1531550637182132,
      "grad_norm": 0.41862618923187256,
      "learning_rate": 9.608934964295552e-06,
      "loss": 1.3361,
      "step": 993
    },
    {
      "epoch": 0.15330929842487806,
      "grad_norm": 0.45245254039764404,
      "learning_rate": 9.607966043649047e-06,
      "loss": 1.3474,
      "step": 994
    },
    {
      "epoch": 0.15346353313154293,
      "grad_norm": 0.4146495759487152,
      "learning_rate": 9.606995973140287e-06,
      "loss": 1.3285,
      "step": 995
    },
    {
      "epoch": 0.1536177678382078,
      "grad_norm": 0.4099392890930176,
      "learning_rate": 9.606024753011346e-06,
      "loss": 1.3385,
      "step": 996
    },
    {
      "epoch": 0.15377200254487267,
      "grad_norm": 0.46234220266342163,
      "learning_rate": 9.605052383504578e-06,
      "loss": 1.3286,
      "step": 997
    },
    {
      "epoch": 0.15392623725153753,
      "grad_norm": 0.3993547558784485,
      "learning_rate": 9.60407886486263e-06,
      "loss": 1.33,
      "step": 998
    },
    {
      "epoch": 0.1540804719582024,
      "grad_norm": 0.43349605798721313,
      "learning_rate": 9.603104197328426e-06,
      "loss": 1.3501,
      "step": 999
    },
    {
      "epoch": 0.15423470666486727,
      "grad_norm": 0.4560033977031708,
      "learning_rate": 9.602128381145186e-06,
      "loss": 1.3747,
      "step": 1000
    },
    {
      "epoch": 0.15438894137153214,
      "grad_norm": 0.43051019310951233,
      "learning_rate": 9.60115141655641e-06,
      "loss": 1.3829,
      "step": 1001
    },
    {
      "epoch": 0.154543176078197,
      "grad_norm": 0.40189671516418457,
      "learning_rate": 9.600173303805893e-06,
      "loss": 1.347,
      "step": 1002
    },
    {
      "epoch": 0.15469741078486188,
      "grad_norm": 0.4193477928638458,
      "learning_rate": 9.599194043137707e-06,
      "loss": 1.3523,
      "step": 1003
    },
    {
      "epoch": 0.15485164549152672,
      "grad_norm": 0.44596585631370544,
      "learning_rate": 9.598213634796218e-06,
      "loss": 1.3744,
      "step": 1004
    },
    {
      "epoch": 0.1550058801981916,
      "grad_norm": 0.4346173405647278,
      "learning_rate": 9.597232079026071e-06,
      "loss": 1.3562,
      "step": 1005
    },
    {
      "epoch": 0.15516011490485646,
      "grad_norm": 0.40672799944877625,
      "learning_rate": 9.596249376072204e-06,
      "loss": 1.383,
      "step": 1006
    },
    {
      "epoch": 0.15531434961152132,
      "grad_norm": 0.4428286850452423,
      "learning_rate": 9.595265526179839e-06,
      "loss": 1.3863,
      "step": 1007
    },
    {
      "epoch": 0.1554685843181862,
      "grad_norm": 0.4054853320121765,
      "learning_rate": 9.59428052959448e-06,
      "loss": 1.3395,
      "step": 1008
    },
    {
      "epoch": 0.15562281902485106,
      "grad_norm": 0.4263022541999817,
      "learning_rate": 9.593294386561926e-06,
      "loss": 1.3331,
      "step": 1009
    },
    {
      "epoch": 0.15577705373151593,
      "grad_norm": 0.4388751983642578,
      "learning_rate": 9.592307097328255e-06,
      "loss": 1.393,
      "step": 1010
    },
    {
      "epoch": 0.1559312884381808,
      "grad_norm": 0.4239785671234131,
      "learning_rate": 9.591318662139833e-06,
      "loss": 1.3521,
      "step": 1011
    },
    {
      "epoch": 0.15608552314484567,
      "grad_norm": 0.4259006381034851,
      "learning_rate": 9.59032908124331e-06,
      "loss": 1.369,
      "step": 1012
    },
    {
      "epoch": 0.15623975785151054,
      "grad_norm": 0.47064313292503357,
      "learning_rate": 9.58933835488563e-06,
      "loss": 1.3671,
      "step": 1013
    },
    {
      "epoch": 0.1563939925581754,
      "grad_norm": 0.4133605659008026,
      "learning_rate": 9.58834648331401e-06,
      "loss": 1.3291,
      "step": 1014
    },
    {
      "epoch": 0.15654822726484027,
      "grad_norm": 0.38037538528442383,
      "learning_rate": 9.587353466775963e-06,
      "loss": 1.3178,
      "step": 1015
    },
    {
      "epoch": 0.15670246197150514,
      "grad_norm": 0.44501787424087524,
      "learning_rate": 9.586359305519283e-06,
      "loss": 1.3593,
      "step": 1016
    },
    {
      "epoch": 0.15685669667817,
      "grad_norm": 0.40758368372917175,
      "learning_rate": 9.585363999792053e-06,
      "loss": 1.3694,
      "step": 1017
    },
    {
      "epoch": 0.15701093138483488,
      "grad_norm": 0.4244682788848877,
      "learning_rate": 9.584367549842638e-06,
      "loss": 1.3232,
      "step": 1018
    },
    {
      "epoch": 0.15716516609149975,
      "grad_norm": 0.5118983387947083,
      "learning_rate": 9.58336995591969e-06,
      "loss": 1.3609,
      "step": 1019
    },
    {
      "epoch": 0.15731940079816462,
      "grad_norm": 0.3951735496520996,
      "learning_rate": 9.582371218272146e-06,
      "loss": 1.3782,
      "step": 1020
    },
    {
      "epoch": 0.15747363550482948,
      "grad_norm": 0.38878145813941956,
      "learning_rate": 9.581371337149232e-06,
      "loss": 1.3093,
      "step": 1021
    },
    {
      "epoch": 0.15762787021149435,
      "grad_norm": 0.4443088471889496,
      "learning_rate": 9.580370312800452e-06,
      "loss": 1.366,
      "step": 1022
    },
    {
      "epoch": 0.15778210491815922,
      "grad_norm": 0.45588213205337524,
      "learning_rate": 9.5793681454756e-06,
      "loss": 1.3797,
      "step": 1023
    },
    {
      "epoch": 0.15793633962482406,
      "grad_norm": 0.3731274902820587,
      "learning_rate": 9.578364835424758e-06,
      "loss": 1.3328,
      "step": 1024
    },
    {
      "epoch": 0.15809057433148893,
      "grad_norm": 0.43674901127815247,
      "learning_rate": 9.577360382898289e-06,
      "loss": 1.3419,
      "step": 1025
    },
    {
      "epoch": 0.1582448090381538,
      "grad_norm": 0.42338722944259644,
      "learning_rate": 9.57635478814684e-06,
      "loss": 1.3628,
      "step": 1026
    },
    {
      "epoch": 0.15839904374481867,
      "grad_norm": 0.38588276505470276,
      "learning_rate": 9.575348051421347e-06,
      "loss": 1.3147,
      "step": 1027
    },
    {
      "epoch": 0.15855327845148354,
      "grad_norm": 0.3909562826156616,
      "learning_rate": 9.574340172973027e-06,
      "loss": 1.3209,
      "step": 1028
    },
    {
      "epoch": 0.1587075131581484,
      "grad_norm": 0.38329264521598816,
      "learning_rate": 9.573331153053385e-06,
      "loss": 1.336,
      "step": 1029
    },
    {
      "epoch": 0.15886174786481327,
      "grad_norm": 0.400493860244751,
      "learning_rate": 9.572320991914209e-06,
      "loss": 1.3121,
      "step": 1030
    },
    {
      "epoch": 0.15901598257147814,
      "grad_norm": 0.42534539103507996,
      "learning_rate": 9.571309689807575e-06,
      "loss": 1.3077,
      "step": 1031
    },
    {
      "epoch": 0.159170217278143,
      "grad_norm": 0.4600400924682617,
      "learning_rate": 9.570297246985838e-06,
      "loss": 1.3919,
      "step": 1032
    },
    {
      "epoch": 0.15932445198480788,
      "grad_norm": 0.3911516070365906,
      "learning_rate": 9.569283663701643e-06,
      "loss": 1.3613,
      "step": 1033
    },
    {
      "epoch": 0.15947868669147275,
      "grad_norm": 0.40016788244247437,
      "learning_rate": 9.568268940207917e-06,
      "loss": 1.3612,
      "step": 1034
    },
    {
      "epoch": 0.15963292139813762,
      "grad_norm": 0.4020020365715027,
      "learning_rate": 9.56725307675787e-06,
      "loss": 1.328,
      "step": 1035
    },
    {
      "epoch": 0.15978715610480249,
      "grad_norm": 0.3938751220703125,
      "learning_rate": 9.566236073605002e-06,
      "loss": 1.2962,
      "step": 1036
    },
    {
      "epoch": 0.15994139081146735,
      "grad_norm": 0.4025926887989044,
      "learning_rate": 9.565217931003092e-06,
      "loss": 1.3177,
      "step": 1037
    },
    {
      "epoch": 0.16009562551813222,
      "grad_norm": 0.4471730589866638,
      "learning_rate": 9.564198649206204e-06,
      "loss": 1.3639,
      "step": 1038
    },
    {
      "epoch": 0.1602498602247971,
      "grad_norm": 0.4198334217071533,
      "learning_rate": 9.563178228468689e-06,
      "loss": 1.3248,
      "step": 1039
    },
    {
      "epoch": 0.16040409493146196,
      "grad_norm": 0.433456689119339,
      "learning_rate": 9.56215666904518e-06,
      "loss": 1.3612,
      "step": 1040
    },
    {
      "epoch": 0.16055832963812683,
      "grad_norm": 0.5922003388404846,
      "learning_rate": 9.561133971190597e-06,
      "loss": 1.3588,
      "step": 1041
    },
    {
      "epoch": 0.1607125643447917,
      "grad_norm": 0.4096234142780304,
      "learning_rate": 9.560110135160139e-06,
      "loss": 1.3321,
      "step": 1042
    },
    {
      "epoch": 0.16086679905145657,
      "grad_norm": 0.41673043370246887,
      "learning_rate": 9.559085161209293e-06,
      "loss": 1.3201,
      "step": 1043
    },
    {
      "epoch": 0.16102103375812143,
      "grad_norm": 0.49432477355003357,
      "learning_rate": 9.558059049593829e-06,
      "loss": 1.3519,
      "step": 1044
    },
    {
      "epoch": 0.16117526846478628,
      "grad_norm": 0.41908732056617737,
      "learning_rate": 9.5570318005698e-06,
      "loss": 1.3613,
      "step": 1045
    },
    {
      "epoch": 0.16132950317145114,
      "grad_norm": 0.4030068814754486,
      "learning_rate": 9.556003414393544e-06,
      "loss": 1.332,
      "step": 1046
    },
    {
      "epoch": 0.161483737878116,
      "grad_norm": 0.45960256457328796,
      "learning_rate": 9.554973891321683e-06,
      "loss": 1.3313,
      "step": 1047
    },
    {
      "epoch": 0.16163797258478088,
      "grad_norm": 0.44315245747566223,
      "learning_rate": 9.553943231611123e-06,
      "loss": 1.36,
      "step": 1048
    },
    {
      "epoch": 0.16179220729144575,
      "grad_norm": 0.43170469999313354,
      "learning_rate": 9.552911435519049e-06,
      "loss": 1.3353,
      "step": 1049
    },
    {
      "epoch": 0.16194644199811062,
      "grad_norm": 0.42984703183174133,
      "learning_rate": 9.551878503302936e-06,
      "loss": 1.3839,
      "step": 1050
    },
    {
      "epoch": 0.1621006767047755,
      "grad_norm": 0.3931952714920044,
      "learning_rate": 9.55084443522054e-06,
      "loss": 1.3711,
      "step": 1051
    },
    {
      "epoch": 0.16225491141144036,
      "grad_norm": 0.4282974600791931,
      "learning_rate": 9.5498092315299e-06,
      "loss": 1.3377,
      "step": 1052
    },
    {
      "epoch": 0.16240914611810522,
      "grad_norm": 0.41101571917533875,
      "learning_rate": 9.548772892489335e-06,
      "loss": 1.3453,
      "step": 1053
    },
    {
      "epoch": 0.1625633808247701,
      "grad_norm": 0.43313929438591003,
      "learning_rate": 9.547735418357458e-06,
      "loss": 1.3655,
      "step": 1054
    },
    {
      "epoch": 0.16271761553143496,
      "grad_norm": 0.4140447676181793,
      "learning_rate": 9.546696809393151e-06,
      "loss": 1.3421,
      "step": 1055
    },
    {
      "epoch": 0.16287185023809983,
      "grad_norm": 0.4221991300582886,
      "learning_rate": 9.54565706585559e-06,
      "loss": 1.3045,
      "step": 1056
    },
    {
      "epoch": 0.1630260849447647,
      "grad_norm": 0.4684382379055023,
      "learning_rate": 9.54461618800423e-06,
      "loss": 1.3527,
      "step": 1057
    },
    {
      "epoch": 0.16318031965142957,
      "grad_norm": 0.42063719034194946,
      "learning_rate": 9.54357417609881e-06,
      "loss": 1.3409,
      "step": 1058
    },
    {
      "epoch": 0.16333455435809444,
      "grad_norm": 0.39833924174308777,
      "learning_rate": 9.54253103039935e-06,
      "loss": 1.3076,
      "step": 1059
    },
    {
      "epoch": 0.1634887890647593,
      "grad_norm": 0.43191730976104736,
      "learning_rate": 9.541486751166156e-06,
      "loss": 1.3582,
      "step": 1060
    },
    {
      "epoch": 0.16364302377142417,
      "grad_norm": 0.39600446820259094,
      "learning_rate": 9.540441338659813e-06,
      "loss": 1.333,
      "step": 1061
    },
    {
      "epoch": 0.16379725847808904,
      "grad_norm": 0.41398027539253235,
      "learning_rate": 9.539394793141193e-06,
      "loss": 1.3444,
      "step": 1062
    },
    {
      "epoch": 0.1639514931847539,
      "grad_norm": 0.4630861282348633,
      "learning_rate": 9.538347114871449e-06,
      "loss": 1.3206,
      "step": 1063
    },
    {
      "epoch": 0.16410572789141878,
      "grad_norm": 0.4108981192111969,
      "learning_rate": 9.537298304112015e-06,
      "loss": 1.3563,
      "step": 1064
    },
    {
      "epoch": 0.16425996259808365,
      "grad_norm": 0.38527730107307434,
      "learning_rate": 9.536248361124611e-06,
      "loss": 1.3242,
      "step": 1065
    },
    {
      "epoch": 0.1644141973047485,
      "grad_norm": 0.460708886384964,
      "learning_rate": 9.535197286171235e-06,
      "loss": 1.3316,
      "step": 1066
    },
    {
      "epoch": 0.16456843201141336,
      "grad_norm": 0.4495241641998291,
      "learning_rate": 9.534145079514172e-06,
      "loss": 1.3554,
      "step": 1067
    },
    {
      "epoch": 0.16472266671807823,
      "grad_norm": 0.39525145292282104,
      "learning_rate": 9.533091741415986e-06,
      "loss": 1.3183,
      "step": 1068
    },
    {
      "epoch": 0.1648769014247431,
      "grad_norm": 0.5031570196151733,
      "learning_rate": 9.532037272139527e-06,
      "loss": 1.371,
      "step": 1069
    },
    {
      "epoch": 0.16503113613140796,
      "grad_norm": 0.42758193612098694,
      "learning_rate": 9.530981671947924e-06,
      "loss": 1.3389,
      "step": 1070
    },
    {
      "epoch": 0.16518537083807283,
      "grad_norm": 0.4300083816051483,
      "learning_rate": 9.529924941104586e-06,
      "loss": 1.3397,
      "step": 1071
    },
    {
      "epoch": 0.1653396055447377,
      "grad_norm": 0.45306211709976196,
      "learning_rate": 9.528867079873214e-06,
      "loss": 1.3425,
      "step": 1072
    },
    {
      "epoch": 0.16549384025140257,
      "grad_norm": 0.42708519101142883,
      "learning_rate": 9.52780808851778e-06,
      "loss": 1.4067,
      "step": 1073
    },
    {
      "epoch": 0.16564807495806744,
      "grad_norm": 0.39134591817855835,
      "learning_rate": 9.526747967302544e-06,
      "loss": 1.3273,
      "step": 1074
    },
    {
      "epoch": 0.1658023096647323,
      "grad_norm": 0.3904189467430115,
      "learning_rate": 9.525686716492044e-06,
      "loss": 1.3488,
      "step": 1075
    },
    {
      "epoch": 0.16595654437139717,
      "grad_norm": 0.40552303194999695,
      "learning_rate": 9.524624336351104e-06,
      "loss": 1.3742,
      "step": 1076
    },
    {
      "epoch": 0.16611077907806204,
      "grad_norm": 0.4283524751663208,
      "learning_rate": 9.523560827144829e-06,
      "loss": 1.3375,
      "step": 1077
    },
    {
      "epoch": 0.1662650137847269,
      "grad_norm": 0.4074269235134125,
      "learning_rate": 9.522496189138606e-06,
      "loss": 1.3307,
      "step": 1078
    },
    {
      "epoch": 0.16641924849139178,
      "grad_norm": 0.41989561915397644,
      "learning_rate": 9.521430422598099e-06,
      "loss": 1.3519,
      "step": 1079
    },
    {
      "epoch": 0.16657348319805665,
      "grad_norm": 0.42565736174583435,
      "learning_rate": 9.520363527789258e-06,
      "loss": 1.3043,
      "step": 1080
    },
    {
      "epoch": 0.16672771790472152,
      "grad_norm": 0.43175825476646423,
      "learning_rate": 9.519295504978316e-06,
      "loss": 1.3264,
      "step": 1081
    },
    {
      "epoch": 0.16688195261138639,
      "grad_norm": 0.4620734751224518,
      "learning_rate": 9.518226354431785e-06,
      "loss": 1.3223,
      "step": 1082
    },
    {
      "epoch": 0.16703618731805125,
      "grad_norm": 0.3752366602420807,
      "learning_rate": 9.517156076416457e-06,
      "loss": 1.3205,
      "step": 1083
    },
    {
      "epoch": 0.16719042202471612,
      "grad_norm": 0.43047410249710083,
      "learning_rate": 9.516084671199405e-06,
      "loss": 1.3581,
      "step": 1084
    },
    {
      "epoch": 0.167344656731381,
      "grad_norm": 0.46193546056747437,
      "learning_rate": 9.515012139047989e-06,
      "loss": 1.3808,
      "step": 1085
    },
    {
      "epoch": 0.16749889143804583,
      "grad_norm": 0.42722025513648987,
      "learning_rate": 9.513938480229845e-06,
      "loss": 1.3486,
      "step": 1086
    },
    {
      "epoch": 0.1676531261447107,
      "grad_norm": 0.3964281976222992,
      "learning_rate": 9.512863695012892e-06,
      "loss": 1.3063,
      "step": 1087
    },
    {
      "epoch": 0.16780736085137557,
      "grad_norm": 0.44473353028297424,
      "learning_rate": 9.511787783665329e-06,
      "loss": 1.3665,
      "step": 1088
    },
    {
      "epoch": 0.16796159555804044,
      "grad_norm": 0.4227943420410156,
      "learning_rate": 9.510710746455636e-06,
      "loss": 1.3032,
      "step": 1089
    },
    {
      "epoch": 0.1681158302647053,
      "grad_norm": 0.41199952363967896,
      "learning_rate": 9.509632583652578e-06,
      "loss": 1.2954,
      "step": 1090
    },
    {
      "epoch": 0.16827006497137018,
      "grad_norm": 0.4591391086578369,
      "learning_rate": 9.508553295525191e-06,
      "loss": 1.3524,
      "step": 1091
    },
    {
      "epoch": 0.16842429967803504,
      "grad_norm": 0.4313882887363434,
      "learning_rate": 9.507472882342805e-06,
      "loss": 1.3684,
      "step": 1092
    },
    {
      "epoch": 0.1685785343846999,
      "grad_norm": 0.3902830481529236,
      "learning_rate": 9.506391344375022e-06,
      "loss": 1.3093,
      "step": 1093
    },
    {
      "epoch": 0.16873276909136478,
      "grad_norm": 0.47131410241127014,
      "learning_rate": 9.505308681891725e-06,
      "loss": 1.342,
      "step": 1094
    },
    {
      "epoch": 0.16888700379802965,
      "grad_norm": 0.42069944739341736,
      "learning_rate": 9.50422489516308e-06,
      "loss": 1.3922,
      "step": 1095
    },
    {
      "epoch": 0.16904123850469452,
      "grad_norm": 0.4462716579437256,
      "learning_rate": 9.503139984459536e-06,
      "loss": 1.3458,
      "step": 1096
    },
    {
      "epoch": 0.1691954732113594,
      "grad_norm": 0.49075546860694885,
      "learning_rate": 9.502053950051815e-06,
      "loss": 1.3583,
      "step": 1097
    },
    {
      "epoch": 0.16934970791802426,
      "grad_norm": 0.40180158615112305,
      "learning_rate": 9.500966792210926e-06,
      "loss": 1.3663,
      "step": 1098
    },
    {
      "epoch": 0.16950394262468912,
      "grad_norm": 0.40839266777038574,
      "learning_rate": 9.499878511208155e-06,
      "loss": 1.3398,
      "step": 1099
    },
    {
      "epoch": 0.169658177331354,
      "grad_norm": 0.45156973600387573,
      "learning_rate": 9.498789107315069e-06,
      "loss": 1.3064,
      "step": 1100
    }
  ],
  "logging_steps": 1,
  "max_steps": 6484,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.375515394495742e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}