{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9995100440960314,
  "eval_steps": 383,
  "global_step": 1530,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006532745386248571,
      "grad_norm": 6.782679973808308e+18,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.1939,
      "step": 1
    },
    {
      "epoch": 0.0006532745386248571,
      "eval_loss": 2.101844310760498,
      "eval_runtime": 159.9057,
      "eval_samples_per_second": 16.122,
      "eval_steps_per_second": 4.034,
      "step": 1
    },
    {
      "epoch": 0.0013065490772497142,
      "grad_norm": 1.0602286061923074e+19,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.3265,
      "step": 2
    },
    {
      "epoch": 0.0019598236158745713,
      "grad_norm": 9.040893238818374e+18,
      "learning_rate": 1.2e-05,
      "loss": 1.2696,
      "step": 3
    },
    {
      "epoch": 0.0026130981544994283,
      "grad_norm": 8.791663639615504e+18,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.4,
      "step": 4
    },
    {
      "epoch": 0.0032663726931242854,
      "grad_norm": 1.0141305916373336e+19,
      "learning_rate": 2e-05,
      "loss": 1.4178,
      "step": 5
    },
    {
      "epoch": 0.0039196472317491425,
      "grad_norm": 1.5361737647030534e+19,
      "learning_rate": 2.4e-05,
      "loss": 1.4652,
      "step": 6
    },
    {
      "epoch": 0.004572921770374,
      "grad_norm": 1.329169150346317e+19,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 1.548,
      "step": 7
    },
    {
      "epoch": 0.005226196308998857,
      "grad_norm": 1.5421850146744304e+19,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 1.5716,
      "step": 8
    },
    {
      "epoch": 0.005879470847623714,
      "grad_norm": Infinity,
      "learning_rate": 3.6e-05,
      "loss": 1.5789,
      "step": 9
    },
    {
      "epoch": 0.006532745386248571,
      "grad_norm": Infinity,
      "learning_rate": 4e-05,
      "loss": 1.6679,
      "step": 10
    },
    {
      "epoch": 0.007186019924873428,
      "grad_norm": Infinity,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 1.7226,
      "step": 11
    },
    {
      "epoch": 0.007839294463498285,
      "grad_norm": Infinity,
      "learning_rate": 4.8e-05,
      "loss": 1.7121,
      "step": 12
    },
    {
      "epoch": 0.008492569002123142,
      "grad_norm": Infinity,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 1.7012,
      "step": 13
    },
    {
      "epoch": 0.009145843540748,
      "grad_norm": Infinity,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 1.8244,
      "step": 14
    },
    {
      "epoch": 0.009799118079372856,
      "grad_norm": Infinity,
      "learning_rate": 6e-05,
      "loss": 1.9075,
      "step": 15
    },
    {
      "epoch": 0.010452392617997713,
      "grad_norm": Infinity,
      "learning_rate": 6.400000000000001e-05,
      "loss": 2.2215,
      "step": 16
    },
    {
      "epoch": 0.01110566715662257,
      "grad_norm": Infinity,
      "learning_rate": 6.800000000000001e-05,
      "loss": 2.3717,
      "step": 17
    },
    {
      "epoch": 0.011758941695247428,
      "grad_norm": Infinity,
      "learning_rate": 7.2e-05,
      "loss": 2.5005,
      "step": 18
    },
    {
      "epoch": 0.012412216233872285,
      "grad_norm": Infinity,
      "learning_rate": 7.6e-05,
      "loss": 2.5937,
      "step": 19
    },
    {
      "epoch": 0.013065490772497142,
      "grad_norm": Infinity,
      "learning_rate": 8e-05,
      "loss": 2.4281,
      "step": 20
    },
    {
      "epoch": 0.013718765311121999,
      "grad_norm": Infinity,
      "learning_rate": 8.4e-05,
      "loss": 2.8442,
      "step": 21
    },
    {
      "epoch": 0.014372039849746856,
      "grad_norm": Infinity,
      "learning_rate": 8.800000000000001e-05,
      "loss": 2.6257,
      "step": 22
    },
    {
      "epoch": 0.015025314388371713,
      "grad_norm": Infinity,
      "learning_rate": 9.200000000000001e-05,
      "loss": 3.0521,
      "step": 23
    },
    {
      "epoch": 0.01567858892699657,
      "grad_norm": Infinity,
      "learning_rate": 9.6e-05,
      "loss": 3.4209,
      "step": 24
    },
    {
      "epoch": 0.01633186346562143,
      "grad_norm": Infinity,
      "learning_rate": 0.0001,
      "loss": 4.3767,
      "step": 25
    },
    {
      "epoch": 0.016985138004246284,
      "grad_norm": 1.6052595987134284e+19,
      "learning_rate": 0.00010400000000000001,
      "loss": 1.2769,
      "step": 26
    },
    {
      "epoch": 0.017638412542871143,
      "grad_norm": 1.6565584132189454e+19,
      "learning_rate": 0.00010800000000000001,
      "loss": 1.3066,
      "step": 27
    },
    {
      "epoch": 0.018291687081496,
      "grad_norm": Infinity,
      "learning_rate": 0.00011200000000000001,
      "loss": 1.3591,
      "step": 28
    },
    {
      "epoch": 0.018944961620120857,
      "grad_norm": Infinity,
      "learning_rate": 0.000116,
      "loss": 1.4514,
      "step": 29
    },
    {
      "epoch": 0.019598236158745713,
      "grad_norm": Infinity,
      "learning_rate": 0.00012,
      "loss": 1.4569,
      "step": 30
    },
    {
      "epoch": 0.02025151069737057,
      "grad_norm": Infinity,
      "learning_rate": 0.000124,
      "loss": 1.4416,
      "step": 31
    },
    {
      "epoch": 0.020904785235995427,
      "grad_norm": Infinity,
      "learning_rate": 0.00012800000000000002,
      "loss": 1.5548,
      "step": 32
    },
    {
      "epoch": 0.021558059774620286,
      "grad_norm": Infinity,
      "learning_rate": 0.000132,
      "loss": 1.4983,
      "step": 33
    },
    {
      "epoch": 0.02221133431324514,
      "grad_norm": Infinity,
      "learning_rate": 0.00013600000000000003,
      "loss": 1.5397,
      "step": 34
    },
    {
      "epoch": 0.02286460885187,
      "grad_norm": Infinity,
      "learning_rate": 0.00014,
      "loss": 1.5866,
      "step": 35
    },
    {
      "epoch": 0.023517883390494855,
      "grad_norm": Infinity,
      "learning_rate": 0.000144,
      "loss": 1.7101,
      "step": 36
    },
    {
      "epoch": 0.024171157929119714,
      "grad_norm": Infinity,
      "learning_rate": 0.000148,
      "loss": 1.7309,
      "step": 37
    },
    {
      "epoch": 0.02482443246774457,
      "grad_norm": Infinity,
      "learning_rate": 0.000152,
      "loss": 1.8801,
      "step": 38
    },
    {
      "epoch": 0.025477707006369428,
      "grad_norm": Infinity,
      "learning_rate": 0.00015600000000000002,
      "loss": 1.9034,
      "step": 39
    },
    {
      "epoch": 0.026130981544994283,
      "grad_norm": Infinity,
      "learning_rate": 0.00016,
      "loss": 1.8048,
      "step": 40
    },
    {
      "epoch": 0.026784256083619142,
      "grad_norm": Infinity,
      "learning_rate": 0.000164,
      "loss": 2.1254,
      "step": 41
    },
    {
      "epoch": 0.027437530622243998,
      "grad_norm": Infinity,
      "learning_rate": 0.000168,
      "loss": 2.3971,
      "step": 42
    },
    {
      "epoch": 0.028090805160868856,
      "grad_norm": Infinity,
      "learning_rate": 0.000172,
      "loss": 2.7006,
      "step": 43
    },
    {
      "epoch": 0.02874407969949371,
      "grad_norm": Infinity,
      "learning_rate": 0.00017600000000000002,
      "loss": 2.474,
      "step": 44
    },
    {
      "epoch": 0.02939735423811857,
      "grad_norm": Infinity,
      "learning_rate": 0.00018,
      "loss": 2.9188,
      "step": 45
    },
    {
      "epoch": 0.030050628776743426,
      "grad_norm": Infinity,
      "learning_rate": 0.00018400000000000003,
      "loss": 3.0924,
      "step": 46
    },
    {
      "epoch": 0.030703903315368285,
      "grad_norm": Infinity,
      "learning_rate": 0.000188,
      "loss": 3.0522,
      "step": 47
    },
    {
      "epoch": 0.03135717785399314,
      "grad_norm": Infinity,
      "learning_rate": 0.000192,
      "loss": 3.1227,
      "step": 48
    },
    {
      "epoch": 0.032010452392617995,
      "grad_norm": Infinity,
      "learning_rate": 0.000196,
      "loss": 3.6011,
      "step": 49
    },
    {
      "epoch": 0.03266372693124286,
      "grad_norm": Infinity,
      "learning_rate": 0.0002,
      "loss": 4.8325,
      "step": 50
    },
    {
      "epoch": 0.03331700146986771,
      "grad_norm": Infinity,
      "learning_rate": 0.00019999977470780007,
      "loss": 1.2339,
      "step": 51
    },
    {
      "epoch": 0.03397027600849257,
      "grad_norm": Infinity,
      "learning_rate": 0.00019999909883221535,
      "loss": 1.3726,
      "step": 52
    },
    {
      "epoch": 0.034623550547117424,
      "grad_norm": Infinity,
      "learning_rate": 0.0001999979723762913,
      "loss": 1.4039,
      "step": 53
    },
    {
      "epoch": 0.035276825085742286,
      "grad_norm": Infinity,
      "learning_rate": 0.00019999639534510347,
      "loss": 1.3955,
      "step": 54
    },
    {
      "epoch": 0.03593009962436714,
      "grad_norm": Infinity,
      "learning_rate": 0.0001999943677457578,
      "loss": 1.4902,
      "step": 55
    },
    {
      "epoch": 0.036583374162992,
      "grad_norm": Infinity,
      "learning_rate": 0.00019999188958739027,
      "loss": 1.4463,
      "step": 56
    },
    {
      "epoch": 0.03723664870161685,
      "grad_norm": Infinity,
      "learning_rate": 0.00019998896088116715,
      "loss": 1.5482,
      "step": 57
    },
    {
      "epoch": 0.037889923240241714,
      "grad_norm": Infinity,
      "learning_rate": 0.00019998558164028465,
      "loss": 1.5274,
      "step": 58
    },
    {
      "epoch": 0.03854319777886657,
      "grad_norm": Infinity,
      "learning_rate": 0.00019998175187996916,
      "loss": 1.5272,
      "step": 59
    },
    {
      "epoch": 0.039196472317491425,
      "grad_norm": Infinity,
      "learning_rate": 0.00019997747161747695,
      "loss": 1.7031,
      "step": 60
    },
    {
      "epoch": 0.03984974685611628,
      "grad_norm": Infinity,
      "learning_rate": 0.00019997274087209423,
      "loss": 1.6404,
      "step": 61
    },
    {
      "epoch": 0.04050302139474114,
      "grad_norm": Infinity,
      "learning_rate": 0.000199967559665137,
      "loss": 1.8366,
      "step": 62
    },
    {
      "epoch": 0.041156295933366,
      "grad_norm": Infinity,
      "learning_rate": 0.00019996192801995097,
      "loss": 1.8651,
      "step": 63
    },
    {
      "epoch": 0.04180957047199085,
      "grad_norm": Infinity,
      "learning_rate": 0.00019995584596191145,
      "loss": 1.9868,
      "step": 64
    },
    {
      "epoch": 0.04246284501061571,
      "grad_norm": Infinity,
      "learning_rate": 0.00019994931351842327,
      "loss": 2.0719,
      "step": 65
    },
    {
      "epoch": 0.04311611954924057,
      "grad_norm": Infinity,
      "learning_rate": 0.00019994233071892056,
      "loss": 2.1893,
      "step": 66
    },
    {
      "epoch": 0.043769394087865426,
      "grad_norm": Infinity,
      "learning_rate": 0.00019993489759486673,
      "loss": 2.3754,
      "step": 67
    },
    {
      "epoch": 0.04442266862649028,
      "grad_norm": Infinity,
      "learning_rate": 0.00019992701417975427,
      "loss": 2.4019,
      "step": 68
    },
    {
      "epoch": 0.04507594316511514,
      "grad_norm": Infinity,
      "learning_rate": 0.0001999186805091047,
      "loss": 2.5356,
      "step": 69
    },
    {
      "epoch": 0.04572921770374,
      "grad_norm": Infinity,
      "learning_rate": 0.00019990989662046818,
      "loss": 2.8341,
      "step": 70
    },
    {
      "epoch": 0.046382492242364855,
      "grad_norm": Infinity,
      "learning_rate": 0.00019990066255342348,
      "loss": 3.1333,
      "step": 71
    },
    {
      "epoch": 0.04703576678098971,
      "grad_norm": Infinity,
      "learning_rate": 0.00019989097834957799,
      "loss": 2.9525,
      "step": 72
    },
    {
      "epoch": 0.047689041319614565,
      "grad_norm": Infinity,
      "learning_rate": 0.00019988084405256714,
      "loss": 3.3206,
      "step": 73
    },
    {
      "epoch": 0.04834231585823943,
      "grad_norm": Infinity,
      "learning_rate": 0.00019987025970805448,
      "loss": 3.5856,
      "step": 74
    },
    {
      "epoch": 0.04899559039686428,
      "grad_norm": Infinity,
      "learning_rate": 0.00019985922536373146,
      "loss": 3.9854,
      "step": 75
    },
    {
      "epoch": 0.04964886493548914,
      "grad_norm": Infinity,
      "learning_rate": 0.00019984774106931714,
      "loss": 1.2761,
      "step": 76
    },
    {
      "epoch": 0.050302139474113994,
      "grad_norm": Infinity,
      "learning_rate": 0.0001998358068765579,
      "loss": 1.426,
      "step": 77
    },
    {
      "epoch": 0.050955414012738856,
      "grad_norm": Infinity,
      "learning_rate": 0.00019982342283922738,
      "loss": 1.4982,
      "step": 78
    },
    {
      "epoch": 0.05160868855136371,
      "grad_norm": Infinity,
      "learning_rate": 0.00019981058901312606,
      "loss": 1.4369,
      "step": 79
    },
    {
      "epoch": 0.05226196308998857,
      "grad_norm": Infinity,
      "learning_rate": 0.00019979730545608126,
      "loss": 1.4131,
      "step": 80
    },
    {
      "epoch": 0.05291523762861342,
      "grad_norm": Infinity,
      "learning_rate": 0.00019978357222794654,
      "loss": 1.4518,
      "step": 81
    },
    {
      "epoch": 0.053568512167238284,
      "grad_norm": Infinity,
      "learning_rate": 0.00019976938939060172,
      "loss": 1.5077,
      "step": 82
    },
    {
      "epoch": 0.05422178670586314,
      "grad_norm": Infinity,
      "learning_rate": 0.00019975475700795246,
      "loss": 1.5611,
      "step": 83
    },
    {
      "epoch": 0.054875061244487995,
      "grad_norm": Infinity,
      "learning_rate": 0.00019973967514592996,
      "loss": 1.467,
      "step": 84
    },
    {
      "epoch": 0.05552833578311285,
      "grad_norm": Infinity,
      "learning_rate": 0.00019972414387249072,
      "loss": 1.5768,
      "step": 85
    },
    {
      "epoch": 0.05618161032173771,
      "grad_norm": Infinity,
      "learning_rate": 0.00019970816325761627,
      "loss": 1.6242,
      "step": 86
    },
    {
      "epoch": 0.05683488486036257,
      "grad_norm": Infinity,
      "learning_rate": 0.0001996917333733128,
      "loss": 1.7764,
      "step": 87
    },
    {
      "epoch": 0.05748815939898742,
      "grad_norm": Infinity,
      "learning_rate": 0.00019967485429361076,
      "loss": 1.8326,
      "step": 88
    },
    {
      "epoch": 0.05814143393761228,
      "grad_norm": Infinity,
      "learning_rate": 0.00019965752609456464,
      "loss": 1.8779,
      "step": 89
    },
    {
      "epoch": 0.05879470847623714,
      "grad_norm": Infinity,
      "learning_rate": 0.00019963974885425266,
      "loss": 2.1007,
      "step": 90
    },
    {
      "epoch": 0.059447983014861996,
      "grad_norm": Infinity,
      "learning_rate": 0.00019962152265277623,
      "loss": 2.2433,
      "step": 91
    },
    {
      "epoch": 0.06010125755348685,
      "grad_norm": Infinity,
      "learning_rate": 0.0001996028475722598,
      "loss": 2.2192,
      "step": 92
    },
    {
      "epoch": 0.06075453209211171,
      "grad_norm": Infinity,
      "learning_rate": 0.00019958372369685033,
      "loss": 2.1113,
      "step": 93
    },
    {
      "epoch": 0.06140780663073657,
      "grad_norm": Infinity,
      "learning_rate": 0.00019956415111271712,
      "loss": 2.6662,
      "step": 94
    },
    {
      "epoch": 0.062061081169361425,
      "grad_norm": Infinity,
      "learning_rate": 0.00019954412990805107,
      "loss": 2.7781,
      "step": 95
    },
    {
      "epoch": 0.06271435570798628,
      "grad_norm": Infinity,
      "learning_rate": 0.00019952366017306466,
      "loss": 2.7183,
      "step": 96
    },
    {
      "epoch": 0.06336763024661114,
      "grad_norm": Infinity,
      "learning_rate": 0.00019950274199999132,
      "loss": 2.7068,
      "step": 97
    },
    {
      "epoch": 0.06402090478523599,
      "grad_norm": Infinity,
      "learning_rate": 0.00019948137548308502,
      "loss": 3.1326,
      "step": 98
    },
    {
      "epoch": 0.06467417932386085,
      "grad_norm": Infinity,
      "learning_rate": 0.00019945956071862003,
      "loss": 3.1585,
      "step": 99
    },
    {
      "epoch": 0.06532745386248572,
      "grad_norm": Infinity,
      "learning_rate": 0.00019943729780489027,
      "loss": 4.3935,
      "step": 100
    },
    {
      "epoch": 0.06598072840111056,
      "grad_norm": Infinity,
      "learning_rate": 0.0001994145868422089,
      "loss": 1.3574,
      "step": 101
    },
    {
      "epoch": 0.06663400293973543,
      "grad_norm": Infinity,
      "learning_rate": 0.00019939142793290798,
      "loss": 1.3933,
      "step": 102
    },
    {
      "epoch": 0.06728727747836027,
      "grad_norm": Infinity,
      "learning_rate": 0.000199367821181338,
      "loss": 1.4518,
      "step": 103
    },
    {
      "epoch": 0.06794055201698514,
      "grad_norm": Infinity,
      "learning_rate": 0.00019934376669386727,
      "loss": 1.3938,
      "step": 104
    },
    {
      "epoch": 0.06859382655561,
      "grad_norm": Infinity,
      "learning_rate": 0.00019931926457888156,
      "loss": 1.4311,
      "step": 105
    },
    {
      "epoch": 0.06924710109423485,
      "grad_norm": Infinity,
      "learning_rate": 0.00019929431494678356,
      "loss": 1.4887,
      "step": 106
    },
    {
      "epoch": 0.06990037563285971,
      "grad_norm": Infinity,
      "learning_rate": 0.00019926891790999243,
      "loss": 1.5323,
      "step": 107
    },
    {
      "epoch": 0.07055365017148457,
      "grad_norm": Infinity,
      "learning_rate": 0.00019924307358294322,
      "loss": 1.5969,
      "step": 108
    },
    {
      "epoch": 0.07120692471010942,
      "grad_norm": Infinity,
      "learning_rate": 0.00019921678208208654,
      "loss": 1.5895,
      "step": 109
    },
    {
      "epoch": 0.07186019924873428,
      "grad_norm": Infinity,
      "learning_rate": 0.00019919004352588767,
      "loss": 1.5687,
      "step": 110
    },
    {
      "epoch": 0.07251347378735913,
      "grad_norm": Infinity,
      "learning_rate": 0.00019916285803482647,
      "loss": 1.7395,
      "step": 111
    },
    {
      "epoch": 0.073166748325984,
      "grad_norm": Infinity,
      "learning_rate": 0.0001991352257313965,
      "loss": 1.8653,
      "step": 112
    },
    {
      "epoch": 0.07382002286460886,
      "grad_norm": Infinity,
      "learning_rate": 0.00019910714674010454,
      "loss": 1.8805,
      "step": 113
    },
    {
      "epoch": 0.0744732974032337,
      "grad_norm": Infinity,
      "learning_rate": 0.00019907862118747022,
      "loss": 1.9816,
      "step": 114
    },
    {
      "epoch": 0.07512657194185857,
      "grad_norm": Infinity,
      "learning_rate": 0.0001990496492020252,
      "loss": 2.0509,
      "step": 115
    },
    {
      "epoch": 0.07577984648048343,
      "grad_norm": Infinity,
      "learning_rate": 0.0001990202309143127,
      "loss": 2.2874,
      "step": 116
    },
    {
      "epoch": 0.07643312101910828,
      "grad_norm": Infinity,
      "learning_rate": 0.000198990366456887,
      "loss": 2.2876,
      "step": 117
    },
    {
      "epoch": 0.07708639555773314,
      "grad_norm": Infinity,
      "learning_rate": 0.00019896005596431264,
      "loss": 2.5601,
      "step": 118
    },
    {
      "epoch": 0.07773967009635799,
      "grad_norm": Infinity,
      "learning_rate": 0.00019892929957316397,
      "loss": 2.3229,
      "step": 119
    },
    {
      "epoch": 0.07839294463498285,
      "grad_norm": Infinity,
      "learning_rate": 0.00019889809742202455,
      "loss": 2.4374,
      "step": 120
    },
    {
      "epoch": 0.07904621917360771,
      "grad_norm": Infinity,
      "learning_rate": 0.0001988664496514863,
      "loss": 2.6645,
      "step": 121
    },
    {
      "epoch": 0.07969949371223256,
      "grad_norm": Infinity,
      "learning_rate": 0.00019883435640414922,
      "loss": 2.954,
      "step": 122
    },
    {
      "epoch": 0.08035276825085742,
      "grad_norm": Infinity,
      "learning_rate": 0.0001988018178246205,
      "loss": 3.252,
      "step": 123
    },
    {
      "epoch": 0.08100604278948229,
      "grad_norm": Infinity,
      "learning_rate": 0.00019876883405951377,
      "loss": 3.2595,
      "step": 124
    },
    {
      "epoch": 0.08165931732810713,
      "grad_norm": Infinity,
      "learning_rate": 0.00019873540525744887,
      "loss": 4.0661,
      "step": 125
    },
    {
      "epoch": 0.082312591866732,
      "grad_norm": Infinity,
      "learning_rate": 0.00019870153156905068,
      "loss": 1.2808,
      "step": 126
    },
    {
      "epoch": 0.08296586640535684,
      "grad_norm": Infinity,
      "learning_rate": 0.00019866721314694882,
      "loss": 1.3348,
      "step": 127
    },
    {
      "epoch": 0.0836191409439817,
      "grad_norm": Infinity,
      "learning_rate": 0.00019863245014577668,
      "loss": 1.3889,
      "step": 128
    },
    {
      "epoch": 0.08427241548260657,
      "grad_norm": Infinity,
      "learning_rate": 0.00019859724272217099,
      "loss": 1.3553,
      "step": 129
    },
    {
      "epoch": 0.08492569002123142,
      "grad_norm": Infinity,
      "learning_rate": 0.00019856159103477086,
      "loss": 1.4116,
      "step": 130
    },
    {
      "epoch": 0.08557896455985628,
      "grad_norm": Infinity,
      "learning_rate": 0.00019852549524421723,
      "loss": 1.5371,
      "step": 131
    },
    {
      "epoch": 0.08623223909848114,
      "grad_norm": Infinity,
      "learning_rate": 0.0001984889555131521,
      "loss": 1.5529,
      "step": 132
    },
    {
      "epoch": 0.08688551363710599,
      "grad_norm": Infinity,
      "learning_rate": 0.00019845197200621785,
      "loss": 1.5542,
      "step": 133
    },
    {
      "epoch": 0.08753878817573085,
      "grad_norm": Infinity,
      "learning_rate": 0.00019841454489005636,
      "loss": 1.5355,
      "step": 134
    },
    {
      "epoch": 0.0881920627143557,
      "grad_norm": Infinity,
      "learning_rate": 0.00019837667433330838,
      "loss": 1.5569,
      "step": 135
    },
    {
      "epoch": 0.08884533725298056,
      "grad_norm": Infinity,
      "learning_rate": 0.0001983383605066127,
      "loss": 1.6497,
      "step": 136
    },
    {
      "epoch": 0.08949861179160543,
      "grad_norm": Infinity,
      "learning_rate": 0.00019829960358260545,
      "loss": 1.8195,
      "step": 137
    },
    {
      "epoch": 0.09015188633023027,
      "grad_norm": Infinity,
      "learning_rate": 0.00019826040373591933,
      "loss": 1.9526,
      "step": 138
    },
    {
      "epoch": 0.09080516086885514,
      "grad_norm": Infinity,
      "learning_rate": 0.0001982207611431827,
      "loss": 2.011,
      "step": 139
    },
    {
      "epoch": 0.09145843540748,
      "grad_norm": Infinity,
      "learning_rate": 0.0001981806759830189,
      "loss": 2.1913,
      "step": 140
    },
    {
      "epoch": 0.09211170994610485,
      "grad_norm": Infinity,
      "learning_rate": 0.00019814014843604543,
      "loss": 2.1102,
      "step": 141
    },
    {
      "epoch": 0.09276498448472971,
      "grad_norm": Infinity,
      "learning_rate": 0.00019809917868487308,
      "loss": 2.3621,
      "step": 142
    },
    {
      "epoch": 0.09341825902335456,
      "grad_norm": Infinity,
      "learning_rate": 0.00019805776691410516,
      "loss": 2.2622,
      "step": 143
    },
    {
      "epoch": 0.09407153356197942,
      "grad_norm": Infinity,
      "learning_rate": 0.00019801591331033663,
      "loss": 2.7022,
      "step": 144
    },
    {
      "epoch": 0.09472480810060428,
      "grad_norm": Infinity,
      "learning_rate": 0.00019797361806215332,
      "loss": 2.9089,
      "step": 145
    },
    {
      "epoch": 0.09537808263922913,
      "grad_norm": Infinity,
      "learning_rate": 0.000197930881360131,
      "loss": 2.8431,
      "step": 146
    },
    {
      "epoch": 0.096031357177854,
      "grad_norm": Infinity,
      "learning_rate": 0.00019788770339683462,
      "loss": 3.0716,
      "step": 147
    },
    {
      "epoch": 0.09668463171647886,
      "grad_norm": Infinity,
      "learning_rate": 0.00019784408436681732,
      "loss": 2.9731,
      "step": 148
    },
    {
      "epoch": 0.0973379062551037,
      "grad_norm": Infinity,
      "learning_rate": 0.00019780002446661966,
      "loss": 3.4067,
      "step": 149
    },
    {
      "epoch": 0.09799118079372857,
      "grad_norm": Infinity,
      "learning_rate": 0.00019775552389476864,
      "loss": 4.0305,
      "step": 150
    },
    {
      "epoch": 0.09864445533235343,
      "grad_norm": Infinity,
      "learning_rate": 0.0001977105828517769,
      "loss": 1.3048,
      "step": 151
    },
    {
      "epoch": 0.09929772987097828,
      "grad_norm": Infinity,
      "learning_rate": 0.00019766520154014183,
      "loss": 1.3672,
      "step": 152
    },
    {
      "epoch": 0.09995100440960314,
      "grad_norm": Infinity,
      "learning_rate": 0.00019761938016434448,
      "loss": 1.3763,
      "step": 153
    },
    {
      "epoch": 0.10060427894822799,
      "grad_norm": Infinity,
      "learning_rate": 0.00019757311893084885,
      "loss": 1.3836,
      "step": 154
    },
    {
      "epoch": 0.10125755348685285,
      "grad_norm": Infinity,
      "learning_rate": 0.00019752641804810084,
      "loss": 1.4765,
      "step": 155
    },
    {
      "epoch": 0.10191082802547771,
      "grad_norm": Infinity,
      "learning_rate": 0.0001974792777265273,
      "loss": 1.6151,
      "step": 156
    },
    {
      "epoch": 0.10256410256410256,
      "grad_norm": Infinity,
      "learning_rate": 0.00019743169817853525,
      "loss": 1.5203,
      "step": 157
    },
    {
      "epoch": 0.10321737710272742,
      "grad_norm": Infinity,
      "learning_rate": 0.00019738367961851064,
      "loss": 1.5259,
      "step": 158
    },
    {
      "epoch": 0.10387065164135229,
      "grad_norm": Infinity,
      "learning_rate": 0.0001973352222628176,
      "loss": 1.6711,
      "step": 159
    },
    {
      "epoch": 0.10452392617997713,
      "grad_norm": Infinity,
      "learning_rate": 0.00019728632632979746,
      "loss": 1.7773,
      "step": 160
    },
    {
      "epoch": 0.105177200718602,
      "grad_norm": Infinity,
      "learning_rate": 0.00019723699203976766,
      "loss": 1.7367,
      "step": 161
    },
    {
      "epoch": 0.10583047525722684,
      "grad_norm": Infinity,
      "learning_rate": 0.0001971872196150208,
      "loss": 1.8355,
      "step": 162
    },
    {
      "epoch": 0.1064837497958517,
      "grad_norm": Infinity,
      "learning_rate": 0.00019713700927982372,
      "loss": 1.8341,
      "step": 163
    },
    {
      "epoch": 0.10713702433447657,
      "grad_norm": Infinity,
      "learning_rate": 0.0001970863612604162,
      "loss": 1.8942,
      "step": 164
    },
    {
      "epoch": 0.10779029887310142,
      "grad_norm": Infinity,
      "learning_rate": 0.0001970352757850105,
      "loss": 2.0185,
      "step": 165
    },
    {
      "epoch": 0.10844357341172628,
      "grad_norm": Infinity,
      "learning_rate": 0.00019698375308378974,
      "loss": 2.1895,
      "step": 166
    },
    {
      "epoch": 0.10909684795035114,
      "grad_norm": Infinity,
      "learning_rate": 0.0001969317933889071,
      "loss": 2.3008,
      "step": 167
    },
    {
      "epoch": 0.10975012248897599,
      "grad_norm": Infinity,
      "learning_rate": 0.00019687939693448494,
      "loss": 2.5866,
      "step": 168
    },
    {
      "epoch": 0.11040339702760085,
      "grad_norm": Infinity,
      "learning_rate": 0.0001968265639566135,
      "loss": 2.3347,
      "step": 169
    },
    {
      "epoch": 0.1110566715662257,
      "grad_norm": Infinity,
      "learning_rate": 0.0001967732946933499,
      "loss": 2.5208,
      "step": 170
    },
    {
      "epoch": 0.11170994610485056,
      "grad_norm": Infinity,
      "learning_rate": 0.00019671958938471715,
      "loss": 2.7925,
      "step": 171
    },
    {
      "epoch": 0.11236322064347543,
      "grad_norm": Infinity,
      "learning_rate": 0.000196665448272703,
      "loss": 3.3472,
      "step": 172
    },
    {
      "epoch": 0.11301649518210027,
      "grad_norm": Infinity,
      "learning_rate": 0.00019661087160125886,
      "loss": 3.157,
      "step": 173
    },
    {
      "epoch": 0.11366976972072514,
      "grad_norm": Infinity,
      "learning_rate": 0.00019655585961629867,
      "loss": 3.5798,
      "step": 174
    },
    {
      "epoch": 0.11432304425935,
      "grad_norm": Infinity,
      "learning_rate": 0.00019650041256569792,
      "loss": 4.4552,
      "step": 175
    },
    {
      "epoch": 0.11497631879797485,
      "grad_norm": Infinity,
      "learning_rate": 0.00019644453069929228,
      "loss": 1.1694,
      "step": 176
    },
    {
      "epoch": 0.11562959333659971,
      "grad_norm": Infinity,
      "learning_rate": 0.00019638821426887673,
      "loss": 1.3668,
      "step": 177
    },
    {
      "epoch": 0.11628286787522456,
      "grad_norm": Infinity,
      "learning_rate": 0.0001963314635282044,
      "loss": 1.4632,
      "step": 178
    },
    {
      "epoch": 0.11693614241384942,
      "grad_norm": Infinity,
      "learning_rate": 0.0001962742787329852,
      "loss": 1.4713,
      "step": 179
    },
    {
      "epoch": 0.11758941695247428,
      "grad_norm": Infinity,
      "learning_rate": 0.00019621666014088494,
      "loss": 1.4127,
      "step": 180
    },
    {
      "epoch": 0.11824269149109913,
      "grad_norm": Infinity,
      "learning_rate": 0.00019615860801152398,
      "loss": 1.5774,
      "step": 181
    },
    {
      "epoch": 0.11889596602972399,
      "grad_norm": Infinity,
      "learning_rate": 0.00019610012260647618,
      "loss": 1.4787,
      "step": 182
    },
    {
      "epoch": 0.11954924056834886,
      "grad_norm": Infinity,
      "learning_rate": 0.00019604120418926764,
      "loss": 1.6504,
      "step": 183
    },
    {
      "epoch": 0.1202025151069737,
      "grad_norm": Infinity,
      "learning_rate": 0.00019598185302537556,
      "loss": 1.6627,
      "step": 184
    },
    {
      "epoch": 0.12085578964559857,
      "grad_norm": Infinity,
      "learning_rate": 0.00019592206938222703,
      "loss": 1.7108,
      "step": 185
    },
    {
      "epoch": 0.12150906418422341,
      "grad_norm": Infinity,
      "learning_rate": 0.0001958618535291978,
      "loss": 1.7505,
      "step": 186
    },
    {
      "epoch": 0.12216233872284828,
      "grad_norm": Infinity,
      "learning_rate": 0.00019580120573761112,
      "loss": 1.7437,
      "step": 187
    },
    {
      "epoch": 0.12281561326147314,
      "grad_norm": Infinity,
      "learning_rate": 0.00019574012628073648,
      "loss": 1.7792,
      "step": 188
    },
    {
      "epoch": 0.12346888780009799,
      "grad_norm": Infinity,
      "learning_rate": 0.00019567861543378837,
      "loss": 1.6856,
      "step": 189
    },
    {
      "epoch": 0.12412216233872285,
      "grad_norm": Infinity,
      "learning_rate": 0.00019561667347392508,
      "loss": 2.2671,
      "step": 190
    },
    {
      "epoch": 0.12477543687734771,
      "grad_norm": Infinity,
      "learning_rate": 0.00019555430068024748,
      "loss": 2.0929,
      "step": 191
    },
    {
      "epoch": 0.12542871141597256,
      "grad_norm": Infinity,
      "learning_rate": 0.00019549149733379755,
      "loss": 2.4075,
      "step": 192
    },
    {
      "epoch": 0.1260819859545974,
      "grad_norm": Infinity,
      "learning_rate": 0.00019542826371755743,
      "loss": 2.4512,
      "step": 193
    },
    {
      "epoch": 0.12673526049322228,
      "grad_norm": Infinity,
      "learning_rate": 0.0001953646001164479,
      "loss": 2.6113,
      "step": 194
    },
    {
      "epoch": 0.12738853503184713,
      "grad_norm": Infinity,
      "learning_rate": 0.0001953005068173272,
      "loss": 2.7185,
      "step": 195
    },
    {
      "epoch": 0.12804180957047198,
      "grad_norm": Infinity,
      "learning_rate": 0.0001952359841089898,
      "loss": 2.7951,
      "step": 196
    },
    {
      "epoch": 0.12869508410909686,
      "grad_norm": Infinity,
      "learning_rate": 0.00019517103228216493,
      "loss": 2.7111,
      "step": 197
    },
    {
      "epoch": 0.1293483586477217,
      "grad_norm": Infinity,
      "learning_rate": 0.00019510565162951537,
      "loss": 2.8432,
      "step": 198
    },
    {
      "epoch": 0.13000163318634655,
      "grad_norm": Infinity,
      "learning_rate": 0.00019503984244563616,
      "loss": 3.8419,
      "step": 199
    },
    {
      "epoch": 0.13065490772497143,
      "grad_norm": Infinity,
      "learning_rate": 0.0001949736050270532,
      "loss": 4.3049,
      "step": 200
    },
    {
      "epoch": 0.13130818226359628,
      "grad_norm": Infinity,
      "learning_rate": 0.00019490693967222199,
      "loss": 1.308,
      "step": 201
    },
    {
      "epoch": 0.13196145680222113,
      "grad_norm": Infinity,
      "learning_rate": 0.00019483984668152617,
      "loss": 1.2917,
      "step": 202
    },
    {
      "epoch": 0.132614731340846,
      "grad_norm": Infinity,
      "learning_rate": 0.00019477232635727637,
      "loss": 1.47,
      "step": 203
    },
    {
      "epoch": 0.13326800587947085,
      "grad_norm": Infinity,
      "learning_rate": 0.00019470437900370857,
      "loss": 1.3956,
      "step": 204
    },
    {
      "epoch": 0.1339212804180957,
      "grad_norm": Infinity,
      "learning_rate": 0.00019463600492698296,
      "loss": 1.4451,
      "step": 205
    },
    {
      "epoch": 0.13457455495672055,
      "grad_norm": Infinity,
      "learning_rate": 0.00019456720443518247,
      "loss": 1.45,
      "step": 206
    },
    {
      "epoch": 0.13522782949534543,
      "grad_norm": Infinity,
      "learning_rate": 0.0001944979778383114,
      "loss": 1.5642,
      "step": 207
    },
    {
      "epoch": 0.13588110403397027,
      "grad_norm": Infinity,
      "learning_rate": 0.00019442832544829398,
      "loss": 1.4935,
      "step": 208
    },
    {
      "epoch": 0.13653437857259512,
      "grad_norm": Infinity,
      "learning_rate": 0.000194358247578973,
      "loss": 1.6163,
      "step": 209
    },
    {
      "epoch": 0.13718765311122,
      "grad_norm": Infinity,
      "learning_rate": 0.00019428774454610843,
      "loss": 1.6751,
      "step": 210
    },
    {
      "epoch": 0.13784092764984485,
      "grad_norm": Infinity,
      "learning_rate": 0.00019421681666737594,
      "loss": 1.714,
      "step": 211
    },
    {
      "epoch": 0.1384942021884697,
      "grad_norm": Infinity,
      "learning_rate": 0.00019414546426236543,
      "loss": 1.8134,
      "step": 212
    },
    {
      "epoch": 0.13914747672709457,
      "grad_norm": Infinity,
      "learning_rate": 0.00019407368765257977,
      "loss": 1.7647,
      "step": 213
    },
    {
      "epoch": 0.13980075126571942,
      "grad_norm": Infinity,
      "learning_rate": 0.00019400148716143317,
      "loss": 1.8783,
      "step": 214
    },
    {
      "epoch": 0.14045402580434427,
      "grad_norm": Infinity,
      "learning_rate": 0.00019392886311424973,
      "loss": 1.9749,
      "step": 215
    },
    {
      "epoch": 0.14110730034296914,
      "grad_norm": Infinity,
      "learning_rate": 0.00019385581583826212,
      "loss": 2.1572,
      "step": 216
    },
    {
      "epoch": 0.141760574881594,
      "grad_norm": Infinity,
      "learning_rate": 0.00019378234566260995,
      "loss": 2.2505,
      "step": 217
    },
    {
      "epoch": 0.14241384942021884,
      "grad_norm": Infinity,
      "learning_rate": 0.00019370845291833837,
      "loss": 2.4754,
      "step": 218
    },
    {
      "epoch": 0.14306712395884372,
      "grad_norm": Infinity,
      "learning_rate": 0.00019363413793839658,
      "loss": 2.6163,
      "step": 219
    },
    {
      "epoch": 0.14372039849746857,
      "grad_norm": Infinity,
      "learning_rate": 0.0001935594010576362,
      "loss": 2.4796,
      "step": 220
    },
    {
      "epoch": 0.14437367303609341,
      "grad_norm": Infinity,
      "learning_rate": 0.0001934842426128101,
      "loss": 2.7531,
      "step": 221
    },
    {
      "epoch": 0.14502694757471826,
      "grad_norm": Infinity,
      "learning_rate": 0.00019340866294257042,
      "loss": 2.954,
      "step": 222
    },
    {
      "epoch": 0.14568022211334314,
      "grad_norm": Infinity,
      "learning_rate": 0.00019333266238746736,
      "loss": 2.8526,
      "step": 223
    },
    {
      "epoch": 0.146333496651968,
      "grad_norm": Infinity,
      "learning_rate": 0.0001932562412899476,
      "loss": 3.3364,
      "step": 224
    },
    {
      "epoch": 0.14698677119059284,
      "grad_norm": Infinity,
      "learning_rate": 0.0001931793999943526,
      "loss": 4.5349,
      "step": 225
    },
    {
      "epoch": 0.1476400457292177,
      "grad_norm": Infinity,
      "learning_rate": 0.0001931021388469174,
      "loss": 1.228,
      "step": 226
    },
    {
      "epoch": 0.14829332026784256,
      "grad_norm": Infinity,
      "learning_rate": 0.00019302445819576855,
      "loss": 1.3118,
      "step": 227
    },
    {
      "epoch": 0.1489465948064674,
      "grad_norm": Infinity,
      "learning_rate": 0.000192946358390923,
      "loss": 1.3919,
      "step": 228
    },
    {
      "epoch": 0.14959986934509228,
      "grad_norm": Infinity,
      "learning_rate": 0.00019286783978428624,
      "loss": 1.5169,
      "step": 229
    },
    {
      "epoch": 0.15025314388371713,
      "grad_norm": Infinity,
      "learning_rate": 0.00019278890272965096,
      "loss": 1.4527,
      "step": 230
    },
    {
      "epoch": 0.15090641842234198,
      "grad_norm": Infinity,
      "learning_rate": 0.00019270954758269512,
      "loss": 1.5709,
      "step": 231
    },
    {
      "epoch": 0.15155969296096686,
      "grad_norm": Infinity,
      "learning_rate": 0.00019262977470098065,
      "loss": 1.5521,
      "step": 232
    },
    {
      "epoch": 0.1522129674995917,
      "grad_norm": Infinity,
      "learning_rate": 0.00019254958444395173,
      "loss": 1.6832,
      "step": 233
    },
    {
      "epoch": 0.15286624203821655,
      "grad_norm": Infinity,
      "learning_rate": 0.00019246897717293315,
      "loss": 1.6709,
      "step": 234
    },
    {
      "epoch": 0.15351951657684143,
      "grad_norm": Infinity,
      "learning_rate": 0.0001923879532511287,
      "loss": 1.6955,
      "step": 235
    },
    {
      "epoch": 0.15417279111546628,
      "grad_norm": Infinity,
      "learning_rate": 0.0001923065130436195,
      "loss": 1.6892,
      "step": 236
    },
    {
      "epoch": 0.15482606565409113,
      "grad_norm": Infinity,
      "learning_rate": 0.00019222465691736247,
      "loss": 1.8645,
      "step": 237
    },
    {
      "epoch": 0.15547934019271598,
      "grad_norm": Infinity,
      "learning_rate": 0.0001921423852411885,
      "loss": 1.8915,
      "step": 238
    },
    {
      "epoch": 0.15613261473134085,
      "grad_norm": Infinity,
      "learning_rate": 0.00019205969838580094,
      "loss": 1.8416,
      "step": 239
    },
    {
      "epoch": 0.1567858892699657,
      "grad_norm": Infinity,
      "learning_rate": 0.0001919765967237739,
      "loss": 1.8057,
      "step": 240
    },
    {
      "epoch": 0.15743916380859055,
      "grad_norm": Infinity,
      "learning_rate": 0.00019189308062955043,
      "loss": 1.9947,
      "step": 241
    },
    {
      "epoch": 0.15809243834721542,
      "grad_norm": Infinity,
      "learning_rate": 0.00019180915047944112,
      "loss": 2.3449,
      "step": 242
    },
    {
      "epoch": 0.15874571288584027,
      "grad_norm": Infinity,
      "learning_rate": 0.000191724806651622,
      "loss": 2.4693,
      "step": 243
    },
    {
      "epoch": 0.15939898742446512,
      "grad_norm": Infinity,
      "learning_rate": 0.00019164004952613336,
      "loss": 2.3417,
      "step": 244
    },
    {
      "epoch": 0.16005226196309,
      "grad_norm": Infinity,
      "learning_rate": 0.00019155487948487748,
      "loss": 2.2969,
      "step": 245
    },
    {
      "epoch": 0.16070553650171485,
      "grad_norm": Infinity,
      "learning_rate": 0.00019146929691161727,
      "loss": 2.8115,
      "step": 246
    },
    {
      "epoch": 0.1613588110403397,
      "grad_norm": Infinity,
      "learning_rate": 0.0001913833021919745,
      "loss": 2.8346,
      "step": 247
    },
    {
      "epoch": 0.16201208557896457,
      "grad_norm": Infinity,
      "learning_rate": 0.00019129689571342797,
      "loss": 3.143,
      "step": 248
    },
    {
      "epoch": 0.16266536011758942,
      "grad_norm": Infinity,
      "learning_rate": 0.00019121007786531178,
      "loss": 3.4042,
      "step": 249
    },
    {
      "epoch": 0.16331863465621427,
      "grad_norm": Infinity,
      "learning_rate": 0.0001911228490388136,
      "loss": 3.7222,
      "step": 250
    },
    {
      "epoch": 0.16397190919483914,
      "grad_norm": Infinity,
      "learning_rate": 0.0001910352096269729,
      "loss": 1.2778,
      "step": 251
    },
    {
      "epoch": 0.164625183733464,
      "grad_norm": Infinity,
      "learning_rate": 0.00019094716002467918,
      "loss": 1.3826,
      "step": 252
    },
    {
      "epoch": 0.16527845827208884,
      "grad_norm": Infinity,
      "learning_rate": 0.0001908587006286703,
      "loss": 1.3929,
      "step": 253
    },
    {
      "epoch": 0.1659317328107137,
      "grad_norm": Infinity,
      "learning_rate": 0.00019076983183753045,
      "loss": 1.4139,
      "step": 254
    },
    {
      "epoch": 0.16658500734933857,
      "grad_norm": Infinity,
      "learning_rate": 0.0001906805540516885,
      "loss": 1.3002,
      "step": 255
    },
    {
      "epoch": 0.1672382818879634,
      "grad_norm": Infinity,
      "learning_rate": 0.00019059086767341627,
      "loss": 1.414,
      "step": 256
    },
    {
      "epoch": 0.16789155642658826,
      "grad_norm": Infinity,
      "learning_rate": 0.00019050077310682657,
      "loss": 1.6117,
      "step": 257
    },
    {
      "epoch": 0.16854483096521314,
      "grad_norm": Infinity,
      "learning_rate": 0.0001904102707578715,
      "loss": 1.5172,
      "step": 258
    },
    {
      "epoch": 0.169198105503838,
      "grad_norm": Infinity,
      "learning_rate": 0.00019031936103434044,
      "loss": 1.6757,
      "step": 259
    },
    {
      "epoch": 0.16985138004246284,
      "grad_norm": Infinity,
      "learning_rate": 0.00019022804434585852,
      "loss": 1.6502,
      "step": 260
    },
    {
      "epoch": 0.1705046545810877,
      "grad_norm": Infinity,
      "learning_rate": 0.00019013632110388446,
      "loss": 1.6107,
      "step": 261
    },
    {
      "epoch": 0.17115792911971256,
      "grad_norm": Infinity,
      "learning_rate": 0.00019004419172170887,
      "loss": 1.6386,
      "step": 262
    },
    {
      "epoch": 0.1718112036583374,
      "grad_norm": Infinity,
      "learning_rate": 0.00018995165661445234,
      "loss": 1.7794,
      "step": 263
    },
    {
      "epoch": 0.17246447819696228,
      "grad_norm": Infinity,
      "learning_rate": 0.0001898587161990637,
      "loss": 1.8321,
      "step": 264
    },
    {
      "epoch": 0.17311775273558713,
      "grad_norm": Infinity,
      "learning_rate": 0.0001897653708943179,
      "loss": 1.9069,
      "step": 265
    },
    {
      "epoch": 0.17377102727421198,
      "grad_norm": Infinity,
      "learning_rate": 0.00018967162112081438,
      "loss": 2.1029,
      "step": 266
    },
    {
      "epoch": 0.17442430181283686,
      "grad_norm": Infinity,
      "learning_rate": 0.00018957746730097495,
      "loss": 2.2505,
      "step": 267
    },
    {
      "epoch": 0.1750775763514617,
      "grad_norm": Infinity,
      "learning_rate": 0.00018948290985904204,
      "loss": 2.4312,
      "step": 268
    },
    {
      "epoch": 0.17573085089008655,
      "grad_norm": Infinity,
      "learning_rate": 0.00018938794922107675,
      "loss": 2.544,
      "step": 269
    },
    {
      "epoch": 0.1763841254287114,
      "grad_norm": Infinity,
      "learning_rate": 0.00018929258581495685,
      "loss": 2.8027,
      "step": 270
    },
    {
      "epoch": 0.17703739996733628,
      "grad_norm": Infinity,
      "learning_rate": 0.00018919682007037506,
      "loss": 2.7801,
      "step": 271
    },
    {
      "epoch": 0.17769067450596113,
      "grad_norm": Infinity,
      "learning_rate": 0.0001891006524188368,
      "loss": 3.1716,
      "step": 272
    },
    {
      "epoch": 0.17834394904458598,
      "grad_norm": Infinity,
      "learning_rate": 0.00018900408329365856,
      "loss": 2.8948,
      "step": 273
    },
    {
      "epoch": 0.17899722358321085,
      "grad_norm": Infinity,
      "learning_rate": 0.0001889071131299657,
      "loss": 3.2573,
      "step": 274
    },
    {
      "epoch": 0.1796504981218357,
      "grad_norm": Infinity,
      "learning_rate": 0.0001888097423646907,
      "loss": 4.1261,
      "step": 275
    },
    {
      "epoch": 0.18030377266046055,
      "grad_norm": Infinity,
      "learning_rate": 0.00018871197143657104,
      "loss": 1.2702,
      "step": 276
    },
    {
      "epoch": 0.18095704719908542,
      "grad_norm": Infinity,
      "learning_rate": 0.00018861380078614726,
      "loss": 1.3885,
      "step": 277
    },
    {
      "epoch": 0.18161032173771027,
      "grad_norm": Infinity,
      "learning_rate": 0.00018851523085576096,
      "loss": 1.3993,
      "step": 278
    },
    {
      "epoch": 0.18226359627633512,
      "grad_norm": Infinity,
      "learning_rate": 0.00018841626208955292,
      "loss": 1.3527,
      "step": 279
    },
    {
      "epoch": 0.18291687081496,
      "grad_norm": Infinity,
      "learning_rate": 0.00018831689493346095,
      "loss": 1.3832,
      "step": 280
    },
    {
      "epoch": 0.18357014535358485,
      "grad_norm": Infinity,
      "learning_rate": 0.0001882171298352179,
      "loss": 1.4736,
      "step": 281
    },
    {
      "epoch": 0.1842234198922097,
      "grad_norm": Infinity,
      "learning_rate": 0.00018811696724434983,
      "loss": 1.5195,
      "step": 282
    },
    {
      "epoch": 0.18487669443083457,
      "grad_norm": Infinity,
      "learning_rate": 0.00018801640761217367,
      "loss": 1.5279,
      "step": 283
    },
    {
      "epoch": 0.18552996896945942,
      "grad_norm": Infinity,
      "learning_rate": 0.00018791545139179548,
      "loss": 1.5788,
      "step": 284
    },
    {
      "epoch": 0.18618324350808427,
      "grad_norm": Infinity,
      "learning_rate": 0.00018781409903810821,
      "loss": 1.6494,
      "step": 285
    },
    {
      "epoch": 0.18683651804670912,
      "grad_norm": Infinity,
      "learning_rate": 0.0001877123510077898,
      "loss": 1.5504,
      "step": 286
    },
    {
      "epoch": 0.187489792585334,
      "grad_norm": Infinity,
      "learning_rate": 0.00018761020775930095,
      "loss": 1.7568,
      "step": 287
    },
    {
      "epoch": 0.18814306712395884,
      "grad_norm": Infinity,
      "learning_rate": 0.0001875076697528832,
      "loss": 1.8331,
      "step": 288
    },
    {
      "epoch": 0.1887963416625837,
      "grad_norm": Infinity,
      "learning_rate": 0.0001874047374505569,
      "loss": 1.8368,
      "step": 289
    },
    {
      "epoch": 0.18944961620120856,
      "grad_norm": Infinity,
      "learning_rate": 0.00018730141131611882,
      "loss": 1.9461,
      "step": 290
    },
    {
      "epoch": 0.1901028907398334,
      "grad_norm": Infinity,
      "learning_rate": 0.0001871976918151405,
      "loss": 2.0115,
      "step": 291
    },
    {
      "epoch": 0.19075616527845826,
      "grad_norm": Infinity,
      "learning_rate": 0.0001870935794149658,
      "loss": 2.0988,
      "step": 292
    },
    {
      "epoch": 0.19140943981708314,
      "grad_norm": Infinity,
      "learning_rate": 0.00018698907458470894,
      "loss": 2.4166,
      "step": 293
    },
    {
      "epoch": 0.192062714355708,
      "grad_norm": Infinity,
      "learning_rate": 0.0001868841777952524,
      "loss": 2.4735,
      "step": 294
    },
    {
      "epoch": 0.19271598889433283,
      "grad_norm": Infinity,
      "learning_rate": 0.00018677888951924474,
      "loss": 2.6137,
      "step": 295
    },
    {
      "epoch": 0.1933692634329577,
      "grad_norm": Infinity,
      "learning_rate": 0.0001866732102310985,
      "loss": 2.9701,
      "step": 296
    },
    {
      "epoch": 0.19402253797158256,
      "grad_norm": Infinity,
      "learning_rate": 0.0001865671404069881,
      "loss": 3.195,
      "step": 297
    },
    {
      "epoch": 0.1946758125102074,
      "grad_norm": Infinity,
      "learning_rate": 0.00018646068052484755,
      "loss": 3.4701,
      "step": 298
    },
    {
      "epoch": 0.19532908704883228,
      "grad_norm": Infinity,
      "learning_rate": 0.00018635383106436855,
      "loss": 3.0494,
      "step": 299
    },
    {
      "epoch": 0.19598236158745713,
      "grad_norm": Infinity,
      "learning_rate": 0.00018624659250699805,
      "loss": 3.8624,
      "step": 300
    },
    {
      "epoch": 0.19663563612608198,
      "grad_norm": Infinity,
      "learning_rate": 0.00018613896533593632,
      "loss": 1.318,
      "step": 301
    },
    {
      "epoch": 0.19728891066470686,
      "grad_norm": Infinity,
      "learning_rate": 0.0001860309500361345,
      "loss": 1.3523,
      "step": 302
    },
    {
      "epoch": 0.1979421852033317,
      "grad_norm": Infinity,
      "learning_rate": 0.0001859225470942928,
      "loss": 1.3658,
      "step": 303
    },
    {
      "epoch": 0.19859545974195655,
      "grad_norm": Infinity,
      "learning_rate": 0.00018581375699885786,
      "loss": 1.4371,
      "step": 304
    },
    {
      "epoch": 0.1992487342805814,
      "grad_norm": Infinity,
      "learning_rate": 0.00018570458024002093,
      "loss": 1.4728,
      "step": 305
    },
    {
      "epoch": 0.19990200881920628,
      "grad_norm": Infinity,
      "learning_rate": 0.00018559501730971544,
      "loss": 1.4962,
      "step": 306
    },
    {
      "epoch": 0.20055528335783113,
      "grad_norm": Infinity,
      "learning_rate": 0.00018548506870161492,
      "loss": 1.5558,
      "step": 307
    },
    {
      "epoch": 0.20120855789645598,
      "grad_norm": Infinity,
      "learning_rate": 0.00018537473491113054,
      "loss": 1.5558,
      "step": 308
    },
    {
      "epoch": 0.20186183243508085,
      "grad_norm": Infinity,
      "learning_rate": 0.00018526401643540922,
      "loss": 1.6514,
      "step": 309
    },
    {
      "epoch": 0.2025151069737057,
      "grad_norm": Infinity,
      "learning_rate": 0.00018515291377333112,
      "loss": 1.6654,
      "step": 310
    },
    {
      "epoch": 0.20316838151233055,
      "grad_norm": Infinity,
      "learning_rate": 0.00018504142742550755,
      "loss": 1.6443,
      "step": 311
    },
    {
      "epoch": 0.20382165605095542,
      "grad_norm": Infinity,
      "learning_rate": 0.0001849295578942785,
      "loss": 1.7981,
      "step": 312
    },
    {
      "epoch": 0.20447493058958027,
      "grad_norm": Infinity,
      "learning_rate": 0.00018481730568371073,
      "loss": 2.0421,
      "step": 313
    },
    {
      "epoch": 0.20512820512820512,
      "grad_norm": Infinity,
      "learning_rate": 0.0001847046712995951,
      "loss": 1.9165,
      "step": 314
    },
    {
      "epoch": 0.20578147966683,
      "grad_norm": Infinity,
      "learning_rate": 0.0001845916552494446,
      "loss": 2.0489,
      "step": 315
    },
    {
      "epoch": 0.20643475420545485,
      "grad_norm": Infinity,
      "learning_rate": 0.00018447825804249201,
      "loss": 2.2374,
      "step": 316
    },
    {
      "epoch": 0.2070880287440797,
      "grad_norm": Infinity,
      "learning_rate": 0.00018436448018968731,
      "loss": 2.232,
      "step": 317
    },
    {
      "epoch": 0.20774130328270457,
      "grad_norm": Infinity,
      "learning_rate": 0.00018425032220369589,
      "loss": 2.3227,
      "step": 318
    },
    {
      "epoch": 0.20839457782132942,
      "grad_norm": Infinity,
      "learning_rate": 0.0001841357845988957,
      "loss": 2.6041,
      "step": 319
    },
    {
      "epoch": 0.20904785235995427,
      "grad_norm": Infinity,
      "learning_rate": 0.00018402086789137546,
      "loss": 2.6858,
      "step": 320
    },
    {
      "epoch": 0.20970112689857912,
      "grad_norm": Infinity,
      "learning_rate": 0.00018390557259893178,
      "loss": 2.9323,
      "step": 321
    },
    {
      "epoch": 0.210354401437204,
      "grad_norm": Infinity,
      "learning_rate": 0.00018378989924106736,
      "loss": 2.9611,
      "step": 322
    },
    {
      "epoch": 0.21100767597582884,
      "grad_norm": Infinity,
      "learning_rate": 0.00018367384833898828,
      "loss": 3.4458,
      "step": 323
    },
    {
      "epoch": 0.2116609505144537,
      "grad_norm": Infinity,
      "learning_rate": 0.0001835574204156018,
      "loss": 3.5276,
      "step": 324
    },
    {
      "epoch": 0.21231422505307856,
      "grad_norm": Infinity,
      "learning_rate": 0.00018344061599551398,
      "loss": 4.3704,
      "step": 325
    },
    {
      "epoch": 0.2129674995917034,
      "grad_norm": Infinity,
      "learning_rate": 0.0001833234356050273,
      "loss": 1.3251,
      "step": 326
    },
    {
      "epoch": 0.21362077413032826,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832058797721383,
      "loss": 1.3373,
      "step": 327
    },
    {
      "epoch": 0.21427404866895314,
      "grad_norm": Infinity,
      "learning_rate": 0.00018308794902653533,
      "loss": 1.3725,
      "step": 328
    },
    {
      "epoch": 0.21492732320757799,
      "grad_norm": Infinity,
      "learning_rate": 0.00018296964389959578,
      "loss": 1.4768,
      "step": 329
    },
    {
      "epoch": 0.21558059774620283,
      "grad_norm": Infinity,
      "learning_rate": 0.00018285096492438424,
      "loss": 1.4613,
      "step": 330
    },
    {
      "epoch": 0.2162338722848277,
      "grad_norm": Infinity,
      "learning_rate": 0.00018273191263564956,
      "loss": 1.4252,
      "step": 331
    },
    {
      "epoch": 0.21688714682345256,
      "grad_norm": Infinity,
      "learning_rate": 0.0001826124875698228,
      "loss": 1.5251,
      "step": 332
    },
    {
      "epoch": 0.2175404213620774,
      "grad_norm": Infinity,
      "learning_rate": 0.00018249269026501472,
      "loss": 1.6811,
      "step": 333
    },
    {
      "epoch": 0.21819369590070228,
      "grad_norm": Infinity,
      "learning_rate": 0.00018237252126101323,
      "loss": 1.6452,
      "step": 334
    },
    {
      "epoch": 0.21884697043932713,
      "grad_norm": Infinity,
      "learning_rate": 0.00018225198109928114,
      "loss": 1.6707,
      "step": 335
    },
    {
      "epoch": 0.21950024497795198,
      "grad_norm": Infinity,
      "learning_rate": 0.00018213107032295363,
      "loss": 1.6945,
      "step": 336
    },
    {
      "epoch": 0.22015351951657683,
      "grad_norm": Infinity,
      "learning_rate": 0.00018200978947683583,
      "loss": 1.8109,
      "step": 337
    },
    {
      "epoch": 0.2208067940552017,
      "grad_norm": Infinity,
      "learning_rate": 0.0001818881391074002,
      "loss": 2.0009,
      "step": 338
    },
    {
      "epoch": 0.22146006859382655,
      "grad_norm": Infinity,
      "learning_rate": 0.00018176611976278441,
      "loss": 2.0717,
      "step": 339
    },
    {
      "epoch": 0.2221133431324514,
      "grad_norm": Infinity,
      "learning_rate": 0.00018164373199278856,
      "loss": 2.1003,
      "step": 340
    },
    {
      "epoch": 0.22276661767107628,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815209763488729,
      "loss": 2.1306,
      "step": 341
    },
    {
      "epoch": 0.22341989220970113,
      "grad_norm": Infinity,
      "learning_rate": 0.00018139785338415517,
      "loss": 2.2117,
      "step": 342
    },
    {
      "epoch": 0.22407316674832597,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812743636534082,
      "loss": 2.5837,
      "step": 343
    },
    {
      "epoch": 0.22472644128695085,
      "grad_norm": Infinity,
      "learning_rate": 0.00018115050771305756,
      "loss": 2.4268,
      "step": 344
    },
    {
      "epoch": 0.2253797158255757,
      "grad_norm": Infinity,
      "learning_rate": 0.00018102628612117865,
      "loss": 2.4221,
      "step": 345
    },
    {
      "epoch": 0.22603299036420055,
      "grad_norm": Infinity,
      "learning_rate": 0.00018090169943749476,
      "loss": 2.8648,
      "step": 346
    },
    {
      "epoch": 0.22668626490282542,
      "grad_norm": Infinity,
      "learning_rate": 0.00018077674822337392,
      "loss": 2.9829,
      "step": 347
    },
    {
      "epoch": 0.22733953944145027,
      "grad_norm": Infinity,
      "learning_rate": 0.00018065143304182683,
      "loss": 3.1063,
      "step": 348
    },
    {
      "epoch": 0.22799281398007512,
      "grad_norm": Infinity,
      "learning_rate": 0.00018052575445750419,
      "loss": 3.2922,
      "step": 349
    },
    {
      "epoch": 0.2286460885187,
      "grad_norm": Infinity,
      "learning_rate": 0.00018039971303669407,
      "loss": 4.3134,
      "step": 350
    },
    {
      "epoch": 0.22929936305732485,
      "grad_norm": Infinity,
      "learning_rate": 0.00018027330934731946,
      "loss": 1.2538,
      "step": 351
    },
    {
      "epoch": 0.2299526375959497,
      "grad_norm": Infinity,
      "learning_rate": 0.00018014654395893563,
      "loss": 1.3942,
      "step": 352
    },
    {
      "epoch": 0.23060591213457454,
      "grad_norm": Infinity,
      "learning_rate": 0.00018001941744272767,
      "loss": 1.3873,
      "step": 353
    },
    {
      "epoch": 0.23125918667319942,
      "grad_norm": Infinity,
      "learning_rate": 0.00017989193037150784,
      "loss": 1.3868,
      "step": 354
    },
    {
      "epoch": 0.23191246121182427,
      "grad_norm": Infinity,
      "learning_rate": 0.00017976408331971298,
      "loss": 1.5054,
      "step": 355
    },
    {
      "epoch": 0.23256573575044912,
      "grad_norm": Infinity,
      "learning_rate": 0.00017963587686340197,
      "loss": 1.4309,
      "step": 356
    },
    {
      "epoch": 0.233219010289074,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795073115802531,
      "loss": 1.5283,
      "step": 357
    },
    {
      "epoch": 0.23387228482769884,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793783880495615,
      "loss": 1.6016,
      "step": 358
    },
    {
      "epoch": 0.2345255593663237,
      "grad_norm": Infinity,
      "learning_rate": 0.00017924910685223643,
      "loss": 1.508,
      "step": 359
    },
    {
      "epoch": 0.23517883390494856,
      "grad_norm": Infinity,
      "learning_rate": 0.00017911946857079888,
      "loss": 1.735,
      "step": 360
    },
    {
      "epoch": 0.2358321084435734,
      "grad_norm": Infinity,
      "learning_rate": 0.0001789894737893786,
      "loss": 1.6061,
      "step": 361
    },
    {
      "epoch": 0.23648538298219826,
      "grad_norm": Infinity,
      "learning_rate": 0.00017885912309371192,
      "loss": 1.6814,
      "step": 362
    },
    {
      "epoch": 0.23713865752082314,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787284170711387,
      "loss": 1.7403,
      "step": 363
    },
    {
      "epoch": 0.23779193205944799,
      "grad_norm": Infinity,
      "learning_rate": 0.00017859735631059985,
      "loss": 1.8771,
      "step": 364
    },
    {
      "epoch": 0.23844520659807283,
      "grad_norm": Infinity,
      "learning_rate": 0.00017846594140263474,
      "loss": 1.954,
      "step": 365
    },
    {
      "epoch": 0.2390984811366977,
      "grad_norm": Infinity,
      "learning_rate": 0.00017833417293937847,
      "loss": 2.1301,
      "step": 366
    },
    {
      "epoch": 0.23975175567532256,
      "grad_norm": Infinity,
      "learning_rate": 0.00017820205151455914,
      "loss": 2.0567,
      "step": 367
    },
    {
      "epoch": 0.2404050302139474,
      "grad_norm": Infinity,
      "learning_rate": 0.00017806957772349526,
      "loss": 2.4691,
      "step": 368
    },
    {
      "epoch": 0.24105830475257226,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779367521630931,
      "loss": 2.4948,
      "step": 369
    },
    {
      "epoch": 0.24171157929119713,
      "grad_norm": Infinity,
      "learning_rate": 0.00017780357543184397,
      "loss": 2.7374,
      "step": 370
    },
    {
      "epoch": 0.24236485382982198,
      "grad_norm": Infinity,
      "learning_rate": 0.00017767004812982133,
      "loss": 2.746,
      "step": 371
    },
    {
      "epoch": 0.24301812836844683,
      "grad_norm": Infinity,
      "learning_rate": 0.00017753617085867847,
      "loss": 2.6322,
      "step": 372
    },
    {
      "epoch": 0.2436714029070717,
      "grad_norm": Infinity,
      "learning_rate": 0.00017740194422164542,
      "loss": 2.9857,
      "step": 373
    },
    {
      "epoch": 0.24432467744569655,
      "grad_norm": Infinity,
      "learning_rate": 0.0001772673688235265,
      "loss": 3.2684,
      "step": 374
    },
    {
      "epoch": 0.2449779519843214,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771324452706975,
      "loss": 4.7209,
      "step": 375
    },
    {
      "epoch": 0.24563122652294628,
      "grad_norm": Infinity,
      "learning_rate": 0.00017699717417110283,
      "loss": 1.3461,
      "step": 376
    },
    {
      "epoch": 0.24628450106157113,
      "grad_norm": Infinity,
      "learning_rate": 0.00017686155613425296,
      "loss": 1.4335,
      "step": 377
    },
    {
      "epoch": 0.24693777560019597,
      "grad_norm": Infinity,
      "learning_rate": 0.00017672559177122165,
      "loss": 1.4385,
      "step": 378
    },
    {
      "epoch": 0.24759105013882085,
      "grad_norm": Infinity,
      "learning_rate": 0.00017658928169464312,
      "loss": 1.3503,
      "step": 379
    },
    {
      "epoch": 0.2482443246774457,
      "grad_norm": Infinity,
      "learning_rate": 0.00017645262651870926,
      "loss": 1.4273,
      "step": 380
    },
    {
      "epoch": 0.24889759921607055,
      "grad_norm": Infinity,
      "learning_rate": 0.00017631562685916703,
      "loss": 1.3989,
      "step": 381
    },
    {
      "epoch": 0.24955087375469542,
      "grad_norm": Infinity,
      "learning_rate": 0.00017617828333331545,
      "loss": 1.5455,
      "step": 382
    },
    {
      "epoch": 0.25020414829332027,
      "grad_norm": Infinity,
      "learning_rate": 0.0001760405965600031,
      "loss": 1.5684,
      "step": 383
    },
    {
      "epoch": 0.25020414829332027,
      "eval_loss": 2.097975254058838,
      "eval_runtime": 160.3657,
      "eval_samples_per_second": 16.076,
      "eval_steps_per_second": 4.022,
      "step": 383
    },
    {
      "epoch": 0.2508574228319451,
      "grad_norm": Infinity,
      "learning_rate": 0.00017590256715962506,
      "loss": 1.4462,
      "step": 384
    },
    {
      "epoch": 0.25151069737056997,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757641957541203,
      "loss": 1.5079,
      "step": 385
    },
    {
      "epoch": 0.2521639719091948,
      "grad_norm": Infinity,
      "learning_rate": 0.00017562548296696875,
      "loss": 1.536,
      "step": 386
    },
    {
      "epoch": 0.2528172464478197,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754864294231886,
      "loss": 1.5538,
      "step": 387
    },
    {
      "epoch": 0.25347052098644457,
      "grad_norm": Infinity,
      "learning_rate": 0.00017534703574933347,
      "loss": 1.8808,
      "step": 388
    },
    {
      "epoch": 0.2541237955250694,
      "grad_norm": Infinity,
      "learning_rate": 0.00017520730257348946,
      "loss": 1.9979,
      "step": 389
    },
    {
      "epoch": 0.25477707006369427,
      "grad_norm": Infinity,
      "learning_rate": 0.00017506723052527242,
      "loss": 1.9746,
      "step": 390
    },
    {
      "epoch": 0.2554303446023191,
      "grad_norm": Infinity,
      "learning_rate": 0.00017492682023582528,
      "loss": 2.23,
      "step": 391
    },
    {
      "epoch": 0.25608361914094396,
      "grad_norm": Infinity,
      "learning_rate": 0.0001747860723378148,
      "loss": 2.2459,
      "step": 392
    },
    {
      "epoch": 0.2567368936795688,
      "grad_norm": Infinity,
      "learning_rate": 0.0001746449874654291,
      "loss": 2.1848,
      "step": 393
    },
    {
      "epoch": 0.2573901682181937,
      "grad_norm": Infinity,
      "learning_rate": 0.0001745035662543745,
      "loss": 2.5955,
      "step": 394
    },
    {
      "epoch": 0.25804344275681856,
      "grad_norm": Infinity,
      "learning_rate": 0.00017436180934187308,
      "loss": 2.6265,
      "step": 395
    },
    {
      "epoch": 0.2586967172954434,
      "grad_norm": Infinity,
      "learning_rate": 0.00017421971736665925,
      "loss": 2.9216,
      "step": 396
    },
    {
      "epoch": 0.25934999183406826,
      "grad_norm": Infinity,
      "learning_rate": 0.00017407729096897737,
      "loss": 2.6603,
      "step": 397
    },
    {
      "epoch": 0.2600032663726931,
      "grad_norm": Infinity,
      "learning_rate": 0.00017393453079057847,
      "loss": 3.1459,
      "step": 398
    },
    {
      "epoch": 0.26065654091131796,
      "grad_norm": Infinity,
      "learning_rate": 0.00017379143747471768,
      "loss": 3.2171,
      "step": 399
    },
    {
      "epoch": 0.26130981544994286,
      "grad_norm": Infinity,
      "learning_rate": 0.00017364801166615124,
      "loss": 4.4529,
      "step": 400
    },
    {
      "epoch": 0.2619630899885677,
      "grad_norm": Infinity,
      "learning_rate": 0.00017350425401113336,
      "loss": 1.3081,
      "step": 401
    },
    {
      "epoch": 0.26261636452719256,
      "grad_norm": Infinity,
      "learning_rate": 0.00017336016515741366,
      "loss": 1.4575,
      "step": 402
    },
    {
      "epoch": 0.2632696390658174,
      "grad_norm": Infinity,
      "learning_rate": 0.00017321574575423406,
      "loss": 1.4,
      "step": 403
    },
    {
      "epoch": 0.26392291360444226,
      "grad_norm": Infinity,
      "learning_rate": 0.00017307099645232578,
      "loss": 1.4742,
      "step": 404
    },
    {
      "epoch": 0.2645761881430671,
      "grad_norm": Infinity,
      "learning_rate": 0.00017292591790390665,
      "loss": 1.4765,
      "step": 405
    },
    {
      "epoch": 0.265229462681692,
      "grad_norm": Infinity,
      "learning_rate": 0.00017278051076267796,
      "loss": 1.5007,
      "step": 406
    },
    {
      "epoch": 0.26588273722031686,
      "grad_norm": Infinity,
      "learning_rate": 0.00017263477568382166,
      "loss": 1.5344,
      "step": 407
    },
    {
      "epoch": 0.2665360117589417,
      "grad_norm": Infinity,
      "learning_rate": 0.0001724887133239972,
      "loss": 1.5091,
      "step": 408
    },
    {
      "epoch": 0.26718928629756655,
      "grad_norm": Infinity,
      "learning_rate": 0.00017234232434133883,
      "loss": 1.6508,
      "step": 409
    },
    {
      "epoch": 0.2678425608361914,
      "grad_norm": Infinity,
      "learning_rate": 0.00017219560939545246,
      "loss": 1.7905,
      "step": 410
    },
    {
      "epoch": 0.26849583537481625,
      "grad_norm": Infinity,
      "learning_rate": 0.00017204856914741274,
      "loss": 1.686,
      "step": 411
    },
    {
      "epoch": 0.2691491099134411,
      "grad_norm": Infinity,
      "learning_rate": 0.0001719012042597601,
      "loss": 1.9091,
      "step": 412
    },
    {
      "epoch": 0.269802384452066,
      "grad_norm": Infinity,
      "learning_rate": 0.00017175351539649774,
      "loss": 1.9509,
      "step": 413
    },
    {
      "epoch": 0.27045565899069085,
      "grad_norm": Infinity,
      "learning_rate": 0.00017160550322308863,
      "loss": 1.9816,
      "step": 414
    },
    {
      "epoch": 0.2711089335293157,
      "grad_norm": Infinity,
      "learning_rate": 0.00017145716840645254,
      "loss": 2.3551,
      "step": 415
    },
    {
      "epoch": 0.27176220806794055,
      "grad_norm": Infinity,
      "learning_rate": 0.000171308511614963,
      "loss": 2.2993,
      "step": 416
    },
    {
      "epoch": 0.2724154826065654,
      "grad_norm": Infinity,
      "learning_rate": 0.00017115953351844434,
      "loss": 2.4097,
      "step": 417
    },
    {
      "epoch": 0.27306875714519024,
      "grad_norm": Infinity,
      "learning_rate": 0.00017101023478816857,
      "loss": 2.6394,
      "step": 418
    },
    {
      "epoch": 0.27372203168381515,
      "grad_norm": Infinity,
      "learning_rate": 0.00017086061609685257,
      "loss": 2.7541,
      "step": 419
    },
    {
      "epoch": 0.27437530622244,
      "grad_norm": Infinity,
      "learning_rate": 0.00017071067811865476,
      "loss": 2.6986,
      "step": 420
    },
    {
      "epoch": 0.27502858076106484,
      "grad_norm": Infinity,
      "learning_rate": 0.0001705604215291723,
      "loss": 2.8947,
      "step": 421
    },
    {
      "epoch": 0.2756818552996897,
      "grad_norm": Infinity,
      "learning_rate": 0.00017040984700543793,
      "loss": 2.5362,
      "step": 422
    },
    {
      "epoch": 0.27633512983831454,
      "grad_norm": Infinity,
      "learning_rate": 0.00017025895522591693,
      "loss": 2.8326,
      "step": 423
    },
    {
      "epoch": 0.2769884043769394,
      "grad_norm": Infinity,
      "learning_rate": 0.00017010774687050418,
      "loss": 3.813,
      "step": 424
    },
    {
      "epoch": 0.27764167891556424,
      "grad_norm": Infinity,
      "learning_rate": 0.00016995622262052092,
      "loss": 3.9088,
      "step": 425
    },
    {
      "epoch": 0.27829495345418914,
      "grad_norm": Infinity,
      "learning_rate": 0.00016980438315871178,
      "loss": 1.4111,
      "step": 426
    },
    {
      "epoch": 0.278948227992814,
      "grad_norm": Infinity,
      "learning_rate": 0.00016965222916924167,
      "loss": 1.3758,
      "step": 427
    },
    {
      "epoch": 0.27960150253143884,
      "grad_norm": Infinity,
      "learning_rate": 0.0001694997613376928,
      "loss": 1.4541,
      "step": 428
    },
    {
      "epoch": 0.2802547770700637,
      "grad_norm": Infinity,
      "learning_rate": 0.00016934698035106133,
      "loss": 1.3758,
      "step": 429
    },
    {
      "epoch": 0.28090805160868854,
      "grad_norm": Infinity,
      "learning_rate": 0.00016919388689775464,
      "loss": 1.491,
      "step": 430
    },
    {
      "epoch": 0.2815613261473134,
      "grad_norm": Infinity,
      "learning_rate": 0.0001690404816675879,
      "loss": 1.5145,
      "step": 431
    },
    {
      "epoch": 0.2822146006859383,
      "grad_norm": Infinity,
      "learning_rate": 0.00016888676535178115,
      "loss": 1.5918,
      "step": 432
    },
    {
      "epoch": 0.28286787522456314,
      "grad_norm": Infinity,
      "learning_rate": 0.00016873273864295612,
      "loss": 1.5333,
      "step": 433
    },
    {
      "epoch": 0.283521149763188,
      "grad_norm": Infinity,
      "learning_rate": 0.00016857840223513315,
      "loss": 1.7168,
      "step": 434
    },
    {
      "epoch": 0.28417442430181283,
      "grad_norm": Infinity,
      "learning_rate": 0.00016842375682372805,
      "loss": 1.5902,
      "step": 435
    },
    {
      "epoch": 0.2848276988404377,
      "grad_norm": Infinity,
      "learning_rate": 0.00016826880310554887,
      "loss": 1.6932,
      "step": 436
    },
    {
      "epoch": 0.28548097337906253,
      "grad_norm": Infinity,
      "learning_rate": 0.00016811354177879287,
      "loss": 1.8251,
      "step": 437
    },
    {
      "epoch": 0.28613424791768743,
      "grad_norm": Infinity,
      "learning_rate": 0.00016795797354304345,
      "loss": 1.7183,
      "step": 438
    },
    {
      "epoch": 0.2867875224563123,
      "grad_norm": Infinity,
      "learning_rate": 0.00016780209909926676,
      "loss": 1.836,
      "step": 439
    },
    {
      "epoch": 0.28744079699493713,
      "grad_norm": Infinity,
      "learning_rate": 0.0001676459191498087,
      "loss": 2.0649,
      "step": 440
    },
    {
      "epoch": 0.288094071533562,
      "grad_norm": Infinity,
      "learning_rate": 0.0001674894343983918,
      "loss": 2.058,
      "step": 441
    },
    {
      "epoch": 0.28874734607218683,
      "grad_norm": Infinity,
      "learning_rate": 0.00016733264555011195,
      "loss": 2.1594,
      "step": 442
    },
    {
      "epoch": 0.2894006206108117,
      "grad_norm": Infinity,
      "learning_rate": 0.0001671755533114352,
      "loss": 2.3795,
      "step": 443
    },
    {
      "epoch": 0.2900538951494365,
      "grad_norm": Infinity,
      "learning_rate": 0.00016701815839019468,
      "loss": 2.6133,
      "step": 444
    },
    {
      "epoch": 0.29070716968806143,
      "grad_norm": Infinity,
      "learning_rate": 0.00016686046149558736,
      "loss": 2.5835,
      "step": 445
    },
    {
      "epoch": 0.2913604442266863,
      "grad_norm": Infinity,
      "learning_rate": 0.00016670246333817088,
      "loss": 2.9254,
      "step": 446
    },
    {
      "epoch": 0.2920137187653111,
      "grad_norm": Infinity,
      "learning_rate": 0.00016654416462986022,
      "loss": 2.9367,
      "step": 447
    },
    {
      "epoch": 0.292666993303936,
      "grad_norm": Infinity,
      "learning_rate": 0.00016638556608392473,
      "loss": 3.2409,
      "step": 448
    },
    {
      "epoch": 0.2933202678425608,
      "grad_norm": Infinity,
      "learning_rate": 0.00016622666841498463,
      "loss": 3.7771,
      "step": 449
    },
    {
      "epoch": 0.29397354238118567,
      "grad_norm": Infinity,
      "learning_rate": 0.00016606747233900815,
      "loss": 4.364,
      "step": 450
    },
    {
      "epoch": 0.2946268169198106,
      "grad_norm": Infinity,
      "learning_rate": 0.00016590797857330789,
      "loss": 1.2995,
      "step": 451
    },
    {
      "epoch": 0.2952800914584354,
      "grad_norm": Infinity,
      "learning_rate": 0.0001657481878365379,
      "loss": 1.4345,
      "step": 452
    },
    {
      "epoch": 0.29593336599706027,
      "grad_norm": Infinity,
      "learning_rate": 0.0001655881008486903,
      "loss": 1.421,
      "step": 453
    },
    {
      "epoch": 0.2965866405356851,
      "grad_norm": Infinity,
      "learning_rate": 0.0001654277183310921,
      "loss": 1.4081,
      "step": 454
    },
    {
      "epoch": 0.29723991507430997,
      "grad_norm": Infinity,
      "learning_rate": 0.0001652670410064019,
      "loss": 1.4488,
      "step": 455
    },
    {
      "epoch": 0.2978931896129348,
      "grad_norm": Infinity,
      "learning_rate": 0.00016510606959860665,
      "loss": 1.4637,
      "step": 456
    },
    {
      "epoch": 0.2985464641515597,
      "grad_norm": Infinity,
      "learning_rate": 0.00016494480483301836,
      "loss": 1.5709,
      "step": 457
    },
    {
      "epoch": 0.29919973869018457,
      "grad_norm": Infinity,
      "learning_rate": 0.00016478324743627101,
      "loss": 1.4784,
      "step": 458
    },
    {
      "epoch": 0.2998530132288094,
      "grad_norm": Infinity,
      "learning_rate": 0.00016462139813631693,
      "loss": 1.5429,
      "step": 459
    },
    {
      "epoch": 0.30050628776743427,
      "grad_norm": Infinity,
      "learning_rate": 0.00016445925766242391,
      "loss": 1.6066,
      "step": 460
    },
    {
      "epoch": 0.3011595623060591,
      "grad_norm": Infinity,
      "learning_rate": 0.00016429682674517156,
      "loss": 1.8577,
      "step": 461
    },
    {
      "epoch": 0.30181283684468396,
      "grad_norm": Infinity,
      "learning_rate": 0.00016413410611644825,
      "loss": 1.7408,
      "step": 462
    },
    {
      "epoch": 0.3024661113833088,
      "grad_norm": Infinity,
      "learning_rate": 0.0001639710965094478,
      "loss": 1.8884,
      "step": 463
    },
    {
      "epoch": 0.3031193859219337,
      "grad_norm": Infinity,
      "learning_rate": 0.00016380779865866603,
      "loss": 2.052,
      "step": 464
    },
    {
      "epoch": 0.30377266046055856,
      "grad_norm": Infinity,
      "learning_rate": 0.00016364421329989755,
      "loss": 1.9189,
      "step": 465
    },
    {
      "epoch": 0.3044259349991834,
      "grad_norm": Infinity,
      "learning_rate": 0.00016348034117023258,
      "loss": 2.2974,
      "step": 466
    },
    {
      "epoch": 0.30507920953780826,
      "grad_norm": Infinity,
      "learning_rate": 0.00016331618300805326,
      "loss": 2.187,
      "step": 467
    },
    {
      "epoch": 0.3057324840764331,
      "grad_norm": Infinity,
      "learning_rate": 0.00016315173955303068,
      "loss": 2.4096,
      "step": 468
    },
    {
      "epoch": 0.30638575861505796,
      "grad_norm": Infinity,
      "learning_rate": 0.00016298701154612147,
      "loss": 2.3334,
      "step": 469
    },
    {
      "epoch": 0.30703903315368286,
      "grad_norm": Infinity,
      "learning_rate": 0.00016282199972956425,
      "loss": 2.7411,
      "step": 470
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": Infinity,
      "learning_rate": 0.00016265670484687654,
      "loss": 2.7242,
      "step": 471
    },
    {
      "epoch": 0.30834558223093256,
      "grad_norm": Infinity,
      "learning_rate": 0.0001624911276428513,
      "loss": 3.1537,
      "step": 472
    },
    {
      "epoch": 0.3089988567695574,
      "grad_norm": Infinity,
      "learning_rate": 0.0001623252688635536,
      "loss": 3.5625,
      "step": 473
    },
    {
      "epoch": 0.30965213130818225,
      "grad_norm": Infinity,
      "learning_rate": 0.00016215912925631723,
      "loss": 3.6483,
      "step": 474
    },
    {
      "epoch": 0.3103054058468071,
      "grad_norm": Infinity,
      "learning_rate": 0.00016199270956974128,
      "loss": 4.2782,
      "step": 475
    },
    {
      "epoch": 0.31095868038543195,
      "grad_norm": Infinity,
      "learning_rate": 0.00016182601055368697,
      "loss": 1.3157,
      "step": 476
    },
    {
      "epoch": 0.31161195492405686,
      "grad_norm": Infinity,
      "learning_rate": 0.00016165903295927401,
      "loss": 1.4153,
      "step": 477
    },
    {
      "epoch": 0.3122652294626817,
      "grad_norm": Infinity,
      "learning_rate": 0.00016149177753887746,
      "loss": 1.4591,
      "step": 478
    },
    {
      "epoch": 0.31291850400130655,
      "grad_norm": Infinity,
      "learning_rate": 0.00016132424504612406,
      "loss": 1.4796,
      "step": 479
    },
    {
      "epoch": 0.3135717785399314,
      "grad_norm": Infinity,
      "learning_rate": 0.00016115643623588915,
      "loss": 1.5419,
      "step": 480
    },
    {
      "epoch": 0.31422505307855625,
      "grad_norm": Infinity,
      "learning_rate": 0.00016098835186429303,
      "loss": 1.4612,
      "step": 481
    },
    {
      "epoch": 0.3148783276171811,
      "grad_norm": Infinity,
      "learning_rate": 0.00016081999268869766,
      "loss": 1.6256,
      "step": 482
    },
    {
      "epoch": 0.315531602155806,
      "grad_norm": Infinity,
      "learning_rate": 0.00016065135946770323,
      "loss": 1.6535,
      "step": 483
    },
    {
      "epoch": 0.31618487669443085,
      "grad_norm": Infinity,
      "learning_rate": 0.00016048245296114468,
      "loss": 1.6035,
      "step": 484
    },
    {
      "epoch": 0.3168381512330557,
      "grad_norm": Infinity,
      "learning_rate": 0.00016031327393008845,
      "loss": 1.575,
      "step": 485
    },
    {
      "epoch": 0.31749142577168055,
      "grad_norm": Infinity,
      "learning_rate": 0.00016014382313682881,
      "loss": 1.792,
      "step": 486
    },
    {
      "epoch": 0.3181447003103054,
      "grad_norm": Infinity,
      "learning_rate": 0.00015997410134488464,
      "loss": 1.7314,
      "step": 487
    },
    {
      "epoch": 0.31879797484893024,
      "grad_norm": Infinity,
      "learning_rate": 0.00015980410931899582,
      "loss": 1.8793,
      "step": 488
    },
    {
      "epoch": 0.31945124938755515,
      "grad_norm": Infinity,
      "learning_rate": 0.00015963384782511993,
      "loss": 1.9586,
      "step": 489
    },
    {
      "epoch": 0.32010452392618,
      "grad_norm": Infinity,
      "learning_rate": 0.00015946331763042867,
      "loss": 1.9951,
      "step": 490
    },
    {
      "epoch": 0.32075779846480484,
      "grad_norm": Infinity,
      "learning_rate": 0.0001592925195033045,
      "loss": 2.1583,
      "step": 491
    },
    {
      "epoch": 0.3214110730034297,
      "grad_norm": Infinity,
      "learning_rate": 0.00015912145421333719,
      "loss": 2.3153,
      "step": 492
    },
    {
      "epoch": 0.32206434754205454,
      "grad_norm": Infinity,
      "learning_rate": 0.00015895012253132017,
      "loss": 2.3933,
      "step": 493
    },
    {
      "epoch": 0.3227176220806794,
      "grad_norm": Infinity,
      "learning_rate": 0.00015877852522924732,
      "loss": 2.3513,
      "step": 494
    },
    {
      "epoch": 0.32337089661930424,
      "grad_norm": Infinity,
      "learning_rate": 0.00015860666308030932,
      "loss": 2.6687,
      "step": 495
    },
    {
      "epoch": 0.32402417115792914,
      "grad_norm": Infinity,
      "learning_rate": 0.00015843453685889016,
      "loss": 2.8344,
      "step": 496
    },
    {
      "epoch": 0.324677445696554,
      "grad_norm": Infinity,
      "learning_rate": 0.00015826214734056374,
      "loss": 2.883,
      "step": 497
    },
    {
      "epoch": 0.32533072023517884,
      "grad_norm": Infinity,
      "learning_rate": 0.0001580894953020904,
      "loss": 3.1467,
      "step": 498
    },
    {
      "epoch": 0.3259839947738037,
      "grad_norm": Infinity,
      "learning_rate": 0.00015791658152141327,
      "loss": 3.3732,
      "step": 499
    },
    {
      "epoch": 0.32663726931242854,
      "grad_norm": Infinity,
      "learning_rate": 0.0001577434067776548,
      "loss": 4.2269,
      "step": 500
    },
    {
      "epoch": 0.3272905438510534,
      "grad_norm": Infinity,
      "learning_rate": 0.00015756997185111348,
      "loss": 1.2486,
      "step": 501
    },
    {
      "epoch": 0.3279438183896783,
      "grad_norm": Infinity,
      "learning_rate": 0.00015739627752325996,
      "loss": 1.3895,
      "step": 502
    },
    {
      "epoch": 0.32859709292830314,
      "grad_norm": Infinity,
      "learning_rate": 0.0001572223245767338,
      "loss": 1.422,
      "step": 503
    },
    {
      "epoch": 0.329250367466928,
      "grad_norm": Infinity,
      "learning_rate": 0.00015704811379533987,
      "loss": 1.4185,
      "step": 504
    },
    {
      "epoch": 0.32990364200555283,
      "grad_norm": Infinity,
      "learning_rate": 0.0001568736459640447,
      "loss": 1.4656,
      "step": 505
    },
    {
      "epoch": 0.3305569165441777,
      "grad_norm": Infinity,
      "learning_rate": 0.00015669892186897318,
      "loss": 1.4505,
      "step": 506
    },
    {
      "epoch": 0.33121019108280253,
      "grad_norm": Infinity,
      "learning_rate": 0.00015652394229740484,
      "loss": 1.6007,
      "step": 507
    },
    {
      "epoch": 0.3318634656214274,
      "grad_norm": Infinity,
      "learning_rate": 0.00015634870803777026,
      "loss": 1.6112,
      "step": 508
    },
    {
      "epoch": 0.3325167401600523,
      "grad_norm": Infinity,
      "learning_rate": 0.00015617321987964776,
      "loss": 1.7202,
      "step": 509
    },
    {
      "epoch": 0.33317001469867713,
      "grad_norm": Infinity,
      "learning_rate": 0.00015599747861375955,
      "loss": 1.6825,
      "step": 510
    },
    {
      "epoch": 0.333823289237302,
      "grad_norm": Infinity,
      "learning_rate": 0.0001558214850319684,
      "loss": 1.8399,
      "step": 511
    },
    {
      "epoch": 0.3344765637759268,
      "grad_norm": Infinity,
      "learning_rate": 0.00015564523992727387,
      "loss": 1.7425,
      "step": 512
    },
    {
      "epoch": 0.3351298383145517,
      "grad_norm": Infinity,
      "learning_rate": 0.00015546874409380897,
      "loss": 1.7817,
      "step": 513
    },
    {
      "epoch": 0.3357831128531765,
      "grad_norm": Infinity,
      "learning_rate": 0.00015529199832683635,
      "loss": 2.0647,
      "step": 514
    },
    {
      "epoch": 0.33643638739180143,
      "grad_norm": Infinity,
      "learning_rate": 0.0001551150034227449,
      "loss": 1.9344,
      "step": 515
    },
    {
      "epoch": 0.3370896619304263,
      "grad_norm": Infinity,
      "learning_rate": 0.00015493776017904602,
      "loss": 2.2739,
      "step": 516
    },
    {
      "epoch": 0.3377429364690511,
      "grad_norm": Infinity,
      "learning_rate": 0.00015476026939437014,
      "loss": 2.4646,
      "step": 517
    },
    {
      "epoch": 0.338396211007676,
      "grad_norm": Infinity,
      "learning_rate": 0.00015458253186846301,
      "loss": 2.5544,
      "step": 518
    },
    {
      "epoch": 0.3390494855463008,
      "grad_norm": Infinity,
      "learning_rate": 0.00015440454840218225,
      "loss": 2.4784,
      "step": 519
    },
    {
      "epoch": 0.33970276008492567,
      "grad_norm": Infinity,
      "learning_rate": 0.00015422631979749354,
      "loss": 2.8688,
      "step": 520
    },
    {
      "epoch": 0.3403560346235506,
      "grad_norm": Infinity,
      "learning_rate": 0.00015404784685746716,
      "loss": 2.8859,
      "step": 521
    },
    {
      "epoch": 0.3410093091621754,
      "grad_norm": Infinity,
      "learning_rate": 0.0001538691303862744,
      "loss": 3.362,
      "step": 522
    },
    {
      "epoch": 0.34166258370080027,
      "grad_norm": Infinity,
      "learning_rate": 0.00015369017118918373,
      "loss": 3.297,
      "step": 523
    },
    {
      "epoch": 0.3423158582394251,
      "grad_norm": Infinity,
      "learning_rate": 0.00015351097007255742,
      "loss": 3.6869,
      "step": 524
    },
    {
      "epoch": 0.34296913277804997,
      "grad_norm": Infinity,
      "learning_rate": 0.00015333152784384777,
      "loss": 4.2406,
      "step": 525
    },
    {
      "epoch": 0.3436224073166748,
      "grad_norm": Infinity,
      "learning_rate": 0.0001531518453115934,
      "loss": 1.3543,
      "step": 526
    },
    {
      "epoch": 0.34427568185529966,
      "grad_norm": Infinity,
      "learning_rate": 0.00015297192328541582,
      "loss": 1.4295,
      "step": 527
    },
    {
      "epoch": 0.34492895639392457,
      "grad_norm": Infinity,
      "learning_rate": 0.00015279176257601557,
      "loss": 1.4386,
      "step": 528
    },
    {
      "epoch": 0.3455822309325494,
      "grad_norm": Infinity,
      "learning_rate": 0.00015261136399516873,
      "loss": 1.4315,
      "step": 529
    },
    {
      "epoch": 0.34623550547117427,
      "grad_norm": Infinity,
      "learning_rate": 0.00015243072835572318,
      "loss": 1.4927,
      "step": 530
    },
    {
      "epoch": 0.3468887800097991,
      "grad_norm": Infinity,
      "learning_rate": 0.0001522498564715949,
      "loss": 1.4866,
      "step": 531
    },
    {
      "epoch": 0.34754205454842396,
      "grad_norm": Infinity,
      "learning_rate": 0.0001520687491577644,
      "loss": 1.5698,
      "step": 532
    },
    {
      "epoch": 0.3481953290870488,
      "grad_norm": Infinity,
      "learning_rate": 0.00015188740723027296,
      "loss": 1.5634,
      "step": 533
    },
    {
      "epoch": 0.3488486036256737,
      "grad_norm": Infinity,
      "learning_rate": 0.00015170583150621905,
      "loss": 1.6559,
      "step": 534
    },
    {
      "epoch": 0.34950187816429856,
      "grad_norm": Infinity,
      "learning_rate": 0.00015152402280375454,
      "loss": 1.726,
      "step": 535
    },
    {
      "epoch": 0.3501551527029234,
      "grad_norm": Infinity,
      "learning_rate": 0.00015134198194208107,
      "loss": 1.752,
      "step": 536
    },
    {
      "epoch": 0.35080842724154826,
      "grad_norm": Infinity,
      "learning_rate": 0.00015115970974144635,
      "loss": 1.7894,
      "step": 537
    },
    {
      "epoch": 0.3514617017801731,
      "grad_norm": Infinity,
      "learning_rate": 0.00015097720702314055,
      "loss": 1.8187,
      "step": 538
    },
    {
      "epoch": 0.35211497631879796,
      "grad_norm": Infinity,
      "learning_rate": 0.00015079447460949238,
      "loss": 1.8663,
      "step": 539
    },
    {
      "epoch": 0.3527682508574228,
      "grad_norm": Infinity,
      "learning_rate": 0.00015061151332386566,
      "loss": 1.8478,
      "step": 540
    },
    {
      "epoch": 0.3534215253960477,
      "grad_norm": Infinity,
      "learning_rate": 0.0001504283239906553,
      "loss": 1.8881,
      "step": 541
    },
    {
      "epoch": 0.35407479993467256,
      "grad_norm": Infinity,
      "learning_rate": 0.00015024490743528393,
      "loss": 2.1571,
      "step": 542
    },
    {
      "epoch": 0.3547280744732974,
      "grad_norm": Infinity,
      "learning_rate": 0.00015006126448419791,
      "loss": 2.4004,
      "step": 543
    },
    {
      "epoch": 0.35538134901192225,
      "grad_norm": Infinity,
      "learning_rate": 0.00014987739596486374,
      "loss": 2.4008,
      "step": 544
    },
    {
      "epoch": 0.3560346235505471,
      "grad_norm": Infinity,
      "learning_rate": 0.00014969330270576427,
      "loss": 2.7406,
      "step": 545
    },
    {
      "epoch": 0.35668789808917195,
      "grad_norm": Infinity,
      "learning_rate": 0.00014950898553639505,
      "loss": 2.6133,
      "step": 546
    },
    {
      "epoch": 0.35734117262779685,
      "grad_norm": Infinity,
      "learning_rate": 0.00014932444528726043,
      "loss": 2.9329,
      "step": 547
    },
    {
      "epoch": 0.3579944471664217,
      "grad_norm": Infinity,
      "learning_rate": 0.00014913968278987004,
      "loss": 3.1703,
      "step": 548
    },
    {
      "epoch": 0.35864772170504655,
      "grad_norm": Infinity,
      "learning_rate": 0.00014895469887673483,
      "loss": 2.9231,
      "step": 549
    },
    {
      "epoch": 0.3593009962436714,
      "grad_norm": Infinity,
      "learning_rate": 0.00014876949438136347,
      "loss": 4.0923,
      "step": 550
    },
    {
      "epoch": 0.35995427078229625,
      "grad_norm": Infinity,
      "learning_rate": 0.00014858407013825854,
      "loss": 1.3462,
      "step": 551
    },
    {
      "epoch": 0.3606075453209211,
      "grad_norm": Infinity,
      "learning_rate": 0.00014839842698291267,
      "loss": 1.4769,
      "step": 552
    },
    {
      "epoch": 0.361260819859546,
      "grad_norm": Infinity,
      "learning_rate": 0.00014821256575180507,
      "loss": 1.3811,
      "step": 553
    },
    {
      "epoch": 0.36191409439817085,
      "grad_norm": Infinity,
      "learning_rate": 0.00014802648728239742,
      "loss": 1.3628,
      "step": 554
    },
    {
      "epoch": 0.3625673689367957,
      "grad_norm": Infinity,
      "learning_rate": 0.00014784019241313026,
      "loss": 1.4687,
      "step": 555
    },
    {
      "epoch": 0.36322064347542055,
      "grad_norm": Infinity,
      "learning_rate": 0.00014765368198341918,
      "loss": 1.4745,
      "step": 556
    },
    {
      "epoch": 0.3638739180140454,
      "grad_norm": Infinity,
      "learning_rate": 0.00014746695683365112,
      "loss": 1.4466,
      "step": 557
    },
    {
      "epoch": 0.36452719255267024,
      "grad_norm": Infinity,
      "learning_rate": 0.0001472800178051805,
      "loss": 1.6015,
      "step": 558
    },
    {
      "epoch": 0.3651804670912951,
      "grad_norm": Infinity,
      "learning_rate": 0.00014709286574032536,
      "loss": 1.5199,
      "step": 559
    },
    {
      "epoch": 0.36583374162992,
      "grad_norm": Infinity,
      "learning_rate": 0.0001469055014823637,
      "loss": 1.5306,
      "step": 560
    },
    {
      "epoch": 0.36648701616854484,
      "grad_norm": Infinity,
      "learning_rate": 0.0001467179258755297,
      "loss": 1.745,
      "step": 561
    },
    {
      "epoch": 0.3671402907071697,
      "grad_norm": Infinity,
      "learning_rate": 0.00014653013976500975,
      "loss": 1.8516,
      "step": 562
    },
    {
      "epoch": 0.36779356524579454,
      "grad_norm": Infinity,
      "learning_rate": 0.0001463421439969388,
      "loss": 1.8714,
      "step": 563
    },
    {
      "epoch": 0.3684468397844194,
      "grad_norm": Infinity,
      "learning_rate": 0.00014615393941839637,
      "loss": 1.8113,
      "step": 564
    },
    {
      "epoch": 0.36910011432304424,
      "grad_norm": Infinity,
      "learning_rate": 0.00014596552687740302,
      "loss": 2.0245,
      "step": 565
    },
    {
      "epoch": 0.36975338886166914,
      "grad_norm": Infinity,
      "learning_rate": 0.00014577690722291622,
      "loss": 1.963,
      "step": 566
    },
    {
      "epoch": 0.370406663400294,
      "grad_norm": Infinity,
      "learning_rate": 0.00014558808130482674,
      "loss": 2.3101,
      "step": 567
    },
    {
      "epoch": 0.37105993793891884,
      "grad_norm": Infinity,
      "learning_rate": 0.00014539904997395468,
      "loss": 2.2081,
      "step": 568
    },
    {
      "epoch": 0.3717132124775437,
      "grad_norm": Infinity,
      "learning_rate": 0.00014520981408204574,
      "loss": 2.5591,
      "step": 569
    },
    {
      "epoch": 0.37236648701616853,
      "grad_norm": Infinity,
      "learning_rate": 0.00014502037448176734,
      "loss": 2.301,
      "step": 570
    },
    {
      "epoch": 0.3730197615547934,
      "grad_norm": Infinity,
      "learning_rate": 0.00014483073202670475,
      "loss": 2.8042,
      "step": 571
    },
    {
      "epoch": 0.37367303609341823,
      "grad_norm": Infinity,
      "learning_rate": 0.00014464088757135728,
      "loss": 2.9381,
      "step": 572
    },
    {
      "epoch": 0.37432631063204314,
      "grad_norm": Infinity,
      "learning_rate": 0.00014445084197113443,
      "loss": 2.9872,
      "step": 573
    },
    {
      "epoch": 0.374979585170668,
      "grad_norm": Infinity,
      "learning_rate": 0.00014426059608235208,
      "loss": 3.4324,
      "step": 574
    },
    {
      "epoch": 0.37563285970929283,
      "grad_norm": Infinity,
      "learning_rate": 0.00014407015076222846,
      "loss": 4.1841,
      "step": 575
    },
    {
      "epoch": 0.3762861342479177,
      "grad_norm": Infinity,
      "learning_rate": 0.00014387950686888047,
      "loss": 1.1931,
      "step": 576
    },
    {
      "epoch": 0.37693940878654253,
      "grad_norm": Infinity,
      "learning_rate": 0.0001436886652613198,
      "loss": 1.3665,
      "step": 577
    },
    {
      "epoch": 0.3775926833251674,
      "grad_norm": Infinity,
      "learning_rate": 0.00014349762679944896,
      "loss": 1.402,
      "step": 578
    },
    {
      "epoch": 0.3782459578637923,
      "grad_norm": Infinity,
      "learning_rate": 0.00014330639234405742,
      "loss": 1.4069,
      "step": 579
    },
    {
      "epoch": 0.37889923240241713,
      "grad_norm": Infinity,
      "learning_rate": 0.00014311496275681783,
      "loss": 1.4134,
      "step": 580
    },
    {
      "epoch": 0.379552506941042,
      "grad_norm": Infinity,
      "learning_rate": 0.00014292333890028204,
      "loss": 1.5248,
      "step": 581
    },
    {
      "epoch": 0.3802057814796668,
      "grad_norm": Infinity,
      "learning_rate": 0.00014273152163787726,
      "loss": 1.5282,
      "step": 582
    },
    {
      "epoch": 0.3808590560182917,
      "grad_norm": Infinity,
      "learning_rate": 0.00014253951183390215,
      "loss": 1.6725,
      "step": 583
    },
    {
      "epoch": 0.3815123305569165,
      "grad_norm": Infinity,
      "learning_rate": 0.0001423473103535229,
      "loss": 1.5775,
      "step": 584
    },
    {
      "epoch": 0.3821656050955414,
      "grad_norm": Infinity,
      "learning_rate": 0.00014215491806276944,
      "loss": 1.7086,
      "step": 585
    },
    {
      "epoch": 0.3828188796341663,
      "grad_norm": Infinity,
      "learning_rate": 0.0001419623358285314,
      "loss": 1.7671,
      "step": 586
    },
    {
      "epoch": 0.3834721541727911,
      "grad_norm": Infinity,
      "learning_rate": 0.0001417695645185543,
      "loss": 1.7763,
      "step": 587
    },
    {
      "epoch": 0.384125428711416,
      "grad_norm": Infinity,
      "learning_rate": 0.00014157660500143553,
      "loss": 1.9369,
      "step": 588
    },
    {
      "epoch": 0.3847787032500408,
      "grad_norm": Infinity,
      "learning_rate": 0.00014138345814662068,
      "loss": 1.8933,
      "step": 589
    },
    {
      "epoch": 0.38543197778866567,
      "grad_norm": Infinity,
      "learning_rate": 0.0001411901248243993,
      "loss": 1.9815,
      "step": 590
    },
    {
      "epoch": 0.3860852523272905,
      "grad_norm": Infinity,
      "learning_rate": 0.00014099660590590114,
      "loss": 2.2974,
      "step": 591
    },
    {
      "epoch": 0.3867385268659154,
      "grad_norm": Infinity,
      "learning_rate": 0.00014080290226309224,
      "loss": 2.2827,
      "step": 592
    },
    {
      "epoch": 0.38739180140454027,
      "grad_norm": Infinity,
      "learning_rate": 0.00014060901476877107,
      "loss": 2.8547,
      "step": 593
    },
    {
      "epoch": 0.3880450759431651,
      "grad_norm": Infinity,
      "learning_rate": 0.00014041494429656442,
      "loss": 2.4635,
      "step": 594
    },
    {
      "epoch": 0.38869835048178997,
      "grad_norm": Infinity,
      "learning_rate": 0.00014022069172092352,
      "loss": 2.9595,
      "step": 595
    },
    {
      "epoch": 0.3893516250204148,
      "grad_norm": Infinity,
      "learning_rate": 0.00014002625791712021,
      "loss": 2.7913,
      "step": 596
    },
    {
      "epoch": 0.39000489955903966,
      "grad_norm": Infinity,
      "learning_rate": 0.00013983164376124286,
      "loss": 2.8025,
      "step": 597
    },
    {
      "epoch": 0.39065817409766457,
      "grad_norm": Infinity,
      "learning_rate": 0.0001396368501301925,
      "loss": 3.4075,
      "step": 598
    },
    {
      "epoch": 0.3913114486362894,
      "grad_norm": Infinity,
      "learning_rate": 0.0001394418779016789,
      "loss": 3.6667,
      "step": 599
    },
    {
      "epoch": 0.39196472317491426,
      "grad_norm": Infinity,
      "learning_rate": 0.00013924672795421637,
      "loss": 4.7289,
      "step": 600
    },
    {
      "epoch": 0.3926179977135391,
      "grad_norm": Infinity,
      "learning_rate": 0.00013905140116712026,
      "loss": 1.3319,
      "step": 601
    },
    {
      "epoch": 0.39327127225216396,
      "grad_norm": Infinity,
      "learning_rate": 0.00013885589842050253,
      "loss": 1.4609,
      "step": 602
    },
    {
      "epoch": 0.3939245467907888,
      "grad_norm": Infinity,
      "learning_rate": 0.0001386602205952681,
      "loss": 1.4258,
      "step": 603
    },
    {
      "epoch": 0.3945778213294137,
      "grad_norm": Infinity,
      "learning_rate": 0.00013846436857311068,
      "loss": 1.4458,
      "step": 604
    },
    {
      "epoch": 0.39523109586803856,
      "grad_norm": Infinity,
      "learning_rate": 0.000138268343236509,
      "loss": 1.2812,
      "step": 605
    },
    {
      "epoch": 0.3958843704066634,
      "grad_norm": Infinity,
      "learning_rate": 0.00013807214546872256,
      "loss": 1.5763,
      "step": 606
    },
    {
      "epoch": 0.39653764494528826,
      "grad_norm": Infinity,
      "learning_rate": 0.00013787577615378792,
      "loss": 1.5288,
      "step": 607
    },
    {
      "epoch": 0.3971909194839131,
      "grad_norm": Infinity,
      "learning_rate": 0.00013767923617651463,
      "loss": 1.5984,
      "step": 608
    },
    {
      "epoch": 0.39784419402253796,
      "grad_norm": Infinity,
      "learning_rate": 0.00013748252642248115,
      "loss": 1.5448,
      "step": 609
    },
    {
      "epoch": 0.3984974685611628,
      "grad_norm": Infinity,
      "learning_rate": 0.00013728564777803088,
      "loss": 1.745,
      "step": 610
    },
    {
      "epoch": 0.3991507430997877,
      "grad_norm": Infinity,
      "learning_rate": 0.00013708860113026834,
      "loss": 1.7803,
      "step": 611
    },
    {
      "epoch": 0.39980401763841256,
      "grad_norm": Infinity,
      "learning_rate": 0.00013689138736705495,
      "loss": 1.7888,
      "step": 612
    },
    {
      "epoch": 0.4004572921770374,
      "grad_norm": Infinity,
      "learning_rate": 0.0001366940073770052,
      "loss": 1.8004,
      "step": 613
    },
    {
      "epoch": 0.40111056671566225,
      "grad_norm": Infinity,
      "learning_rate": 0.00013649646204948255,
      "loss": 1.7229,
      "step": 614
    },
    {
      "epoch": 0.4017638412542871,
      "grad_norm": Infinity,
      "learning_rate": 0.00013629875227459532,
      "loss": 2.0557,
      "step": 615
    },
    {
      "epoch": 0.40241711579291195,
      "grad_norm": Infinity,
      "learning_rate": 0.00013610087894319302,
      "loss": 1.9848,
      "step": 616
    },
    {
      "epoch": 0.40307039033153685,
      "grad_norm": Infinity,
      "learning_rate": 0.00013590284294686203,
      "loss": 2.3338,
      "step": 617
    },
    {
      "epoch": 0.4037236648701617,
      "grad_norm": Infinity,
      "learning_rate": 0.00013570464517792153,
      "loss": 2.2821,
      "step": 618
    },
    {
      "epoch": 0.40437693940878655,
      "grad_norm": Infinity,
      "learning_rate": 0.00013550628652941985,
      "loss": 2.5272,
      "step": 619
    },
    {
      "epoch": 0.4050302139474114,
      "grad_norm": Infinity,
      "learning_rate": 0.0001353077678951301,
      "loss": 2.847,
      "step": 620
    },
    {
      "epoch": 0.40568348848603625,
      "grad_norm": Infinity,
      "learning_rate": 0.00013510909016954624,
      "loss": 3.0231,
      "step": 621
    },
    {
      "epoch": 0.4063367630246611,
      "grad_norm": Infinity,
      "learning_rate": 0.00013491025424787915,
      "loss": 2.7372,
      "step": 622
    },
    {
      "epoch": 0.40699003756328594,
      "grad_norm": Infinity,
      "learning_rate": 0.00013471126102605245,
      "loss": 3.2824,
      "step": 623
    },
    {
      "epoch": 0.40764331210191085,
      "grad_norm": Infinity,
      "learning_rate": 0.00013451211140069858,
      "loss": 3.2836,
      "step": 624
    },
    {
      "epoch": 0.4082965866405357,
      "grad_norm": Infinity,
      "learning_rate": 0.00013431280626915467,
      "loss": 4.2905,
      "step": 625
    },
    {
      "epoch": 0.40894986117916055,
      "grad_norm": Infinity,
      "learning_rate": 0.0001341133465294585,
      "loss": 1.1817,
      "step": 626
    },
    {
      "epoch": 0.4096031357177854,
      "grad_norm": Infinity,
      "learning_rate": 0.0001339137330803446,
      "loss": 1.3825,
      "step": 627
    },
    {
      "epoch": 0.41025641025641024,
      "grad_norm": Infinity,
      "learning_rate": 0.00013371396682124005,
      "loss": 1.4202,
      "step": 628
    },
    {
      "epoch": 0.4109096847950351,
      "grad_norm": Infinity,
      "learning_rate": 0.0001335140486522604,
      "loss": 1.4288,
      "step": 629
    },
    {
      "epoch": 0.41156295933366,
      "grad_norm": Infinity,
      "learning_rate": 0.00013331397947420576,
      "loss": 1.4444,
      "step": 630
    },
    {
      "epoch": 0.41221623387228484,
      "grad_norm": Infinity,
      "learning_rate": 0.00013311376018855663,
      "loss": 1.5484,
      "step": 631
    },
    {
      "epoch": 0.4128695084109097,
      "grad_norm": Infinity,
      "learning_rate": 0.0001329133916974699,
      "loss": 1.5702,
      "step": 632
    },
    {
      "epoch": 0.41352278294953454,
      "grad_norm": Infinity,
      "learning_rate": 0.00013271287490377467,
      "loss": 1.6131,
      "step": 633
    },
    {
      "epoch": 0.4141760574881594,
      "grad_norm": Infinity,
      "learning_rate": 0.00013251221071096836,
      "loss": 1.6361,
      "step": 634
    },
    {
      "epoch": 0.41482933202678424,
      "grad_norm": Infinity,
      "learning_rate": 0.00013231140002321253,
      "loss": 1.6556,
      "step": 635
    },
    {
      "epoch": 0.41548260656540914,
      "grad_norm": Infinity,
      "learning_rate": 0.00013211044374532882,
      "loss": 1.785,
      "step": 636
    },
    {
      "epoch": 0.416135881104034,
      "grad_norm": Infinity,
      "learning_rate": 0.00013190934278279487,
      "loss": 1.7989,
      "step": 637
    },
    {
      "epoch": 0.41678915564265884,
      "grad_norm": Infinity,
      "learning_rate": 0.00013170809804174022,
      "loss": 1.8345,
      "step": 638
    },
    {
      "epoch": 0.4174424301812837,
      "grad_norm": Infinity,
      "learning_rate": 0.00013150671042894228,
      "loss": 1.8782,
      "step": 639
    },
    {
      "epoch": 0.41809570471990853,
      "grad_norm": Infinity,
      "learning_rate": 0.00013130518085182225,
      "loss": 1.9949,
      "step": 640
    },
    {
      "epoch": 0.4187489792585334,
      "grad_norm": Infinity,
      "learning_rate": 0.00013110351021844094,
      "loss": 2.1345,
      "step": 641
    },
    {
      "epoch": 0.41940225379715823,
      "grad_norm": Infinity,
      "learning_rate": 0.00013090169943749476,
      "loss": 2.5038,
      "step": 642
    },
    {
      "epoch": 0.42005552833578313,
      "grad_norm": Infinity,
      "learning_rate": 0.0001306997494183116,
      "loss": 2.7144,
      "step": 643
    },
    {
      "epoch": 0.420708802874408,
      "grad_norm": Infinity,
      "learning_rate": 0.00013049766107084678,
      "loss": 2.7682,
      "step": 644
    },
    {
      "epoch": 0.42136207741303283,
      "grad_norm": Infinity,
      "learning_rate": 0.00013029543530567884,
      "loss": 2.5713,
      "step": 645
    },
    {
      "epoch": 0.4220153519516577,
      "grad_norm": Infinity,
      "learning_rate": 0.00013009307303400556,
      "loss": 2.7619,
      "step": 646
    },
    {
      "epoch": 0.42266862649028253,
      "grad_norm": Infinity,
      "learning_rate": 0.0001298905751676397,
      "loss": 2.9728,
      "step": 647
    },
    {
      "epoch": 0.4233219010289074,
      "grad_norm": Infinity,
      "learning_rate": 0.0001296879426190051,
      "loss": 2.9242,
      "step": 648
    },
    {
      "epoch": 0.4239751755675323,
      "grad_norm": Infinity,
      "learning_rate": 0.00012948517630113245,
      "loss": 3.7497,
      "step": 649
    },
    {
      "epoch": 0.42462845010615713,
      "grad_norm": Infinity,
      "learning_rate": 0.00012928227712765504,
      "loss": 3.9534,
      "step": 650
    },
    {
      "epoch": 0.425281724644782,
      "grad_norm": Infinity,
      "learning_rate": 0.00012907924601280498,
      "loss": 1.3583,
      "step": 651
    },
    {
      "epoch": 0.4259349991834068,
      "grad_norm": Infinity,
      "learning_rate": 0.0001288760838714088,
      "loss": 1.3782,
      "step": 652
    },
    {
      "epoch": 0.4265882737220317,
      "grad_norm": Infinity,
      "learning_rate": 0.0001286727916188834,
      "loss": 1.3568,
      "step": 653
    },
    {
      "epoch": 0.4272415482606565,
      "grad_norm": Infinity,
      "learning_rate": 0.00012846937017123197,
      "loss": 1.3555,
      "step": 654
    },
    {
      "epoch": 0.42789482279928137,
      "grad_norm": Infinity,
      "learning_rate": 0.00012826582044503978,
      "loss": 1.5263,
      "step": 655
    },
    {
      "epoch": 0.4285480973379063,
      "grad_norm": Infinity,
      "learning_rate": 0.00012806214335747018,
      "loss": 1.502,
      "step": 656
    },
    {
      "epoch": 0.4292013718765311,
      "grad_norm": Infinity,
      "learning_rate": 0.00012785833982626032,
      "loss": 1.4846,
      "step": 657
    },
    {
      "epoch": 0.42985464641515597,
      "grad_norm": Infinity,
      "learning_rate": 0.00012765441076971712,
      "loss": 1.5687,
      "step": 658
    },
    {
      "epoch": 0.4305079209537808,
      "grad_norm": Infinity,
      "learning_rate": 0.0001274503571067131,
      "loss": 1.6626,
      "step": 659
    },
    {
      "epoch": 0.43116119549240567,
      "grad_norm": Infinity,
      "learning_rate": 0.0001272461797566823,
      "loss": 1.7197,
      "step": 660
    },
    {
      "epoch": 0.4318144700310305,
      "grad_norm": Infinity,
      "learning_rate": 0.0001270418796396159,
      "loss": 1.7319,
      "step": 661
    },
    {
      "epoch": 0.4324677445696554,
      "grad_norm": Infinity,
      "learning_rate": 0.00012683745767605846,
      "loss": 1.8091,
      "step": 662
    },
    {
      "epoch": 0.43312101910828027,
      "grad_norm": Infinity,
      "learning_rate": 0.00012663291478710336,
      "loss": 1.8593,
      "step": 663
    },
    {
      "epoch": 0.4337742936469051,
      "grad_norm": Infinity,
      "learning_rate": 0.00012642825189438902,
      "loss": 1.9488,
      "step": 664
    },
    {
      "epoch": 0.43442756818552997,
      "grad_norm": Infinity,
      "learning_rate": 0.00012622346992009447,
      "loss": 2.1199,
      "step": 665
    },
    {
      "epoch": 0.4350808427241548,
      "grad_norm": Infinity,
      "learning_rate": 0.0001260185697869353,
      "loss": 2.0434,
      "step": 666
    },
    {
      "epoch": 0.43573411726277966,
      "grad_norm": Infinity,
      "learning_rate": 0.00012581355241815965,
      "loss": 2.3479,
      "step": 667
    },
    {
      "epoch": 0.43638739180140457,
      "grad_norm": Infinity,
      "learning_rate": 0.00012560841873754375,
      "loss": 2.2125,
      "step": 668
    },
    {
      "epoch": 0.4370406663400294,
      "grad_norm": Infinity,
      "learning_rate": 0.00012540316966938795,
      "loss": 2.3785,
      "step": 669
    },
    {
      "epoch": 0.43769394087865426,
      "grad_norm": Infinity,
      "learning_rate": 0.00012519780613851254,
      "loss": 2.6029,
      "step": 670
    },
    {
      "epoch": 0.4383472154172791,
      "grad_norm": Infinity,
      "learning_rate": 0.00012499232907025353,
      "loss": 2.5433,
      "step": 671
    },
    {
      "epoch": 0.43900048995590396,
      "grad_norm": Infinity,
      "learning_rate": 0.00012478673939045863,
      "loss": 2.7926,
      "step": 672
    },
    {
      "epoch": 0.4396537644945288,
      "grad_norm": Infinity,
      "learning_rate": 0.00012458103802548274,
      "loss": 3.0005,
      "step": 673
    },
    {
      "epoch": 0.44030703903315366,
      "grad_norm": Infinity,
      "learning_rate": 0.00012437522590218417,
      "loss": 3.1366,
      "step": 674
    },
    {
      "epoch": 0.44096031357177856,
      "grad_norm": Infinity,
      "learning_rate": 0.00012416930394792026,
      "loss": 4.2293,
      "step": 675
    },
    {
      "epoch": 0.4416135881104034,
      "grad_norm": Infinity,
      "learning_rate": 0.0001239632730905432,
      "loss": 1.2318,
      "step": 676
    },
    {
      "epoch": 0.44226686264902826,
      "grad_norm": Infinity,
      "learning_rate": 0.0001237571342583959,
      "loss": 1.3872,
      "step": 677
    },
    {
      "epoch": 0.4429201371876531,
      "grad_norm": Infinity,
      "learning_rate": 0.00012355088838030776,
      "loss": 1.4916,
      "step": 678
    },
    {
      "epoch": 0.44357341172627796,
      "grad_norm": Infinity,
      "learning_rate": 0.00012334453638559057,
      "loss": 1.434,
      "step": 679
    },
    {
      "epoch": 0.4442266862649028,
      "grad_norm": Infinity,
      "learning_rate": 0.00012313807920403419,
      "loss": 1.457,
      "step": 680
    },
    {
      "epoch": 0.4448799608035277,
      "grad_norm": Infinity,
      "learning_rate": 0.00012293151776590245,
      "loss": 1.4525,
      "step": 681
    },
    {
      "epoch": 0.44553323534215256,
      "grad_norm": Infinity,
      "learning_rate": 0.00012272485300192902,
      "loss": 1.4693,
      "step": 682
    },
    {
      "epoch": 0.4461865098807774,
      "grad_norm": Infinity,
      "learning_rate": 0.00012251808584331304,
      "loss": 1.493,
      "step": 683
    },
    {
      "epoch": 0.44683978441940225,
      "grad_norm": Infinity,
      "learning_rate": 0.00012231121722171512,
      "loss": 1.5844,
      "step": 684
    },
    {
      "epoch": 0.4474930589580271,
      "grad_norm": Infinity,
      "learning_rate": 0.00012210424806925301,
      "loss": 1.6257,
      "step": 685
    },
    {
      "epoch": 0.44814633349665195,
      "grad_norm": Infinity,
      "learning_rate": 0.00012189717931849731,
      "loss": 1.6306,
      "step": 686
    },
    {
      "epoch": 0.4487996080352768,
      "grad_norm": Infinity,
      "learning_rate": 0.00012169001190246765,
      "loss": 1.7162,
      "step": 687
    },
    {
      "epoch": 0.4494528825739017,
      "grad_norm": Infinity,
      "learning_rate": 0.00012148274675462801,
      "loss": 1.8626,
      "step": 688
    },
    {
      "epoch": 0.45010615711252655,
      "grad_norm": Infinity,
      "learning_rate": 0.00012127538480888283,
      "loss": 1.857,
      "step": 689
    },
    {
      "epoch": 0.4507594316511514,
      "grad_norm": Infinity,
      "learning_rate": 0.00012106792699957263,
      "loss": 1.9262,
      "step": 690
    },
    {
      "epoch": 0.45141270618977625,
      "grad_norm": Infinity,
      "learning_rate": 0.00012086037426147003,
      "loss": 2.0395,
      "step": 691
    },
    {
      "epoch": 0.4520659807284011,
      "grad_norm": Infinity,
      "learning_rate": 0.00012065272752977526,
      "loss": 2.0832,
      "step": 692
    },
    {
      "epoch": 0.45271925526702594,
      "grad_norm": Infinity,
      "learning_rate": 0.00012044498774011209,
      "loss": 2.5049,
      "step": 693
    },
    {
      "epoch": 0.45337252980565085,
      "grad_norm": Infinity,
      "learning_rate": 0.00012023715582852357,
      "loss": 2.4738,
      "step": 694
    },
    {
      "epoch": 0.4540258043442757,
      "grad_norm": Infinity,
      "learning_rate": 0.00012002923273146794,
      "loss": 2.2563,
      "step": 695
    },
    {
      "epoch": 0.45467907888290054,
      "grad_norm": Infinity,
      "learning_rate": 0.0001198212193858142,
      "loss": 2.936,
      "step": 696
    },
    {
      "epoch": 0.4553323534215254,
      "grad_norm": Infinity,
      "learning_rate": 0.00011961311672883804,
      "loss": 3.3187,
      "step": 697
    },
    {
      "epoch": 0.45598562796015024,
      "grad_norm": Infinity,
      "learning_rate": 0.00011940492569821753,
      "loss": 3.1426,
      "step": 698
    },
    {
      "epoch": 0.4566389024987751,
      "grad_norm": Infinity,
      "learning_rate": 0.00011919664723202906,
      "loss": 3.4786,
      "step": 699
    },
    {
      "epoch": 0.4572921770374,
      "grad_norm": Infinity,
      "learning_rate": 0.00011898828226874284,
      "loss": 4.6187,
      "step": 700
    },
    {
      "epoch": 0.45794545157602484,
      "grad_norm": Infinity,
      "learning_rate": 0.00011877983174721892,
      "loss": 1.315,
      "step": 701
    },
    {
      "epoch": 0.4585987261146497,
      "grad_norm": Infinity,
      "learning_rate": 0.00011857129660670281,
      "loss": 1.3934,
      "step": 702
    },
    {
      "epoch": 0.45925200065327454,
      "grad_norm": Infinity,
      "learning_rate": 0.00011836267778682133,
      "loss": 1.4157,
      "step": 703
    },
    {
      "epoch": 0.4599052751918994,
      "grad_norm": Infinity,
      "learning_rate": 0.00011815397622757838,
      "loss": 1.3945,
      "step": 704
    },
    {
      "epoch": 0.46055854973052424,
      "grad_norm": Infinity,
      "learning_rate": 0.00011794519286935055,
      "loss": 1.4258,
      "step": 705
    },
    {
      "epoch": 0.4612118242691491,
      "grad_norm": Infinity,
      "learning_rate": 0.00011773632865288309,
      "loss": 1.4199,
      "step": 706
    },
    {
      "epoch": 0.461865098807774,
      "grad_norm": Infinity,
      "learning_rate": 0.00011752738451928566,
      "loss": 1.5205,
      "step": 707
    },
    {
      "epoch": 0.46251837334639884,
      "grad_norm": Infinity,
      "learning_rate": 0.00011731836141002787,
      "loss": 1.5761,
      "step": 708
    },
    {
      "epoch": 0.4631716478850237,
      "grad_norm": Infinity,
      "learning_rate": 0.00011710926026693525,
      "loss": 1.6457,
      "step": 709
    },
    {
      "epoch": 0.46382492242364853,
      "grad_norm": Infinity,
      "learning_rate": 0.00011690008203218493,
      "loss": 1.8223,
      "step": 710
    },
    {
      "epoch": 0.4644781969622734,
      "grad_norm": Infinity,
      "learning_rate": 0.0001166908276483014,
      "loss": 1.7146,
      "step": 711
    },
    {
      "epoch": 0.46513147150089823,
      "grad_norm": Infinity,
      "learning_rate": 0.00011648149805815227,
      "loss": 1.6361,
      "step": 712
    },
    {
      "epoch": 0.46578474603952313,
      "grad_norm": Infinity,
      "learning_rate": 0.000116272094204944,
      "loss": 1.7237,
      "step": 713
    },
    {
      "epoch": 0.466438020578148,
      "grad_norm": Infinity,
      "learning_rate": 0.00011606261703221772,
      "loss": 1.8705,
      "step": 714
    },
    {
      "epoch": 0.46709129511677283,
      "grad_norm": Infinity,
      "learning_rate": 0.0001158530674838449,
      "loss": 1.9701,
      "step": 715
    },
    {
      "epoch": 0.4677445696553977,
      "grad_norm": Infinity,
      "learning_rate": 0.0001156434465040231,
      "loss": 2.136,
      "step": 716
    },
    {
      "epoch": 0.46839784419402253,
      "grad_norm": Infinity,
      "learning_rate": 0.00011543375503727174,
      "loss": 2.0878,
      "step": 717
    },
    {
      "epoch": 0.4690511187326474,
      "grad_norm": Infinity,
      "learning_rate": 0.00011522399402842783,
      "loss": 2.4607,
      "step": 718
    },
    {
      "epoch": 0.4697043932712722,
      "grad_norm": Infinity,
      "learning_rate": 0.00011501416442264184,
      "loss": 2.4705,
      "step": 719
    },
    {
      "epoch": 0.47035766780989713,
      "grad_norm": Infinity,
      "learning_rate": 0.00011480426716537315,
      "loss": 2.6055,
      "step": 720
    },
    {
      "epoch": 0.471010942348522,
      "grad_norm": Infinity,
      "learning_rate": 0.00011459430320238611,
      "loss": 2.7481,
      "step": 721
    },
    {
      "epoch": 0.4716642168871468,
      "grad_norm": Infinity,
      "learning_rate": 0.00011438427347974554,
      "loss": 2.9075,
      "step": 722
    },
    {
      "epoch": 0.4723174914257717,
      "grad_norm": Infinity,
      "learning_rate": 0.00011417417894381268,
      "loss": 3.147,
      "step": 723
    },
    {
      "epoch": 0.4729707659643965,
      "grad_norm": Infinity,
      "learning_rate": 0.00011396402054124067,
      "loss": 3.0189,
      "step": 724
    },
    {
      "epoch": 0.47362404050302137,
      "grad_norm": Infinity,
      "learning_rate": 0.00011375379921897051,
      "loss": 4.2631,
      "step": 725
    },
    {
      "epoch": 0.4742773150416463,
      "grad_norm": Infinity,
      "learning_rate": 0.00011354351592422665,
      "loss": 1.4092,
      "step": 726
    },
    {
      "epoch": 0.4749305895802711,
      "grad_norm": Infinity,
      "learning_rate": 0.00011333317160451286,
      "loss": 1.4541,
      "step": 727
    },
    {
      "epoch": 0.47558386411889597,
      "grad_norm": Infinity,
      "learning_rate": 0.00011312276720760782,
      "loss": 1.4517,
      "step": 728
    },
    {
      "epoch": 0.4762371386575208,
      "grad_norm": Infinity,
      "learning_rate": 0.00011291230368156087,
      "loss": 1.3585,
      "step": 729
    },
    {
      "epoch": 0.47689041319614567,
      "grad_norm": Infinity,
      "learning_rate": 0.00011270178197468789,
      "loss": 1.4466,
      "step": 730
    },
    {
      "epoch": 0.4775436877347705,
      "grad_norm": Infinity,
      "learning_rate": 0.00011249120303556681,
      "loss": 1.6079,
      "step": 731
    },
    {
      "epoch": 0.4781969622733954,
      "grad_norm": Infinity,
      "learning_rate": 0.0001122805678130335,
      "loss": 1.5759,
      "step": 732
    },
    {
      "epoch": 0.47885023681202027,
      "grad_norm": Infinity,
      "learning_rate": 0.00011206987725617741,
      "loss": 1.4571,
      "step": 733
    },
    {
      "epoch": 0.4795035113506451,
      "grad_norm": Infinity,
      "learning_rate": 0.00011185913231433733,
      "loss": 1.6658,
      "step": 734
    },
    {
      "epoch": 0.48015678588926997,
      "grad_norm": Infinity,
      "learning_rate": 0.00011164833393709706,
      "loss": 1.7304,
      "step": 735
    },
    {
      "epoch": 0.4808100604278948,
      "grad_norm": Infinity,
      "learning_rate": 0.00011143748307428126,
      "loss": 1.6036,
      "step": 736
    },
    {
      "epoch": 0.48146333496651966,
      "grad_norm": Infinity,
      "learning_rate": 0.00011122658067595094,
      "loss": 1.7269,
      "step": 737
    },
    {
      "epoch": 0.4821166095051445,
      "grad_norm": Infinity,
      "learning_rate": 0.00011101562769239946,
      "loss": 1.7812,
      "step": 738
    },
    {
      "epoch": 0.4827698840437694,
      "grad_norm": Infinity,
      "learning_rate": 0.00011080462507414806,
      "loss": 1.9052,
      "step": 739
    },
    {
      "epoch": 0.48342315858239426,
      "grad_norm": Infinity,
      "learning_rate": 0.00011059357377194161,
      "loss": 2.0699,
      "step": 740
    },
    {
      "epoch": 0.4840764331210191,
      "grad_norm": Infinity,
      "learning_rate": 0.00011038247473674434,
      "loss": 2.2282,
      "step": 741
    },
    {
      "epoch": 0.48472970765964396,
      "grad_norm": Infinity,
      "learning_rate": 0.0001101713289197356,
      "loss": 2.5375,
      "step": 742
    },
    {
      "epoch": 0.4853829821982688,
      "grad_norm": Infinity,
      "learning_rate": 0.0001099601372723055,
      "loss": 2.5702,
      "step": 743
    },
    {
      "epoch": 0.48603625673689366,
      "grad_norm": Infinity,
      "learning_rate": 0.00010974890074605062,
      "loss": 2.7113,
      "step": 744
    },
    {
      "epoch": 0.48668953127551856,
      "grad_norm": Infinity,
      "learning_rate": 0.00010953762029276982,
      "loss": 3.0058,
      "step": 745
    },
    {
      "epoch": 0.4873428058141434,
      "grad_norm": Infinity,
      "learning_rate": 0.00010932629686445986,
      "loss": 2.9393,
      "step": 746
    },
    {
      "epoch": 0.48799608035276826,
      "grad_norm": Infinity,
      "learning_rate": 0.00010911493141331113,
      "loss": 3.0283,
      "step": 747
    },
    {
      "epoch": 0.4886493548913931,
      "grad_norm": Infinity,
      "learning_rate": 0.00010890352489170341,
      "loss": 3.3168,
      "step": 748
    },
    {
      "epoch": 0.48930262943001795,
      "grad_norm": Infinity,
      "learning_rate": 0.00010869207825220147,
      "loss": 3.5297,
      "step": 749
    },
    {
      "epoch": 0.4899559039686428,
      "grad_norm": Infinity,
      "learning_rate": 0.00010848059244755093,
      "loss": 4.1073,
      "step": 750
    },
    {
      "epoch": 0.4906091785072677,
      "grad_norm": Infinity,
      "learning_rate": 0.0001082690684306738,
      "loss": 1.3489,
      "step": 751
    },
    {
      "epoch": 0.49126245304589256,
      "grad_norm": Infinity,
      "learning_rate": 0.00010805750715466429,
      "loss": 1.3607,
      "step": 752
    },
    {
      "epoch": 0.4919157275845174,
      "grad_norm": Infinity,
      "learning_rate": 0.0001078459095727845,
      "loss": 1.4329,
      "step": 753
    },
    {
      "epoch": 0.49256900212314225,
      "grad_norm": Infinity,
      "learning_rate": 0.00010763427663846015,
      "loss": 1.4151,
      "step": 754
    },
    {
      "epoch": 0.4932222766617671,
      "grad_norm": Infinity,
      "learning_rate": 0.00010742260930527625,
      "loss": 1.3808,
      "step": 755
    },
    {
      "epoch": 0.49387555120039195,
      "grad_norm": Infinity,
      "learning_rate": 0.00010721090852697275,
      "loss": 1.5492,
      "step": 756
    },
    {
      "epoch": 0.4945288257390168,
      "grad_norm": Infinity,
      "learning_rate": 0.00010699917525744032,
      "loss": 1.4695,
      "step": 757
    },
    {
      "epoch": 0.4951821002776417,
      "grad_norm": Infinity,
      "learning_rate": 0.00010678741045071609,
      "loss": 1.5456,
      "step": 758
    },
    {
      "epoch": 0.49583537481626655,
      "grad_norm": Infinity,
      "learning_rate": 0.0001065756150609792,
      "loss": 1.5857,
      "step": 759
    },
    {
      "epoch": 0.4964886493548914,
      "grad_norm": Infinity,
      "learning_rate": 0.00010636379004254664,
      "loss": 1.4613,
      "step": 760
    },
    {
      "epoch": 0.49714192389351625,
      "grad_norm": Infinity,
      "learning_rate": 0.0001061519363498689,
      "loss": 1.704,
      "step": 761
    },
    {
      "epoch": 0.4977951984321411,
      "grad_norm": Infinity,
      "learning_rate": 0.00010594005493752568,
      "loss": 1.6429,
      "step": 762
    },
    {
      "epoch": 0.49844847297076594,
      "grad_norm": Infinity,
      "learning_rate": 0.00010572814676022158,
      "loss": 1.7032,
      "step": 763
    },
    {
      "epoch": 0.49910174750939085,
      "grad_norm": Infinity,
      "learning_rate": 0.00010551621277278176,
      "loss": 1.6818,
      "step": 764
    },
    {
      "epoch": 0.4997550220480157,
      "grad_norm": Infinity,
      "learning_rate": 0.00010530425393014774,
      "loss": 2.0321,
      "step": 765
    },
    {
      "epoch": 0.5004082965866405,
      "grad_norm": Infinity,
      "learning_rate": 0.00010509227118737298,
      "loss": 1.8955,
      "step": 766
    },
    {
      "epoch": 0.5004082965866405,
      "eval_loss": 2.0970346927642822,
      "eval_runtime": 161.1777,
      "eval_samples_per_second": 15.995,
      "eval_steps_per_second": 4.002,
      "step": 766
    },
    {
      "epoch": 0.5010615711252654,
      "grad_norm": Infinity,
      "learning_rate": 0.00010488026549961863,
      "loss": 2.3278,
      "step": 767
    },
    {
      "epoch": 0.5017148456638902,
      "grad_norm": Infinity,
      "learning_rate": 0.00010466823782214927,
      "loss": 2.3015,
      "step": 768
    },
    {
      "epoch": 0.5023681202025151,
      "grad_norm": Infinity,
      "learning_rate": 0.00010445618911032853,
      "loss": 2.6376,
      "step": 769
    },
    {
      "epoch": 0.5030213947411399,
      "grad_norm": Infinity,
      "learning_rate": 0.00010424412031961484,
      "loss": 2.5291,
      "step": 770
    },
    {
      "epoch": 0.5036746692797648,
      "grad_norm": Infinity,
      "learning_rate": 0.00010403203240555707,
      "loss": 2.7269,
      "step": 771
    },
    {
      "epoch": 0.5043279438183896,
      "grad_norm": Infinity,
      "learning_rate": 0.0001038199263237903,
      "loss": 2.5242,
      "step": 772
    },
    {
      "epoch": 0.5049812183570145,
      "grad_norm": Infinity,
      "learning_rate": 0.00010360780303003139,
      "loss": 3.0819,
      "step": 773
    },
    {
      "epoch": 0.5056344928956394,
      "grad_norm": Infinity,
      "learning_rate": 0.00010339566348007487,
      "loss": 3.4961,
      "step": 774
    },
    {
      "epoch": 0.5062877674342643,
      "grad_norm": Infinity,
      "learning_rate": 0.00010318350862978848,
      "loss": 4.0663,
      "step": 775
    },
    {
      "epoch": 0.5069410419728891,
      "grad_norm": Infinity,
      "learning_rate": 0.00010297133943510879,
      "loss": 1.2678,
      "step": 776
    },
    {
      "epoch": 0.507594316511514,
      "grad_norm": Infinity,
      "learning_rate": 0.00010275915685203712,
      "loss": 1.3923,
      "step": 777
    },
    {
      "epoch": 0.5082475910501388,
      "grad_norm": Infinity,
      "learning_rate": 0.00010254696183663511,
      "loss": 1.403,
      "step": 778
    },
    {
      "epoch": 0.5089008655887637,
      "grad_norm": Infinity,
      "learning_rate": 0.00010233475534502042,
      "loss": 1.4813,
      "step": 779
    },
    {
      "epoch": 0.5095541401273885,
      "grad_norm": Infinity,
      "learning_rate": 0.00010212253833336237,
      "loss": 1.454,
      "step": 780
    },
    {
      "epoch": 0.5102074146660134,
      "grad_norm": Infinity,
      "learning_rate": 0.00010191031175787768,
      "loss": 1.4327,
      "step": 781
    },
    {
      "epoch": 0.5108606892046382,
      "grad_norm": Infinity,
      "learning_rate": 0.00010169807657482623,
      "loss": 1.4784,
      "step": 782
    },
    {
      "epoch": 0.5115139637432631,
      "grad_norm": Infinity,
      "learning_rate": 0.00010148583374050667,
      "loss": 1.6763,
      "step": 783
    },
    {
      "epoch": 0.5121672382818879,
      "grad_norm": Infinity,
      "learning_rate": 0.00010127358421125204,
      "loss": 1.6503,
      "step": 784
    },
    {
      "epoch": 0.5128205128205128,
      "grad_norm": Infinity,
      "learning_rate": 0.00010106132894342564,
      "loss": 1.8225,
      "step": 785
    },
    {
      "epoch": 0.5134737873591376,
      "grad_norm": Infinity,
      "learning_rate": 0.00010084906889341656,
      "loss": 1.5835,
      "step": 786
    },
    {
      "epoch": 0.5141270618977626,
      "grad_norm": Infinity,
      "learning_rate": 0.00010063680501763552,
      "loss": 1.8056,
      "step": 787
    },
    {
      "epoch": 0.5147803364363874,
      "grad_norm": Infinity,
      "learning_rate": 0.00010042453827251044,
      "loss": 1.9522,
      "step": 788
    },
    {
      "epoch": 0.5154336109750123,
      "grad_norm": Infinity,
      "learning_rate": 0.00010021226961448209,
      "loss": 2.0843,
      "step": 789
    },
    {
      "epoch": 0.5160868855136371,
      "grad_norm": Infinity,
      "learning_rate": 0.0001,
      "loss": 2.0254,
      "step": 790
    },
    {
      "epoch": 0.516740160052262,
      "grad_norm": Infinity,
      "learning_rate": 9.97877303855179e-05,
      "loss": 2.1822,
      "step": 791
    },
    {
      "epoch": 0.5173934345908868,
      "grad_norm": Infinity,
      "learning_rate": 9.957546172748958e-05,
      "loss": 2.2943,
      "step": 792
    },
    {
      "epoch": 0.5180467091295117,
      "grad_norm": Infinity,
      "learning_rate": 9.936319498236446e-05,
      "loss": 2.3479,
      "step": 793
    },
    {
      "epoch": 0.5186999836681365,
      "grad_norm": Infinity,
      "learning_rate": 9.915093110658346e-05,
      "loss": 2.7068,
      "step": 794
    },
    {
      "epoch": 0.5193532582067614,
      "grad_norm": Infinity,
      "learning_rate": 9.89386710565744e-05,
      "loss": 2.4189,
      "step": 795
    },
    {
      "epoch": 0.5200065327453862,
      "grad_norm": Infinity,
      "learning_rate": 9.8726415788748e-05,
      "loss": 3.0649,
      "step": 796
    },
    {
      "epoch": 0.5206598072840111,
      "grad_norm": Infinity,
      "learning_rate": 9.851416625949334e-05,
      "loss": 3.1619,
      "step": 797
    },
    {
      "epoch": 0.5213130818226359,
      "grad_norm": Infinity,
      "learning_rate": 9.830192342517379e-05,
      "loss": 3.4088,
      "step": 798
    },
    {
      "epoch": 0.5219663563612609,
      "grad_norm": Infinity,
      "learning_rate": 9.808968824212234e-05,
      "loss": 3.5337,
      "step": 799
    },
    {
      "epoch": 0.5226196308998857,
      "grad_norm": Infinity,
      "learning_rate": 9.787746166663764e-05,
      "loss": 4.3343,
      "step": 800
    },
    {
      "epoch": 0.5232729054385106,
      "grad_norm": Infinity,
      "learning_rate": 9.76652446549796e-05,
      "loss": 1.4279,
      "step": 801
    },
    {
      "epoch": 0.5239261799771354,
      "grad_norm": Infinity,
      "learning_rate": 9.745303816336489e-05,
      "loss": 1.3824,
      "step": 802
    },
    {
      "epoch": 0.5245794545157603,
      "grad_norm": Infinity,
      "learning_rate": 9.724084314796292e-05,
      "loss": 1.3597,
      "step": 803
    },
    {
      "epoch": 0.5252327290543851,
      "grad_norm": Infinity,
      "learning_rate": 9.702866056489125e-05,
      "loss": 1.3498,
      "step": 804
    },
    {
      "epoch": 0.52588600359301,
      "grad_norm": Infinity,
      "learning_rate": 9.681649137021158e-05,
      "loss": 1.3747,
      "step": 805
    },
    {
      "epoch": 0.5265392781316348,
      "grad_norm": Infinity,
      "learning_rate": 9.660433651992514e-05,
      "loss": 1.3791,
      "step": 806
    },
    {
      "epoch": 0.5271925526702597,
      "grad_norm": Infinity,
      "learning_rate": 9.639219696996861e-05,
      "loss": 1.5428,
      "step": 807
    },
    {
      "epoch": 0.5278458272088845,
      "grad_norm": Infinity,
      "learning_rate": 9.618007367620972e-05,
      "loss": 1.586,
      "step": 808
    },
    {
      "epoch": 0.5284991017475094,
      "grad_norm": Infinity,
      "learning_rate": 9.596796759444293e-05,
      "loss": 1.5683,
      "step": 809
    },
    {
      "epoch": 0.5291523762861342,
      "grad_norm": Infinity,
      "learning_rate": 9.57558796803852e-05,
      "loss": 1.6566,
      "step": 810
    },
    {
      "epoch": 0.5298056508247591,
      "grad_norm": Infinity,
      "learning_rate": 9.554381088967148e-05,
      "loss": 1.7084,
      "step": 811
    },
    {
      "epoch": 0.530458925363384,
      "grad_norm": Infinity,
      "learning_rate": 9.533176217785075e-05,
      "loss": 1.9305,
      "step": 812
    },
    {
      "epoch": 0.5311121999020089,
      "grad_norm": Infinity,
      "learning_rate": 9.51197345003814e-05,
      "loss": 1.8239,
      "step": 813
    },
    {
      "epoch": 0.5317654744406337,
      "grad_norm": Infinity,
      "learning_rate": 9.490772881262709e-05,
      "loss": 1.7874,
      "step": 814
    },
    {
      "epoch": 0.5324187489792586,
      "grad_norm": Infinity,
      "learning_rate": 9.46957460698523e-05,
      "loss": 1.8129,
      "step": 815
    },
    {
      "epoch": 0.5330720235178834,
      "grad_norm": Infinity,
      "learning_rate": 9.448378722721825e-05,
      "loss": 2.0876,
      "step": 816
    },
    {
      "epoch": 0.5337252980565083,
      "grad_norm": Infinity,
      "learning_rate": 9.427185323977845e-05,
      "loss": 2.2503,
      "step": 817
    },
    {
      "epoch": 0.5343785725951331,
      "grad_norm": Infinity,
      "learning_rate": 9.405994506247432e-05,
      "loss": 2.2582,
      "step": 818
    },
    {
      "epoch": 0.535031847133758,
      "grad_norm": Infinity,
      "learning_rate": 9.384806365013113e-05,
      "loss": 2.4585,
      "step": 819
    },
    {
      "epoch": 0.5356851216723828,
      "grad_norm": Infinity,
      "learning_rate": 9.363620995745337e-05,
      "loss": 2.5637,
      "step": 820
    },
    {
      "epoch": 0.5363383962110077,
      "grad_norm": Infinity,
      "learning_rate": 9.342438493902085e-05,
      "loss": 2.8806,
      "step": 821
    },
    {
      "epoch": 0.5369916707496325,
      "grad_norm": Infinity,
      "learning_rate": 9.321258954928393e-05,
      "loss": 2.8588,
      "step": 822
    },
    {
      "epoch": 0.5376449452882573,
      "grad_norm": Infinity,
      "learning_rate": 9.300082474255967e-05,
      "loss": 3.0883,
      "step": 823
    },
    {
      "epoch": 0.5382982198268822,
      "grad_norm": Infinity,
      "learning_rate": 9.278909147302727e-05,
      "loss": 3.3387,
      "step": 824
    },
    {
      "epoch": 0.5389514943655072,
      "grad_norm": Infinity,
      "learning_rate": 9.257739069472374e-05,
      "loss": 4.0844,
      "step": 825
    },
    {
      "epoch": 0.539604768904132,
      "grad_norm": Infinity,
      "learning_rate": 9.236572336153986e-05,
      "loss": 1.3864,
      "step": 826
    },
    {
      "epoch": 0.5402580434427569,
      "grad_norm": Infinity,
      "learning_rate": 9.215409042721552e-05,
      "loss": 1.3945,
      "step": 827
    },
    {
      "epoch": 0.5409113179813817,
      "grad_norm": Infinity,
      "learning_rate": 9.194249284533576e-05,
      "loss": 1.4255,
      "step": 828
    },
    {
      "epoch": 0.5415645925200065,
      "grad_norm": Infinity,
      "learning_rate": 9.173093156932623e-05,
      "loss": 1.3899,
      "step": 829
    },
    {
      "epoch": 0.5422178670586314,
      "grad_norm": Infinity,
      "learning_rate": 9.151940755244912e-05,
      "loss": 1.4801,
      "step": 830
    },
    {
      "epoch": 0.5428711415972562,
      "grad_norm": Infinity,
      "learning_rate": 9.130792174779854e-05,
      "loss": 1.5015,
      "step": 831
    },
    {
      "epoch": 0.5435244161358811,
      "grad_norm": Infinity,
      "learning_rate": 9.109647510829657e-05,
      "loss": 1.5265,
      "step": 832
    },
    {
      "epoch": 0.5441776906745059,
      "grad_norm": Infinity,
      "learning_rate": 9.088506858668888e-05,
      "loss": 1.5976,
      "step": 833
    },
    {
      "epoch": 0.5448309652131308,
      "grad_norm": Infinity,
      "learning_rate": 9.067370313554015e-05,
      "loss": 1.5211,
      "step": 834
    },
    {
      "epoch": 0.5454842397517556,
      "grad_norm": Infinity,
      "learning_rate": 9.04623797072302e-05,
      "loss": 1.7078,
      "step": 835
    },
    {
      "epoch": 0.5461375142903805,
      "grad_norm": Infinity,
      "learning_rate": 9.025109925394939e-05,
      "loss": 1.6801,
      "step": 836
    },
    {
      "epoch": 0.5467907888290053,
      "grad_norm": Infinity,
      "learning_rate": 9.003986272769455e-05,
      "loss": 1.755,
      "step": 837
    },
    {
      "epoch": 0.5474440633676303,
      "grad_norm": Infinity,
      "learning_rate": 8.982867108026442e-05,
      "loss": 1.796,
      "step": 838
    },
    {
      "epoch": 0.5480973379062551,
      "grad_norm": Infinity,
      "learning_rate": 8.961752526325565e-05,
      "loss": 1.9277,
      "step": 839
    },
    {
      "epoch": 0.54875061244488,
      "grad_norm": Infinity,
      "learning_rate": 8.94064262280584e-05,
      "loss": 2.0079,
      "step": 840
    },
    {
      "epoch": 0.5494038869835048,
      "grad_norm": Infinity,
      "learning_rate": 8.919537492585194e-05,
      "loss": 2.1747,
      "step": 841
    },
    {
      "epoch": 0.5500571615221297,
      "grad_norm": Infinity,
      "learning_rate": 8.898437230760058e-05,
      "loss": 2.2313,
      "step": 842
    },
    {
      "epoch": 0.5507104360607545,
      "grad_norm": Infinity,
      "learning_rate": 8.877341932404909e-05,
      "loss": 2.6037,
      "step": 843
    },
    {
      "epoch": 0.5513637105993794,
      "grad_norm": Infinity,
      "learning_rate": 8.856251692571879e-05,
      "loss": 2.5185,
      "step": 844
    },
    {
      "epoch": 0.5520169851380042,
      "grad_norm": Infinity,
      "learning_rate": 8.835166606290295e-05,
      "loss": 2.5605,
      "step": 845
    },
    {
      "epoch": 0.5526702596766291,
      "grad_norm": Infinity,
      "learning_rate": 8.814086768566272e-05,
      "loss": 3.0754,
      "step": 846
    },
    {
      "epoch": 0.5533235342152539,
      "grad_norm": Infinity,
      "learning_rate": 8.793012274382261e-05,
      "loss": 3.1223,
      "step": 847
    },
    {
      "epoch": 0.5539768087538788,
      "grad_norm": Infinity,
      "learning_rate": 8.771943218696649e-05,
      "loss": 3.3635,
      "step": 848
    },
    {
      "epoch": 0.5546300832925036,
      "grad_norm": Infinity,
      "learning_rate": 8.750879696443321e-05,
      "loss": 3.5695,
      "step": 849
    },
    {
      "epoch": 0.5552833578311285,
      "grad_norm": Infinity,
      "learning_rate": 8.729821802531212e-05,
      "loss": 4.3607,
      "step": 850
    },
    {
      "epoch": 0.5559366323697534,
      "grad_norm": Infinity,
      "learning_rate": 8.708769631843916e-05,
      "loss": 1.1649,
      "step": 851
    },
    {
      "epoch": 0.5565899069083783,
      "grad_norm": Infinity,
      "learning_rate": 8.687723279239222e-05,
      "loss": 1.412,
      "step": 852
    },
    {
      "epoch": 0.5572431814470031,
      "grad_norm": Infinity,
      "learning_rate": 8.666682839548719e-05,
      "loss": 1.341,
      "step": 853
    },
    {
      "epoch": 0.557896455985628,
      "grad_norm": Infinity,
      "learning_rate": 8.645648407577338e-05,
      "loss": 1.3507,
      "step": 854
    },
    {
      "epoch": 0.5585497305242528,
      "grad_norm": Infinity,
      "learning_rate": 8.624620078102951e-05,
      "loss": 1.4897,
      "step": 855
    },
    {
      "epoch": 0.5592030050628777,
      "grad_norm": Infinity,
      "learning_rate": 8.603597945875935e-05,
      "loss": 1.4492,
      "step": 856
    },
    {
      "epoch": 0.5598562796015025,
      "grad_norm": Infinity,
      "learning_rate": 8.582582105618733e-05,
      "loss": 1.4236,
      "step": 857
    },
    {
      "epoch": 0.5605095541401274,
      "grad_norm": Infinity,
      "learning_rate": 8.561572652025447e-05,
      "loss": 1.5401,
      "step": 858
    },
    {
      "epoch": 0.5611628286787522,
      "grad_norm": Infinity,
      "learning_rate": 8.540569679761391e-05,
      "loss": 1.6108,
      "step": 859
    },
    {
      "epoch": 0.5618161032173771,
      "grad_norm": Infinity,
      "learning_rate": 8.519573283462687e-05,
      "loss": 1.582,
      "step": 860
    },
    {
      "epoch": 0.5624693777560019,
      "grad_norm": Infinity,
      "learning_rate": 8.498583557735819e-05,
      "loss": 1.6815,
      "step": 861
    },
    {
      "epoch": 0.5631226522946268,
      "grad_norm": Infinity,
      "learning_rate": 8.47760059715722e-05,
      "loss": 1.5396,
      "step": 862
    },
    {
      "epoch": 0.5637759268332517,
      "grad_norm": Infinity,
      "learning_rate": 8.456624496272829e-05,
      "loss": 1.7314,
      "step": 863
    },
    {
      "epoch": 0.5644292013718766,
      "grad_norm": Infinity,
      "learning_rate": 8.435655349597689e-05,
      "loss": 1.9425,
      "step": 864
    },
    {
      "epoch": 0.5650824759105014,
      "grad_norm": Infinity,
      "learning_rate": 8.414693251615512e-05,
      "loss": 1.8167,
      "step": 865
    },
    {
      "epoch": 0.5657357504491263,
      "grad_norm": Infinity,
      "learning_rate": 8.393738296778228e-05,
      "loss": 2.0253,
      "step": 866
    },
    {
      "epoch": 0.5663890249877511,
      "grad_norm": Infinity,
      "learning_rate": 8.3727905795056e-05,
      "loss": 2.0881,
      "step": 867
    },
    {
      "epoch": 0.567042299526376,
      "grad_norm": Infinity,
      "learning_rate": 8.351850194184775e-05,
      "loss": 2.3032,
      "step": 868
    },
    {
      "epoch": 0.5676955740650008,
      "grad_norm": Infinity,
      "learning_rate": 8.330917235169867e-05,
      "loss": 2.4988,
      "step": 869
    },
    {
      "epoch": 0.5683488486036257,
      "grad_norm": Infinity,
      "learning_rate": 8.309991796781511e-05,
      "loss": 2.4876,
      "step": 870
    },
    {
      "epoch": 0.5690021231422505,
      "grad_norm": Infinity,
      "learning_rate": 8.289073973306478e-05,
      "loss": 2.951,
      "step": 871
    },
    {
      "epoch": 0.5696553976808754,
      "grad_norm": Infinity,
      "learning_rate": 8.268163858997215e-05,
      "loss": 2.8837,
      "step": 872
    },
    {
      "epoch": 0.5703086722195002,
      "grad_norm": Infinity,
      "learning_rate": 8.247261548071435e-05,
      "loss": 3.1574,
      "step": 873
    },
    {
      "epoch": 0.5709619467581251,
      "grad_norm": Infinity,
      "learning_rate": 8.226367134711692e-05,
      "loss": 3.2576,
      "step": 874
    },
    {
      "epoch": 0.5716152212967499,
      "grad_norm": Infinity,
      "learning_rate": 8.205480713064946e-05,
      "loss": 4.0353,
      "step": 875
    },
    {
      "epoch": 0.5722684958353749,
      "grad_norm": Infinity,
      "learning_rate": 8.184602377242168e-05,
      "loss": 1.2324,
      "step": 876
    },
    {
      "epoch": 0.5729217703739997,
      "grad_norm": Infinity,
      "learning_rate": 8.163732221317868e-05,
      "loss": 1.3859,
      "step": 877
    },
    {
      "epoch": 0.5735750449126246,
      "grad_norm": Infinity,
      "learning_rate": 8.142870339329723e-05,
      "loss": 1.4323,
      "step": 878
    },
    {
      "epoch": 0.5742283194512494,
      "grad_norm": Infinity,
      "learning_rate": 8.12201682527811e-05,
      "loss": 1.4522,
      "step": 879
    },
    {
      "epoch": 0.5748815939898743,
      "grad_norm": Infinity,
      "learning_rate": 8.101171773125716e-05,
      "loss": 1.5134,
      "step": 880
    },
    {
      "epoch": 0.5755348685284991,
      "grad_norm": Infinity,
      "learning_rate": 8.080335276797097e-05,
      "loss": 1.5799,
      "step": 881
    },
    {
      "epoch": 0.576188143067124,
      "grad_norm": Infinity,
      "learning_rate": 8.059507430178247e-05,
      "loss": 1.4346,
      "step": 882
    },
    {
      "epoch": 0.5768414176057488,
      "grad_norm": Infinity,
      "learning_rate": 8.0386883271162e-05,
      "loss": 1.5796,
      "step": 883
    },
    {
      "epoch": 0.5774946921443737,
      "grad_norm": Infinity,
      "learning_rate": 8.017878061418582e-05,
      "loss": 1.6774,
      "step": 884
    },
    {
      "epoch": 0.5781479666829985,
      "grad_norm": Infinity,
      "learning_rate": 7.99707672685321e-05,
      "loss": 1.6591,
      "step": 885
    },
    {
      "epoch": 0.5788012412216234,
      "grad_norm": Infinity,
      "learning_rate": 7.976284417147644e-05,
      "loss": 1.7876,
      "step": 886
    },
    {
      "epoch": 0.5794545157602482,
      "grad_norm": Infinity,
      "learning_rate": 7.955501225988792e-05,
      "loss": 1.7762,
      "step": 887
    },
    {
      "epoch": 0.580107790298873,
      "grad_norm": Infinity,
      "learning_rate": 7.934727247022475e-05,
      "loss": 1.8609,
      "step": 888
    },
    {
      "epoch": 0.580761064837498,
      "grad_norm": Infinity,
      "learning_rate": 7.913962573852996e-05,
      "loss": 1.9347,
      "step": 889
    },
    {
      "epoch": 0.5814143393761229,
      "grad_norm": Infinity,
      "learning_rate": 7.89320730004274e-05,
      "loss": 2.2418,
      "step": 890
    },
    {
      "epoch": 0.5820676139147477,
      "grad_norm": Infinity,
      "learning_rate": 7.87246151911172e-05,
      "loss": 2.1152,
      "step": 891
    },
    {
      "epoch": 0.5827208884533726,
      "grad_norm": Infinity,
      "learning_rate": 7.851725324537204e-05,
      "loss": 2.2936,
      "step": 892
    },
    {
      "epoch": 0.5833741629919974,
      "grad_norm": Infinity,
      "learning_rate": 7.830998809753237e-05,
      "loss": 2.5237,
      "step": 893
    },
    {
      "epoch": 0.5840274375306223,
      "grad_norm": Infinity,
      "learning_rate": 7.81028206815027e-05,
      "loss": 2.33,
      "step": 894
    },
    {
      "epoch": 0.5846807120692471,
      "grad_norm": Infinity,
      "learning_rate": 7.789575193074704e-05,
      "loss": 2.8523,
      "step": 895
    },
    {
      "epoch": 0.585333986607872,
      "grad_norm": Infinity,
      "learning_rate": 7.768878277828486e-05,
      "loss": 2.9459,
      "step": 896
    },
    {
      "epoch": 0.5859872611464968,
      "grad_norm": Infinity,
      "learning_rate": 7.748191415668698e-05,
      "loss": 3.0787,
      "step": 897
    },
    {
      "epoch": 0.5866405356851216,
      "grad_norm": Infinity,
      "learning_rate": 7.727514699807102e-05,
      "loss": 2.9497,
      "step": 898
    },
    {
      "epoch": 0.5872938102237465,
      "grad_norm": Infinity,
      "learning_rate": 7.706848223409759e-05,
      "loss": 3.8379,
      "step": 899
    },
    {
      "epoch": 0.5879470847623713,
      "grad_norm": Infinity,
      "learning_rate": 7.686192079596586e-05,
      "loss": 4.5886,
      "step": 900
    },
    {
      "epoch": 0.5886003593009962,
      "grad_norm": Infinity,
      "learning_rate": 7.66554636144095e-05,
      "loss": 1.2221,
      "step": 901
    },
    {
      "epoch": 0.5892536338396211,
      "grad_norm": Infinity,
      "learning_rate": 7.644911161969226e-05,
      "loss": 1.3855,
      "step": 902
    },
    {
      "epoch": 0.589906908378246,
      "grad_norm": Infinity,
      "learning_rate": 7.624286574160409e-05,
      "loss": 1.3057,
      "step": 903
    },
    {
      "epoch": 0.5905601829168708,
      "grad_norm": Infinity,
      "learning_rate": 7.603672690945682e-05,
      "loss": 1.4719,
      "step": 904
    },
    {
      "epoch": 0.5912134574554957,
      "grad_norm": Infinity,
      "learning_rate": 7.583069605207975e-05,
      "loss": 1.3908,
      "step": 905
    },
    {
      "epoch": 0.5918667319941205,
      "grad_norm": Infinity,
      "learning_rate": 7.562477409781586e-05,
      "loss": 1.5683,
      "step": 906
    },
    {
      "epoch": 0.5925200065327454,
      "grad_norm": Infinity,
      "learning_rate": 7.541896197451727e-05,
      "loss": 1.4981,
      "step": 907
    },
    {
      "epoch": 0.5931732810713702,
      "grad_norm": Infinity,
      "learning_rate": 7.521326060954143e-05,
      "loss": 1.5204,
      "step": 908
    },
    {
      "epoch": 0.5938265556099951,
      "grad_norm": Infinity,
      "learning_rate": 7.500767092974647e-05,
      "loss": 1.6046,
      "step": 909
    },
    {
      "epoch": 0.5944798301486199,
      "grad_norm": Infinity,
      "learning_rate": 7.48021938614875e-05,
      "loss": 1.4955,
      "step": 910
    },
    {
      "epoch": 0.5951331046872448,
      "grad_norm": Infinity,
      "learning_rate": 7.459683033061208e-05,
      "loss": 1.6148,
      "step": 911
    },
    {
      "epoch": 0.5957863792258696,
      "grad_norm": Infinity,
      "learning_rate": 7.439158126245627e-05,
      "loss": 1.6611,
      "step": 912
    },
    {
      "epoch": 0.5964396537644945,
      "grad_norm": Infinity,
      "learning_rate": 7.418644758184038e-05,
      "loss": 1.9497,
      "step": 913
    },
    {
      "epoch": 0.5970929283031194,
      "grad_norm": Infinity,
      "learning_rate": 7.398143021306471e-05,
      "loss": 1.964,
      "step": 914
    },
    {
      "epoch": 0.5977462028417443,
      "grad_norm": Infinity,
      "learning_rate": 7.37765300799056e-05,
      "loss": 2.007,
      "step": 915
    },
    {
      "epoch": 0.5983994773803691,
      "grad_norm": Infinity,
      "learning_rate": 7.357174810561103e-05,
      "loss": 2.3996,
      "step": 916
    },
    {
      "epoch": 0.599052751918994,
      "grad_norm": Infinity,
      "learning_rate": 7.336708521289669e-05,
      "loss": 2.3556,
      "step": 917
    },
    {
      "epoch": 0.5997060264576188,
      "grad_norm": Infinity,
      "learning_rate": 7.316254232394158e-05,
      "loss": 2.3301,
      "step": 918
    },
    {
      "epoch": 0.6003593009962437,
      "grad_norm": Infinity,
      "learning_rate": 7.295812036038407e-05,
      "loss": 2.5678,
      "step": 919
    },
    {
      "epoch": 0.6010125755348685,
      "grad_norm": Infinity,
      "learning_rate": 7.275382024331772e-05,
      "loss": 2.7143,
      "step": 920
    },
    {
      "epoch": 0.6016658500734934,
      "grad_norm": Infinity,
      "learning_rate": 7.254964289328688e-05,
      "loss": 2.7722,
      "step": 921
    },
    {
      "epoch": 0.6023191246121182,
      "grad_norm": Infinity,
      "learning_rate": 7.234558923028289e-05,
      "loss": 3.4389,
      "step": 922
    },
    {
      "epoch": 0.6029723991507431,
      "grad_norm": Infinity,
      "learning_rate": 7.21416601737397e-05,
      "loss": 3.2806,
      "step": 923
    },
    {
      "epoch": 0.6036256736893679,
      "grad_norm": Infinity,
      "learning_rate": 7.193785664252987e-05,
      "loss": 3.4732,
      "step": 924
    },
    {
      "epoch": 0.6042789482279928,
      "grad_norm": Infinity,
      "learning_rate": 7.173417955496024e-05,
      "loss": 4.3247,
      "step": 925
    },
    {
      "epoch": 0.6049322227666176,
      "grad_norm": Infinity,
      "learning_rate": 7.153062982876805e-05,
      "loss": 1.3828,
      "step": 926
    },
    {
      "epoch": 0.6055854973052426,
      "grad_norm": Infinity,
      "learning_rate": 7.132720838111659e-05,
      "loss": 1.3161,
      "step": 927
    },
    {
      "epoch": 0.6062387718438674,
      "grad_norm": Infinity,
      "learning_rate": 7.112391612859118e-05,
      "loss": 1.4399,
      "step": 928
    },
    {
      "epoch": 0.6068920463824923,
      "grad_norm": Infinity,
      "learning_rate": 7.092075398719502e-05,
      "loss": 1.3693,
      "step": 929
    },
    {
      "epoch": 0.6075453209211171,
      "grad_norm": Infinity,
      "learning_rate": 7.071772287234497e-05,
      "loss": 1.453,
      "step": 930
    },
    {
      "epoch": 0.608198595459742,
      "grad_norm": Infinity,
      "learning_rate": 7.051482369886761e-05,
      "loss": 1.4425,
      "step": 931
    },
    {
      "epoch": 0.6088518699983668,
      "grad_norm": Infinity,
      "learning_rate": 7.031205738099491e-05,
      "loss": 1.5857,
      "step": 932
    },
    {
      "epoch": 0.6095051445369917,
      "grad_norm": Infinity,
      "learning_rate": 7.010942483236036e-05,
      "loss": 1.6388,
      "step": 933
    },
    {
      "epoch": 0.6101584190756165,
      "grad_norm": Infinity,
      "learning_rate": 6.990692696599447e-05,
      "loss": 1.6966,
      "step": 934
    },
    {
      "epoch": 0.6108116936142414,
      "grad_norm": Infinity,
      "learning_rate": 6.970456469432117e-05,
      "loss": 1.7996,
      "step": 935
    },
    {
      "epoch": 0.6114649681528662,
      "grad_norm": Infinity,
      "learning_rate": 6.950233892915323e-05,
      "loss": 1.6195,
      "step": 936
    },
    {
      "epoch": 0.6121182426914911,
      "grad_norm": Infinity,
      "learning_rate": 6.93002505816884e-05,
      "loss": 1.6846,
      "step": 937
    },
    {
      "epoch": 0.6127715172301159,
      "grad_norm": Infinity,
      "learning_rate": 6.909830056250527e-05,
      "loss": 1.9402,
      "step": 938
    },
    {
      "epoch": 0.6134247917687408,
      "grad_norm": Infinity,
      "learning_rate": 6.889648978155909e-05,
      "loss": 1.8907,
      "step": 939
    },
    {
      "epoch": 0.6140780663073657,
      "grad_norm": Infinity,
      "learning_rate": 6.869481914817779e-05,
      "loss": 1.855,
      "step": 940
    },
    {
      "epoch": 0.6147313408459906,
      "grad_norm": Infinity,
      "learning_rate": 6.849328957105773e-05,
      "loss": 2.1814,
      "step": 941
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": Infinity,
      "learning_rate": 6.829190195825983e-05,
      "loss": 2.4298,
      "step": 942
    },
    {
      "epoch": 0.6160378899232403,
      "grad_norm": Infinity,
      "learning_rate": 6.809065721720516e-05,
      "loss": 2.5652,
      "step": 943
    },
    {
      "epoch": 0.6166911644618651,
      "grad_norm": Infinity,
      "learning_rate": 6.788955625467119e-05,
      "loss": 2.3691,
      "step": 944
    },
    {
      "epoch": 0.61734443900049,
      "grad_norm": Infinity,
      "learning_rate": 6.768859997678751e-05,
      "loss": 2.7052,
      "step": 945
    },
    {
      "epoch": 0.6179977135391148,
      "grad_norm": Infinity,
      "learning_rate": 6.748778928903166e-05,
      "loss": 3.2163,
      "step": 946
    },
    {
      "epoch": 0.6186509880777397,
      "grad_norm": Infinity,
      "learning_rate": 6.728712509622539e-05,
      "loss": 2.8301,
      "step": 947
    },
    {
      "epoch": 0.6193042626163645,
      "grad_norm": Infinity,
      "learning_rate": 6.708660830253015e-05,
      "loss": 3.2148,
      "step": 948
    },
    {
      "epoch": 0.6199575371549894,
      "grad_norm": Infinity,
      "learning_rate": 6.688623981144339e-05,
      "loss": 3.767,
      "step": 949
    },
    {
      "epoch": 0.6206108116936142,
      "grad_norm": Infinity,
      "learning_rate": 6.668602052579424e-05,
      "loss": 4.4411,
      "step": 950
    },
    {
      "epoch": 0.621264086232239,
      "grad_norm": Infinity,
      "learning_rate": 6.648595134773958e-05,
      "loss": 1.2227,
      "step": 951
    },
    {
      "epoch": 0.6219173607708639,
      "grad_norm": Infinity,
      "learning_rate": 6.628603317875996e-05,
      "loss": 1.466,
      "step": 952
    },
    {
      "epoch": 0.6225706353094889,
      "grad_norm": Infinity,
      "learning_rate": 6.608626691965541e-05,
      "loss": 1.3718,
      "step": 953
    },
    {
      "epoch": 0.6232239098481137,
      "grad_norm": Infinity,
      "learning_rate": 6.588665347054153e-05,
      "loss": 1.4628,
      "step": 954
    },
    {
      "epoch": 0.6238771843867386,
      "grad_norm": Infinity,
      "learning_rate": 6.568719373084538e-05,
      "loss": 1.4955,
      "step": 955
    },
    {
      "epoch": 0.6245304589253634,
      "grad_norm": Infinity,
      "learning_rate": 6.548788859930146e-05,
      "loss": 1.5999,
      "step": 956
    },
    {
      "epoch": 0.6251837334639883,
      "grad_norm": Infinity,
      "learning_rate": 6.528873897394757e-05,
      "loss": 1.5084,
      "step": 957
    },
    {
      "epoch": 0.6258370080026131,
      "grad_norm": Infinity,
      "learning_rate": 6.508974575212088e-05,
      "loss": 1.4152,
      "step": 958
    },
    {
      "epoch": 0.626490282541238,
      "grad_norm": Infinity,
      "learning_rate": 6.489090983045379e-05,
      "loss": 1.6017,
      "step": 959
    },
    {
      "epoch": 0.6271435570798628,
      "grad_norm": Infinity,
      "learning_rate": 6.469223210486992e-05,
      "loss": 1.642,
      "step": 960
    },
    {
      "epoch": 0.6277968316184876,
      "grad_norm": Infinity,
      "learning_rate": 6.449371347058019e-05,
      "loss": 1.6396,
      "step": 961
    },
    {
      "epoch": 0.6284501061571125,
      "grad_norm": Infinity,
      "learning_rate": 6.429535482207846e-05,
      "loss": 1.8072,
      "step": 962
    },
    {
      "epoch": 0.6291033806957373,
      "grad_norm": Infinity,
      "learning_rate": 6.409715705313804e-05,
      "loss": 1.7597,
      "step": 963
    },
    {
      "epoch": 0.6297566552343622,
      "grad_norm": Infinity,
      "learning_rate": 6.389912105680697e-05,
      "loss": 1.8747,
      "step": 964
    },
    {
      "epoch": 0.630409929772987,
      "grad_norm": Infinity,
      "learning_rate": 6.370124772540469e-05,
      "loss": 2.0892,
      "step": 965
    },
    {
      "epoch": 0.631063204311612,
      "grad_norm": Infinity,
      "learning_rate": 6.350353795051748e-05,
      "loss": 2.1664,
      "step": 966
    },
    {
      "epoch": 0.6317164788502369,
      "grad_norm": Infinity,
      "learning_rate": 6.330599262299479e-05,
      "loss": 2.094,
      "step": 967
    },
    {
      "epoch": 0.6323697533888617,
      "grad_norm": Infinity,
      "learning_rate": 6.310861263294507e-05,
      "loss": 2.3747,
      "step": 968
    },
    {
      "epoch": 0.6330230279274865,
      "grad_norm": Infinity,
      "learning_rate": 6.291139886973169e-05,
      "loss": 2.2346,
      "step": 969
    },
    {
      "epoch": 0.6336763024661114,
      "grad_norm": Infinity,
      "learning_rate": 6.271435222196916e-05,
      "loss": 2.6554,
      "step": 970
    },
    {
      "epoch": 0.6343295770047362,
      "grad_norm": Infinity,
      "learning_rate": 6.251747357751889e-05,
      "loss": 2.7962,
      "step": 971
    },
    {
      "epoch": 0.6349828515433611,
      "grad_norm": Infinity,
      "learning_rate": 6.232076382348539e-05,
      "loss": 3.1112,
      "step": 972
    },
    {
      "epoch": 0.6356361260819859,
      "grad_norm": Infinity,
      "learning_rate": 6.212422384621208e-05,
      "loss": 3.0868,
      "step": 973
    },
    {
      "epoch": 0.6362894006206108,
      "grad_norm": Infinity,
      "learning_rate": 6.192785453127748e-05,
      "loss": 3.2682,
      "step": 974
    },
    {
      "epoch": 0.6369426751592356,
      "grad_norm": Infinity,
      "learning_rate": 6.173165676349103e-05,
      "loss": 4.1133,
      "step": 975
    },
    {
      "epoch": 0.6375959496978605,
      "grad_norm": Infinity,
      "learning_rate": 6.15356314268893e-05,
      "loss": 1.2235,
      "step": 976
    },
    {
      "epoch": 0.6382492242364853,
      "grad_norm": Infinity,
      "learning_rate": 6.13397794047319e-05,
      "loss": 1.3234,
      "step": 977
    },
    {
      "epoch": 0.6389024987751103,
      "grad_norm": Infinity,
      "learning_rate": 6.114410157949745e-05,
      "loss": 1.4325,
      "step": 978
    },
    {
      "epoch": 0.6395557733137351,
      "grad_norm": Infinity,
      "learning_rate": 6.094859883287977e-05,
      "loss": 1.3914,
      "step": 979
    },
    {
      "epoch": 0.64020904785236,
      "grad_norm": Infinity,
      "learning_rate": 6.0753272045783625e-05,
      "loss": 1.4688,
      "step": 980
    },
    {
      "epoch": 0.6408623223909848,
      "grad_norm": Infinity,
      "learning_rate": 6.0558122098321144e-05,
      "loss": 1.4596,
      "step": 981
    },
    {
      "epoch": 0.6415155969296097,
      "grad_norm": Infinity,
      "learning_rate": 6.036314986980749e-05,
      "loss": 1.5476,
      "step": 982
    },
    {
      "epoch": 0.6421688714682345,
      "grad_norm": Infinity,
      "learning_rate": 6.016835623875716e-05,
      "loss": 1.5854,
      "step": 983
    },
    {
      "epoch": 0.6428221460068594,
      "grad_norm": Infinity,
      "learning_rate": 5.9973742082879815e-05,
      "loss": 1.7013,
      "step": 984
    },
    {
      "epoch": 0.6434754205454842,
      "grad_norm": Infinity,
      "learning_rate": 5.977930827907649e-05,
      "loss": 1.7004,
      "step": 985
    },
    {
      "epoch": 0.6441286950841091,
      "grad_norm": Infinity,
      "learning_rate": 5.958505570343561e-05,
      "loss": 1.7823,
      "step": 986
    },
    {
      "epoch": 0.6447819696227339,
      "grad_norm": Infinity,
      "learning_rate": 5.9390985231228936e-05,
      "loss": 1.7044,
      "step": 987
    },
    {
      "epoch": 0.6454352441613588,
      "grad_norm": Infinity,
      "learning_rate": 5.91970977369078e-05,
      "loss": 1.9262,
      "step": 988
    },
    {
      "epoch": 0.6460885186999836,
      "grad_norm": Infinity,
      "learning_rate": 5.90033940940989e-05,
      "loss": 1.9572,
      "step": 989
    },
    {
      "epoch": 0.6467417932386085,
      "grad_norm": Infinity,
      "learning_rate": 5.880987517560075e-05,
      "loss": 1.9857,
      "step": 990
    },
    {
      "epoch": 0.6473950677772334,
      "grad_norm": Infinity,
      "learning_rate": 5.861654185337933e-05,
      "loss": 2.3491,
      "step": 991
    },
    {
      "epoch": 0.6480483423158583,
      "grad_norm": Infinity,
      "learning_rate": 5.842339499856444e-05,
      "loss": 2.3709,
      "step": 992
    },
    {
      "epoch": 0.6487016168544831,
      "grad_norm": Infinity,
      "learning_rate": 5.8230435481445734e-05,
      "loss": 2.3352,
      "step": 993
    },
    {
      "epoch": 0.649354891393108,
      "grad_norm": Infinity,
      "learning_rate": 5.8037664171468596e-05,
      "loss": 2.7182,
      "step": 994
    },
    {
      "epoch": 0.6500081659317328,
      "grad_norm": Infinity,
      "learning_rate": 5.784508193723057e-05,
      "loss": 2.823,
      "step": 995
    },
    {
      "epoch": 0.6506614404703577,
      "grad_norm": Infinity,
      "learning_rate": 5.765268964647712e-05,
      "loss": 2.8539,
      "step": 996
    },
    {
      "epoch": 0.6513147150089825,
      "grad_norm": Infinity,
      "learning_rate": 5.746048816609788e-05,
      "loss": 2.871,
      "step": 997
    },
    {
      "epoch": 0.6519679895476074,
      "grad_norm": Infinity,
      "learning_rate": 5.726847836212277e-05,
      "loss": 3.035,
      "step": 998
    },
    {
      "epoch": 0.6526212640862322,
      "grad_norm": Infinity,
      "learning_rate": 5.7076661099717986e-05,
      "loss": 4.0633,
      "step": 999
    },
    {
      "epoch": 0.6532745386248571,
      "grad_norm": Infinity,
      "learning_rate": 5.688503724318217e-05,
      "loss": 3.8549,
      "step": 1000
    },
    {
      "epoch": 0.6539278131634819,
      "grad_norm": Infinity,
      "learning_rate": 5.669360765594258e-05,
      "loss": 1.2725,
      "step": 1001
    },
    {
      "epoch": 0.6545810877021068,
      "grad_norm": Infinity,
      "learning_rate": 5.6502373200551065e-05,
      "loss": 1.3742,
      "step": 1002
    },
    {
      "epoch": 0.6552343622407316,
      "grad_norm": Infinity,
      "learning_rate": 5.631133473868018e-05,
      "loss": 1.3706,
      "step": 1003
    },
    {
      "epoch": 0.6558876367793566,
      "grad_norm": Infinity,
      "learning_rate": 5.6120493131119555e-05,
      "loss": 1.4821,
      "step": 1004
    },
    {
      "epoch": 0.6565409113179814,
      "grad_norm": Infinity,
      "learning_rate": 5.5929849237771556e-05,
      "loss": 1.4066,
      "step": 1005
    },
    {
      "epoch": 0.6571941858566063,
      "grad_norm": Infinity,
      "learning_rate": 5.573940391764796e-05,
      "loss": 1.4638,
      "step": 1006
    },
    {
      "epoch": 0.6578474603952311,
      "grad_norm": Infinity,
      "learning_rate": 5.554915802886558e-05,
      "loss": 1.4518,
      "step": 1007
    },
    {
      "epoch": 0.658500734933856,
      "grad_norm": Infinity,
      "learning_rate": 5.535911242864271e-05,
      "loss": 1.515,
      "step": 1008
    },
    {
      "epoch": 0.6591540094724808,
      "grad_norm": Infinity,
      "learning_rate": 5.5169267973295294e-05,
      "loss": 1.5506,
      "step": 1009
    },
    {
      "epoch": 0.6598072840111057,
      "grad_norm": Infinity,
      "learning_rate": 5.497962551823266e-05,
      "loss": 1.6251,
      "step": 1010
    },
    {
      "epoch": 0.6604605585497305,
      "grad_norm": Infinity,
      "learning_rate": 5.479018591795426e-05,
      "loss": 1.5996,
      "step": 1011
    },
    {
      "epoch": 0.6611138330883554,
      "grad_norm": Infinity,
      "learning_rate": 5.4600950026045326e-05,
      "loss": 1.6873,
      "step": 1012
    },
    {
      "epoch": 0.6617671076269802,
      "grad_norm": Infinity,
      "learning_rate": 5.441191869517328e-05,
      "loss": 1.7156,
      "step": 1013
    },
    {
      "epoch": 0.6624203821656051,
      "grad_norm": Infinity,
      "learning_rate": 5.422309277708379e-05,
      "loss": 1.8523,
      "step": 1014
    },
    {
      "epoch": 0.6630736567042299,
      "grad_norm": Infinity,
      "learning_rate": 5.403447312259702e-05,
      "loss": 2.0574,
      "step": 1015
    },
    {
      "epoch": 0.6637269312428548,
      "grad_norm": Infinity,
      "learning_rate": 5.384606058160363e-05,
      "loss": 1.8924,
      "step": 1016
    },
    {
      "epoch": 0.6643802057814797,
      "grad_norm": Infinity,
      "learning_rate": 5.365785600306124e-05,
      "loss": 2.3307,
      "step": 1017
    },
    {
      "epoch": 0.6650334803201046,
      "grad_norm": Infinity,
      "learning_rate": 5.346986023499027e-05,
      "loss": 2.5785,
      "step": 1018
    },
    {
      "epoch": 0.6656867548587294,
      "grad_norm": Infinity,
      "learning_rate": 5.3282074124470284e-05,
      "loss": 2.4408,
      "step": 1019
    },
    {
      "epoch": 0.6663400293973543,
      "grad_norm": Infinity,
      "learning_rate": 5.309449851763633e-05,
      "loss": 2.4936,
      "step": 1020
    },
    {
      "epoch": 0.6669933039359791,
      "grad_norm": Infinity,
      "learning_rate": 5.290713425967466e-05,
      "loss": 2.7817,
      "step": 1021
    },
    {
      "epoch": 0.667646578474604,
      "grad_norm": Infinity,
      "learning_rate": 5.271998219481953e-05,
      "loss": 3.0024,
      "step": 1022
    },
    {
      "epoch": 0.6682998530132288,
      "grad_norm": Infinity,
      "learning_rate": 5.2533043166348886e-05,
      "loss": 2.867,
      "step": 1023
    },
    {
      "epoch": 0.6689531275518537,
      "grad_norm": Infinity,
      "learning_rate": 5.234631801658081e-05,
      "loss": 3.1437,
      "step": 1024
    },
    {
      "epoch": 0.6696064020904785,
      "grad_norm": Infinity,
      "learning_rate": 5.215980758686978e-05,
      "loss": 3.8861,
      "step": 1025
    },
    {
      "epoch": 0.6702596766291034,
      "grad_norm": Infinity,
      "learning_rate": 5.197351271760258e-05,
      "loss": 1.2455,
      "step": 1026
    },
    {
      "epoch": 0.6709129511677282,
      "grad_norm": Infinity,
      "learning_rate": 5.178743424819492e-05,
      "loss": 1.4044,
      "step": 1027
    },
    {
      "epoch": 0.671566225706353,
      "grad_norm": Infinity,
      "learning_rate": 5.160157301708732e-05,
      "loss": 1.4414,
      "step": 1028
    },
    {
      "epoch": 0.6722195002449779,
      "grad_norm": Infinity,
      "learning_rate": 5.141592986174151e-05,
      "loss": 1.4843,
      "step": 1029
    },
    {
      "epoch": 0.6728727747836029,
      "grad_norm": Infinity,
      "learning_rate": 5.123050561863657e-05,
      "loss": 1.473,
      "step": 1030
    },
    {
      "epoch": 0.6735260493222277,
      "grad_norm": Infinity,
      "learning_rate": 5.104530112326522e-05,
      "loss": 1.5881,
      "step": 1031
    },
    {
      "epoch": 0.6741793238608526,
      "grad_norm": Infinity,
      "learning_rate": 5.086031721012998e-05,
      "loss": 1.5053,
      "step": 1032
    },
    {
      "epoch": 0.6748325983994774,
      "grad_norm": Infinity,
      "learning_rate": 5.067555471273957e-05,
      "loss": 1.5942,
      "step": 1033
    },
    {
      "epoch": 0.6754858729381022,
      "grad_norm": Infinity,
      "learning_rate": 5.049101446360498e-05,
      "loss": 1.5184,
      "step": 1034
    },
    {
      "epoch": 0.6761391474767271,
      "grad_norm": Infinity,
      "learning_rate": 5.0306697294235714e-05,
      "loss": 1.6699,
      "step": 1035
    },
    {
      "epoch": 0.676792422015352,
      "grad_norm": Infinity,
      "learning_rate": 5.01226040351363e-05,
      "loss": 1.7125,
      "step": 1036
    },
    {
      "epoch": 0.6774456965539768,
      "grad_norm": Infinity,
      "learning_rate": 4.99387355158021e-05,
      "loss": 1.8265,
      "step": 1037
    },
    {
      "epoch": 0.6780989710926016,
      "grad_norm": Infinity,
      "learning_rate": 4.97550925647161e-05,
      "loss": 1.8673,
      "step": 1038
    },
    {
      "epoch": 0.6787522456312265,
      "grad_norm": Infinity,
      "learning_rate": 4.957167600934474e-05,
      "loss": 1.935,
      "step": 1039
    },
    {
      "epoch": 0.6794055201698513,
      "grad_norm": Infinity,
      "learning_rate": 4.938848667613436e-05,
      "loss": 2.0304,
      "step": 1040
    },
    {
      "epoch": 0.6800587947084762,
      "grad_norm": Infinity,
      "learning_rate": 4.9205525390507644e-05,
      "loss": 1.9897,
      "step": 1041
    },
    {
      "epoch": 0.6807120692471011,
      "grad_norm": Infinity,
      "learning_rate": 4.9022792976859455e-05,
      "loss": 2.2204,
      "step": 1042
    },
    {
      "epoch": 0.681365343785726,
      "grad_norm": Infinity,
      "learning_rate": 4.884029025855364e-05,
      "loss": 2.268,
      "step": 1043
    },
    {
      "epoch": 0.6820186183243508,
      "grad_norm": Infinity,
      "learning_rate": 4.8658018057918964e-05,
      "loss": 2.5273,
      "step": 1044
    },
    {
      "epoch": 0.6826718928629757,
      "grad_norm": Infinity,
      "learning_rate": 4.8475977196245504e-05,
      "loss": 2.3542,
      "step": 1045
    },
    {
      "epoch": 0.6833251674016005,
      "grad_norm": Infinity,
      "learning_rate": 4.8294168493780955e-05,
      "loss": 2.7903,
      "step": 1046
    },
    {
      "epoch": 0.6839784419402254,
      "grad_norm": Infinity,
      "learning_rate": 4.8112592769727085e-05,
      "loss": 2.9665,
      "step": 1047
    },
    {
      "epoch": 0.6846317164788502,
      "grad_norm": Infinity,
      "learning_rate": 4.793125084223562e-05,
      "loss": 3.3895,
      "step": 1048
    },
    {
      "epoch": 0.6852849910174751,
      "grad_norm": Infinity,
      "learning_rate": 4.7750143528405126e-05,
      "loss": 3.2519,
      "step": 1049
    },
    {
      "epoch": 0.6859382655560999,
      "grad_norm": Infinity,
      "learning_rate": 4.756927164427685e-05,
      "loss": 4.7098,
      "step": 1050
    },
    {
      "epoch": 0.6865915400947248,
      "grad_norm": Infinity,
      "learning_rate": 4.738863600483125e-05,
      "loss": 1.2615,
      "step": 1051
    },
    {
      "epoch": 0.6872448146333496,
      "grad_norm": Infinity,
      "learning_rate": 4.720823742398447e-05,
      "loss": 1.3003,
      "step": 1052
    },
    {
      "epoch": 0.6878980891719745,
      "grad_norm": Infinity,
      "learning_rate": 4.70280767145842e-05,
      "loss": 1.46,
      "step": 1053
    },
    {
      "epoch": 0.6885513637105993,
      "grad_norm": Infinity,
      "learning_rate": 4.684815468840662e-05,
      "loss": 1.4526,
      "step": 1054
    },
    {
      "epoch": 0.6892046382492243,
      "grad_norm": Infinity,
      "learning_rate": 4.666847215615226e-05,
      "loss": 1.4448,
      "step": 1055
    },
    {
      "epoch": 0.6898579127878491,
      "grad_norm": Infinity,
      "learning_rate": 4.648902992744255e-05,
      "loss": 1.4982,
      "step": 1056
    },
    {
      "epoch": 0.690511187326474,
      "grad_norm": Infinity,
      "learning_rate": 4.630982881081629e-05,
      "loss": 1.5638,
      "step": 1057
    },
    {
      "epoch": 0.6911644618650988,
      "grad_norm": Infinity,
      "learning_rate": 4.6130869613725614e-05,
      "loss": 1.5881,
      "step": 1058
    },
    {
      "epoch": 0.6918177364037237,
      "grad_norm": Infinity,
      "learning_rate": 4.595215314253285e-05,
      "loss": 1.6855,
      "step": 1059
    },
    {
      "epoch": 0.6924710109423485,
      "grad_norm": Infinity,
      "learning_rate": 4.57736802025065e-05,
      "loss": 1.7135,
      "step": 1060
    },
    {
      "epoch": 0.6931242854809734,
      "grad_norm": Infinity,
      "learning_rate": 4.5595451597817795e-05,
      "loss": 1.6494,
      "step": 1061
    },
    {
      "epoch": 0.6937775600195982,
      "grad_norm": Infinity,
      "learning_rate": 4.5417468131536975e-05,
      "loss": 1.7395,
      "step": 1062
    },
    {
      "epoch": 0.6944308345582231,
      "grad_norm": Infinity,
      "learning_rate": 4.52397306056299e-05,
      "loss": 1.8408,
      "step": 1063
    },
    {
      "epoch": 0.6950841090968479,
      "grad_norm": Infinity,
      "learning_rate": 4.5062239820953986e-05,
      "loss": 1.9947,
      "step": 1064
    },
    {
      "epoch": 0.6957373836354728,
      "grad_norm": Infinity,
      "learning_rate": 4.488499657725511e-05,
      "loss": 2.0566,
      "step": 1065
    },
    {
      "epoch": 0.6963906581740976,
      "grad_norm": Infinity,
      "learning_rate": 4.470800167316367e-05,
      "loss": 2.0069,
      "step": 1066
    },
    {
      "epoch": 0.6970439327127225,
      "grad_norm": Infinity,
      "learning_rate": 4.453125590619104e-05,
      "loss": 2.2135,
      "step": 1067
    },
    {
      "epoch": 0.6976972072513474,
      "grad_norm": Infinity,
      "learning_rate": 4.4354760072726176e-05,
      "loss": 2.2858,
      "step": 1068
    },
    {
      "epoch": 0.6983504817899723,
      "grad_norm": Infinity,
      "learning_rate": 4.417851496803164e-05,
      "loss": 2.7049,
      "step": 1069
    },
    {
      "epoch": 0.6990037563285971,
      "grad_norm": Infinity,
      "learning_rate": 4.4002521386240466e-05,
      "loss": 2.6026,
      "step": 1070
    },
    {
      "epoch": 0.699657030867222,
      "grad_norm": Infinity,
      "learning_rate": 4.382678012035227e-05,
      "loss": 2.8106,
      "step": 1071
    },
    {
      "epoch": 0.7003103054058468,
      "grad_norm": Infinity,
      "learning_rate": 4.365129196222972e-05,
      "loss": 2.8605,
      "step": 1072
    },
    {
      "epoch": 0.7009635799444717,
      "grad_norm": Infinity,
      "learning_rate": 4.347605770259522e-05,
      "loss": 3.2817,
      "step": 1073
    },
    {
      "epoch": 0.7016168544830965,
      "grad_norm": Infinity,
      "learning_rate": 4.3301078131026826e-05,
      "loss": 3.3869,
      "step": 1074
    },
    {
      "epoch": 0.7022701290217214,
      "grad_norm": Infinity,
      "learning_rate": 4.312635403595532e-05,
      "loss": 4.4504,
      "step": 1075
    },
    {
      "epoch": 0.7029234035603462,
      "grad_norm": Infinity,
      "learning_rate": 4.2951886204660175e-05,
      "loss": 1.2931,
      "step": 1076
    },
    {
      "epoch": 0.7035766780989711,
      "grad_norm": Infinity,
      "learning_rate": 4.2777675423266227e-05,
      "loss": 1.3579,
      "step": 1077
    },
    {
      "epoch": 0.7042299526375959,
      "grad_norm": Infinity,
      "learning_rate": 4.260372247674004e-05,
      "loss": 1.4018,
      "step": 1078
    },
    {
      "epoch": 0.7048832271762208,
      "grad_norm": Infinity,
      "learning_rate": 4.243002814888656e-05,
      "loss": 1.415,
      "step": 1079
    },
    {
      "epoch": 0.7055365017148456,
      "grad_norm": Infinity,
      "learning_rate": 4.2256593222345185e-05,
      "loss": 1.4629,
      "step": 1080
    },
    {
      "epoch": 0.7061897762534706,
      "grad_norm": Infinity,
      "learning_rate": 4.2083418478586755e-05,
      "loss": 1.478,
      "step": 1081
    },
    {
      "epoch": 0.7068430507920954,
      "grad_norm": Infinity,
      "learning_rate": 4.1910504697909614e-05,
      "loss": 1.5336,
      "step": 1082
    },
    {
      "epoch": 0.7074963253307203,
      "grad_norm": Infinity,
      "learning_rate": 4.173785265943624e-05,
      "loss": 1.5161,
      "step": 1083
    },
    {
      "epoch": 0.7081495998693451,
      "grad_norm": Infinity,
      "learning_rate": 4.1565463141109894e-05,
      "loss": 1.5435,
      "step": 1084
    },
    {
      "epoch": 0.70880287440797,
      "grad_norm": Infinity,
      "learning_rate": 4.139333691969071e-05,
      "loss": 1.7314,
      "step": 1085
    },
    {
      "epoch": 0.7094561489465948,
      "grad_norm": Infinity,
      "learning_rate": 4.12214747707527e-05,
      "loss": 1.6485,
      "step": 1086
    },
    {
      "epoch": 0.7101094234852197,
      "grad_norm": Infinity,
      "learning_rate": 4.1049877468679856e-05,
      "loss": 1.8288,
      "step": 1087
    },
    {
      "epoch": 0.7107626980238445,
      "grad_norm": Infinity,
      "learning_rate": 4.087854578666282e-05,
      "loss": 1.945,
      "step": 1088
    },
    {
      "epoch": 0.7114159725624694,
      "grad_norm": Infinity,
      "learning_rate": 4.0707480496695514e-05,
      "loss": 2.0181,
      "step": 1089
    },
    {
      "epoch": 0.7120692471010942,
      "grad_norm": Infinity,
      "learning_rate": 4.053668236957134e-05,
      "loss": 2.1645,
      "step": 1090
    },
    {
      "epoch": 0.712722521639719,
      "grad_norm": Infinity,
      "learning_rate": 4.036615217488009e-05,
      "loss": 2.2705,
      "step": 1091
    },
    {
      "epoch": 0.7133757961783439,
      "grad_norm": Infinity,
      "learning_rate": 4.01958906810042e-05,
      "loss": 2.2283,
      "step": 1092
    },
    {
      "epoch": 0.7140290707169689,
      "grad_norm": Infinity,
      "learning_rate": 4.0025898655115394e-05,
      "loss": 2.4277,
      "step": 1093
    },
    {
      "epoch": 0.7146823452555937,
      "grad_norm": Infinity,
      "learning_rate": 3.985617686317118e-05,
      "loss": 2.6549,
      "step": 1094
    },
    {
      "epoch": 0.7153356197942186,
      "grad_norm": Infinity,
      "learning_rate": 3.96867260699116e-05,
      "loss": 2.3398,
      "step": 1095
    },
    {
      "epoch": 0.7159888943328434,
      "grad_norm": Infinity,
      "learning_rate": 3.951754703885533e-05,
      "loss": 2.6731,
      "step": 1096
    },
    {
      "epoch": 0.7166421688714683,
      "grad_norm": Infinity,
      "learning_rate": 3.934864053229681e-05,
      "loss": 2.6785,
      "step": 1097
    },
    {
      "epoch": 0.7172954434100931,
      "grad_norm": Infinity,
      "learning_rate": 3.918000731130238e-05,
      "loss": 3.2614,
      "step": 1098
    },
    {
      "epoch": 0.717948717948718,
      "grad_norm": Infinity,
      "learning_rate": 3.9011648135706966e-05,
      "loss": 3.9284,
      "step": 1099
    },
    {
      "epoch": 0.7186019924873428,
      "grad_norm": Infinity,
      "learning_rate": 3.884356376411089e-05,
      "loss": 4.2049,
      "step": 1100
    },
    {
      "epoch": 0.7192552670259676,
      "grad_norm": Infinity,
      "learning_rate": 3.8675754953875956e-05,
      "loss": 1.3125,
      "step": 1101
    },
    {
      "epoch": 0.7199085415645925,
      "grad_norm": Infinity,
      "learning_rate": 3.8508222461122567e-05,
      "loss": 1.3905,
      "step": 1102
    },
    {
      "epoch": 0.7205618161032173,
      "grad_norm": Infinity,
      "learning_rate": 3.8340967040725995e-05,
      "loss": 1.4399,
      "step": 1103
    },
    {
      "epoch": 0.7212150906418422,
      "grad_norm": Infinity,
      "learning_rate": 3.817398944631302e-05,
      "loss": 1.4085,
      "step": 1104
    },
    {
      "epoch": 0.721868365180467,
      "grad_norm": Infinity,
      "learning_rate": 3.800729043025871e-05,
      "loss": 1.463,
      "step": 1105
    },
    {
      "epoch": 0.722521639719092,
      "grad_norm": Infinity,
      "learning_rate": 3.784087074368279e-05,
      "loss": 1.4755,
      "step": 1106
    },
    {
      "epoch": 0.7231749142577168,
      "grad_norm": Infinity,
      "learning_rate": 3.767473113644641e-05,
      "loss": 1.5932,
      "step": 1107
    },
    {
      "epoch": 0.7238281887963417,
      "grad_norm": Infinity,
      "learning_rate": 3.750887235714872e-05,
      "loss": 1.6179,
      "step": 1108
    },
    {
      "epoch": 0.7244814633349665,
      "grad_norm": Infinity,
      "learning_rate": 3.734329515312349e-05,
      "loss": 1.5271,
      "step": 1109
    },
    {
      "epoch": 0.7251347378735914,
      "grad_norm": Infinity,
      "learning_rate": 3.717800027043576e-05,
      "loss": 1.5595,
      "step": 1110
    },
    {
      "epoch": 0.7257880124122162,
      "grad_norm": Infinity,
      "learning_rate": 3.7012988453878586e-05,
      "loss": 1.6898,
      "step": 1111
    },
    {
      "epoch": 0.7264412869508411,
      "grad_norm": Infinity,
      "learning_rate": 3.6848260446969306e-05,
      "loss": 1.7962,
      "step": 1112
    },
    {
      "epoch": 0.7270945614894659,
      "grad_norm": Infinity,
      "learning_rate": 3.668381699194676e-05,
      "loss": 1.7829,
      "step": 1113
    },
    {
      "epoch": 0.7277478360280908,
      "grad_norm": Infinity,
      "learning_rate": 3.6519658829767455e-05,
      "loss": 1.874,
      "step": 1114
    },
    {
      "epoch": 0.7284011105667156,
      "grad_norm": Infinity,
      "learning_rate": 3.635578670010242e-05,
      "loss": 1.9904,
      "step": 1115
    },
    {
      "epoch": 0.7290543851053405,
      "grad_norm": Infinity,
      "learning_rate": 3.619220134133402e-05,
      "loss": 1.9767,
      "step": 1116
    },
    {
      "epoch": 0.7297076596439653,
      "grad_norm": Infinity,
      "learning_rate": 3.602890349055221e-05,
      "loss": 2.2286,
      "step": 1117
    },
    {
      "epoch": 0.7303609341825902,
      "grad_norm": Infinity,
      "learning_rate": 3.586589388355176e-05,
      "loss": 2.4116,
      "step": 1118
    },
    {
      "epoch": 0.7310142087212151,
      "grad_norm": Infinity,
      "learning_rate": 3.570317325482847e-05,
      "loss": 2.5585,
      "step": 1119
    },
    {
      "epoch": 0.73166748325984,
      "grad_norm": Infinity,
      "learning_rate": 3.554074233757608e-05,
      "loss": 2.3875,
      "step": 1120
    },
    {
      "epoch": 0.7323207577984648,
      "grad_norm": Infinity,
      "learning_rate": 3.537860186368305e-05,
      "loss": 2.9333,
      "step": 1121
    },
    {
      "epoch": 0.7329740323370897,
      "grad_norm": Infinity,
      "learning_rate": 3.5216752563729e-05,
      "loss": 3.1705,
      "step": 1122
    },
    {
      "epoch": 0.7336273068757145,
      "grad_norm": Infinity,
      "learning_rate": 3.5055195166981645e-05,
      "loss": 3.3577,
      "step": 1123
    },
    {
      "epoch": 0.7342805814143394,
      "grad_norm": Infinity,
      "learning_rate": 3.4893930401393406e-05,
      "loss": 3.4905,
      "step": 1124
    },
    {
      "epoch": 0.7349338559529642,
      "grad_norm": Infinity,
      "learning_rate": 3.4732958993598154e-05,
      "loss": 4.1618,
      "step": 1125
    },
    {
      "epoch": 0.7355871304915891,
      "grad_norm": Infinity,
      "learning_rate": 3.457228166890791e-05,
      "loss": 1.2767,
      "step": 1126
    },
    {
      "epoch": 0.7362404050302139,
      "grad_norm": Infinity,
      "learning_rate": 3.441189915130974e-05,
      "loss": 1.3949,
      "step": 1127
    },
    {
      "epoch": 0.7368936795688388,
      "grad_norm": Infinity,
      "learning_rate": 3.425181216346213e-05,
      "loss": 1.4078,
      "step": 1128
    },
    {
      "epoch": 0.7375469541074636,
      "grad_norm": Infinity,
      "learning_rate": 3.409202142669213e-05,
      "loss": 1.4977,
      "step": 1129
    },
    {
      "epoch": 0.7382002286460885,
      "grad_norm": Infinity,
      "learning_rate": 3.393252766099187e-05,
      "loss": 1.4244,
      "step": 1130
    },
    {
      "epoch": 0.7388535031847133,
      "grad_norm": Infinity,
      "learning_rate": 3.377333158501534e-05,
      "loss": 1.6322,
      "step": 1131
    },
    {
      "epoch": 0.7395067777233383,
      "grad_norm": Infinity,
      "learning_rate": 3.3614433916075326e-05,
      "loss": 1.5867,
      "step": 1132
    },
    {
      "epoch": 0.7401600522619631,
      "grad_norm": Infinity,
      "learning_rate": 3.34558353701398e-05,
      "loss": 1.539,
      "step": 1133
    },
    {
      "epoch": 0.740813326800588,
      "grad_norm": Infinity,
      "learning_rate": 3.3297536661829155e-05,
      "loss": 1.5412,
      "step": 1134
    },
    {
      "epoch": 0.7414666013392128,
      "grad_norm": Infinity,
      "learning_rate": 3.313953850441266e-05,
      "loss": 1.6732,
      "step": 1135
    },
    {
      "epoch": 0.7421198758778377,
      "grad_norm": Infinity,
      "learning_rate": 3.298184160980532e-05,
      "loss": 1.6536,
      "step": 1136
    },
    {
      "epoch": 0.7427731504164625,
      "grad_norm": Infinity,
      "learning_rate": 3.2824446688564815e-05,
      "loss": 1.7995,
      "step": 1137
    },
    {
      "epoch": 0.7434264249550874,
      "grad_norm": Infinity,
      "learning_rate": 3.266735444988808e-05,
      "loss": 1.7265,
      "step": 1138
    },
    {
      "epoch": 0.7440796994937122,
      "grad_norm": Infinity,
      "learning_rate": 3.251056560160821e-05,
      "loss": 1.9421,
      "step": 1139
    },
    {
      "epoch": 0.7447329740323371,
      "grad_norm": Infinity,
      "learning_rate": 3.2354080850191324e-05,
      "loss": 2.1221,
      "step": 1140
    },
    {
      "epoch": 0.7453862485709619,
      "grad_norm": Infinity,
      "learning_rate": 3.219790090073329e-05,
      "loss": 2.1755,
      "step": 1141
    },
    {
      "epoch": 0.7460395231095868,
      "grad_norm": Infinity,
      "learning_rate": 3.2042026456956555e-05,
      "loss": 2.3494,
      "step": 1142
    },
    {
      "epoch": 0.7466927976482116,
      "grad_norm": Infinity,
      "learning_rate": 3.188645822120716e-05,
      "loss": 2.5545,
      "step": 1143
    },
    {
      "epoch": 0.7473460721868365,
      "grad_norm": Infinity,
      "learning_rate": 3.1731196894451154e-05,
      "loss": 2.7081,
      "step": 1144
    },
    {
      "epoch": 0.7479993467254614,
      "grad_norm": Infinity,
      "learning_rate": 3.157624317627195e-05,
      "loss": 2.7034,
      "step": 1145
    },
    {
      "epoch": 0.7486526212640863,
      "grad_norm": Infinity,
      "learning_rate": 3.1421597764866864e-05,
      "loss": 2.9047,
      "step": 1146
    },
    {
      "epoch": 0.7493058958027111,
      "grad_norm": Infinity,
      "learning_rate": 3.126726135704389e-05,
      "loss": 2.915,
      "step": 1147
    },
    {
      "epoch": 0.749959170341336,
      "grad_norm": Infinity,
      "learning_rate": 3.1113234648218883e-05,
      "loss": 3.1268,
      "step": 1148
    },
    {
      "epoch": 0.7506124448799608,
      "grad_norm": Infinity,
      "learning_rate": 3.095951833241213e-05,
      "loss": 3.4791,
      "step": 1149
    },
    {
      "epoch": 0.7506124448799608,
      "eval_loss": 2.0990819931030273,
      "eval_runtime": 161.0223,
      "eval_samples_per_second": 16.01,
      "eval_steps_per_second": 4.006,
      "step": 1149
    },
    {
      "epoch": 0.7512657194185857,
      "grad_norm": Infinity,
      "learning_rate": 3.080611310224539e-05,
      "loss": 4.3268,
      "step": 1150
    },
    {
      "epoch": 0.7519189939572105,
      "grad_norm": Infinity,
      "learning_rate": 3.0653019648938685e-05,
      "loss": 1.2606,
      "step": 1151
    },
    {
      "epoch": 0.7525722684958354,
      "grad_norm": Infinity,
      "learning_rate": 3.0500238662307212e-05,
      "loss": 1.3563,
      "step": 1152
    },
    {
      "epoch": 0.7532255430344602,
      "grad_norm": Infinity,
      "learning_rate": 3.0347770830758316e-05,
      "loss": 1.4777,
      "step": 1153
    },
    {
      "epoch": 0.7538788175730851,
      "grad_norm": Infinity,
      "learning_rate": 3.019561684128823e-05,
      "loss": 1.373,
      "step": 1154
    },
    {
      "epoch": 0.7545320921117099,
      "grad_norm": Infinity,
      "learning_rate": 3.0043777379479098e-05,
      "loss": 1.4508,
      "step": 1155
    },
    {
      "epoch": 0.7551853666503348,
      "grad_norm": Infinity,
      "learning_rate": 2.989225312949584e-05,
      "loss": 1.4653,
      "step": 1156
    },
    {
      "epoch": 0.7558386411889597,
      "grad_norm": Infinity,
      "learning_rate": 2.9741044774083094e-05,
      "loss": 1.5171,
      "step": 1157
    },
    {
      "epoch": 0.7564919157275846,
      "grad_norm": Infinity,
      "learning_rate": 2.9590152994562104e-05,
      "loss": 1.6195,
      "step": 1158
    },
    {
      "epoch": 0.7571451902662094,
      "grad_norm": Infinity,
      "learning_rate": 2.9439578470827755e-05,
      "loss": 1.5933,
      "step": 1159
    },
    {
      "epoch": 0.7577984648048343,
      "grad_norm": Infinity,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 1.6688,
      "step": 1160
    },
    {
      "epoch": 0.7584517393434591,
      "grad_norm": Infinity,
      "learning_rate": 2.913938390314741e-05,
      "loss": 1.7732,
      "step": 1161
    },
    {
      "epoch": 0.759105013882084,
      "grad_norm": Infinity,
      "learning_rate": 2.8989765211831433e-05,
      "loss": 1.6398,
      "step": 1162
    },
    {
      "epoch": 0.7597582884207088,
      "grad_norm": Infinity,
      "learning_rate": 2.8840466481555672e-05,
      "loss": 1.8894,
      "step": 1163
    },
    {
      "epoch": 0.7604115629593337,
      "grad_norm": Infinity,
      "learning_rate": 2.8691488385037012e-05,
      "loss": 1.773,
      "step": 1164
    },
    {
      "epoch": 0.7610648374979585,
      "grad_norm": Infinity,
      "learning_rate": 2.854283159354748e-05,
      "loss": 1.9317,
      "step": 1165
    },
    {
      "epoch": 0.7617181120365834,
      "grad_norm": Infinity,
      "learning_rate": 2.8394496776911382e-05,
      "loss": 2.2086,
      "step": 1166
    },
    {
      "epoch": 0.7623713865752082,
      "grad_norm": Infinity,
      "learning_rate": 2.8246484603502275e-05,
      "loss": 2.3171,
      "step": 1167
    },
    {
      "epoch": 0.763024661113833,
      "grad_norm": Infinity,
      "learning_rate": 2.8098795740239923e-05,
      "loss": 2.2845,
      "step": 1168
    },
    {
      "epoch": 0.7636779356524579,
      "grad_norm": Infinity,
      "learning_rate": 2.7951430852587268e-05,
      "loss": 2.4874,
      "step": 1169
    },
    {
      "epoch": 0.7643312101910829,
      "grad_norm": Infinity,
      "learning_rate": 2.7804390604547557e-05,
      "loss": 2.7623,
      "step": 1170
    },
    {
      "epoch": 0.7649844847297077,
      "grad_norm": Infinity,
      "learning_rate": 2.7657675658661198e-05,
      "loss": 2.576,
      "step": 1171
    },
    {
      "epoch": 0.7656377592683326,
      "grad_norm": Infinity,
      "learning_rate": 2.7511286676002823e-05,
      "loss": 2.7432,
      "step": 1172
    },
    {
      "epoch": 0.7662910338069574,
      "grad_norm": Infinity,
      "learning_rate": 2.7365224316178384e-05,
      "loss": 2.7977,
      "step": 1173
    },
    {
      "epoch": 0.7669443083455822,
      "grad_norm": Infinity,
      "learning_rate": 2.7219489237322026e-05,
      "loss": 3.5219,
      "step": 1174
    },
    {
      "epoch": 0.7675975828842071,
      "grad_norm": Infinity,
      "learning_rate": 2.707408209609339e-05,
      "loss": 3.8416,
      "step": 1175
    },
    {
      "epoch": 0.768250857422832,
      "grad_norm": Infinity,
      "learning_rate": 2.692900354767425e-05,
      "loss": 1.4184,
      "step": 1176
    },
    {
      "epoch": 0.7689041319614568,
      "grad_norm": Infinity,
      "learning_rate": 2.678425424576596e-05,
      "loss": 1.3855,
      "step": 1177
    },
    {
      "epoch": 0.7695574065000816,
      "grad_norm": Infinity,
      "learning_rate": 2.6639834842586365e-05,
      "loss": 1.4003,
      "step": 1178
    },
    {
      "epoch": 0.7702106810387065,
      "grad_norm": Infinity,
      "learning_rate": 2.649574598886665e-05,
      "loss": 1.5033,
      "step": 1179
    },
    {
      "epoch": 0.7708639555773313,
      "grad_norm": Infinity,
      "learning_rate": 2.6351988333848788e-05,
      "loss": 1.4147,
      "step": 1180
    },
    {
      "epoch": 0.7715172301159562,
      "grad_norm": Infinity,
      "learning_rate": 2.620856252528232e-05,
      "loss": 1.4997,
      "step": 1181
    },
    {
      "epoch": 0.772170504654581,
      "grad_norm": Infinity,
      "learning_rate": 2.6065469209421566e-05,
      "loss": 1.5917,
      "step": 1182
    },
    {
      "epoch": 0.772823779193206,
      "grad_norm": Infinity,
      "learning_rate": 2.5922709031022686e-05,
      "loss": 1.5591,
      "step": 1183
    },
    {
      "epoch": 0.7734770537318308,
      "grad_norm": Infinity,
      "learning_rate": 2.578028263334078e-05,
      "loss": 1.5877,
      "step": 1184
    },
    {
      "epoch": 0.7741303282704557,
      "grad_norm": Infinity,
      "learning_rate": 2.5638190658126938e-05,
      "loss": 1.6238,
      "step": 1185
    },
    {
      "epoch": 0.7747836028090805,
      "grad_norm": Infinity,
      "learning_rate": 2.549643374562549e-05,
      "loss": 1.8098,
      "step": 1186
    },
    {
      "epoch": 0.7754368773477054,
      "grad_norm": Infinity,
      "learning_rate": 2.5355012534570953e-05,
      "loss": 1.8082,
      "step": 1187
    },
    {
      "epoch": 0.7760901518863302,
      "grad_norm": Infinity,
      "learning_rate": 2.52139276621852e-05,
      "loss": 1.8164,
      "step": 1188
    },
    {
      "epoch": 0.7767434264249551,
      "grad_norm": Infinity,
      "learning_rate": 2.507317976417475e-05,
      "loss": 2.0629,
      "step": 1189
    },
    {
      "epoch": 0.7773967009635799,
      "grad_norm": Infinity,
      "learning_rate": 2.493276947472756e-05,
      "loss": 2.1745,
      "step": 1190
    },
    {
      "epoch": 0.7780499755022048,
      "grad_norm": Infinity,
      "learning_rate": 2.4792697426510582e-05,
      "loss": 2.1568,
      "step": 1191
    },
    {
      "epoch": 0.7787032500408296,
      "grad_norm": Infinity,
      "learning_rate": 2.4652964250666567e-05,
      "loss": 2.3944,
      "step": 1192
    },
    {
      "epoch": 0.7793565245794545,
      "grad_norm": Infinity,
      "learning_rate": 2.4513570576811395e-05,
      "loss": 2.4762,
      "step": 1193
    },
    {
      "epoch": 0.7800097991180793,
      "grad_norm": Infinity,
      "learning_rate": 2.4374517033031285e-05,
      "loss": 2.3231,
      "step": 1194
    },
    {
      "epoch": 0.7806630736567042,
      "grad_norm": Infinity,
      "learning_rate": 2.4235804245879723e-05,
      "loss": 2.4375,
      "step": 1195
    },
    {
      "epoch": 0.7813163481953291,
      "grad_norm": Infinity,
      "learning_rate": 2.4097432840374945e-05,
      "loss": 2.8067,
      "step": 1196
    },
    {
      "epoch": 0.781969622733954,
      "grad_norm": Infinity,
      "learning_rate": 2.3959403439996907e-05,
      "loss": 2.957,
      "step": 1197
    },
    {
      "epoch": 0.7826228972725788,
      "grad_norm": Infinity,
      "learning_rate": 2.382171666668456e-05,
      "loss": 3.3291,
      "step": 1198
    },
    {
      "epoch": 0.7832761718112037,
      "grad_norm": Infinity,
      "learning_rate": 2.3684373140833016e-05,
      "loss": 3.5427,
      "step": 1199
    },
    {
      "epoch": 0.7839294463498285,
      "grad_norm": Infinity,
      "learning_rate": 2.354737348129077e-05,
      "loss": 3.9099,
      "step": 1200
    },
    {
      "epoch": 0.7845827208884534,
      "grad_norm": Infinity,
      "learning_rate": 2.3410718305356894e-05,
      "loss": 1.3826,
      "step": 1201
    },
    {
      "epoch": 0.7852359954270782,
      "grad_norm": Infinity,
      "learning_rate": 2.3274408228778355e-05,
      "loss": 1.4272,
      "step": 1202
    },
    {
      "epoch": 0.7858892699657031,
      "grad_norm": Infinity,
      "learning_rate": 2.3138443865747062e-05,
      "loss": 1.5302,
      "step": 1203
    },
    {
      "epoch": 0.7865425445043279,
      "grad_norm": Infinity,
      "learning_rate": 2.300282582889719e-05,
      "loss": 1.4341,
      "step": 1204
    },
    {
      "epoch": 0.7871958190429528,
      "grad_norm": Infinity,
      "learning_rate": 2.2867554729302542e-05,
      "loss": 1.5308,
      "step": 1205
    },
    {
      "epoch": 0.7878490935815776,
      "grad_norm": Infinity,
      "learning_rate": 2.2732631176473485e-05,
      "loss": 1.4427,
      "step": 1206
    },
    {
      "epoch": 0.7885023681202025,
      "grad_norm": Infinity,
      "learning_rate": 2.2598055778354587e-05,
      "loss": 1.4804,
      "step": 1207
    },
    {
      "epoch": 0.7891556426588274,
      "grad_norm": Infinity,
      "learning_rate": 2.2463829141321548e-05,
      "loss": 1.6638,
      "step": 1208
    },
    {
      "epoch": 0.7898089171974523,
      "grad_norm": Infinity,
      "learning_rate": 2.2329951870178655e-05,
      "loss": 1.5226,
      "step": 1209
    },
    {
      "epoch": 0.7904621917360771,
      "grad_norm": Infinity,
      "learning_rate": 2.2196424568156073e-05,
      "loss": 1.7148,
      "step": 1210
    },
    {
      "epoch": 0.791115466274702,
      "grad_norm": Infinity,
      "learning_rate": 2.2063247836906886e-05,
      "loss": 1.6323,
      "step": 1211
    },
    {
      "epoch": 0.7917687408133268,
      "grad_norm": Infinity,
      "learning_rate": 2.1930422276504747e-05,
      "loss": 1.7264,
      "step": 1212
    },
    {
      "epoch": 0.7924220153519517,
      "grad_norm": Infinity,
      "learning_rate": 2.17979484854409e-05,
      "loss": 1.8764,
      "step": 1213
    },
    {
      "epoch": 0.7930752898905765,
      "grad_norm": Infinity,
      "learning_rate": 2.1665827060621568e-05,
      "loss": 1.8152,
      "step": 1214
    },
    {
      "epoch": 0.7937285644292014,
      "grad_norm": Infinity,
      "learning_rate": 2.153405859736528e-05,
      "loss": 2.0223,
      "step": 1215
    },
    {
      "epoch": 0.7943818389678262,
      "grad_norm": Infinity,
      "learning_rate": 2.1402643689400192e-05,
      "loss": 1.9798,
      "step": 1216
    },
    {
      "epoch": 0.7950351135064511,
      "grad_norm": Infinity,
      "learning_rate": 2.1271582928861323e-05,
      "loss": 2.2652,
      "step": 1217
    },
    {
      "epoch": 0.7956883880450759,
      "grad_norm": Infinity,
      "learning_rate": 2.1140876906288086e-05,
      "loss": 2.5652,
      "step": 1218
    },
    {
      "epoch": 0.7963416625837008,
      "grad_norm": Infinity,
      "learning_rate": 2.1010526210621406e-05,
      "loss": 2.6046,
      "step": 1219
    },
    {
      "epoch": 0.7969949371223256,
      "grad_norm": Infinity,
      "learning_rate": 2.0880531429201145e-05,
      "loss": 2.4434,
      "step": 1220
    },
    {
      "epoch": 0.7976482116609506,
      "grad_norm": Infinity,
      "learning_rate": 2.0750893147763596e-05,
      "loss": 2.7545,
      "step": 1221
    },
    {
      "epoch": 0.7983014861995754,
      "grad_norm": Infinity,
      "learning_rate": 2.0621611950438512e-05,
      "loss": 3.1435,
      "step": 1222
    },
    {
      "epoch": 0.7989547607382003,
      "grad_norm": Infinity,
      "learning_rate": 2.0492688419746897e-05,
      "loss": 3.2472,
      "step": 1223
    },
    {
      "epoch": 0.7996080352768251,
      "grad_norm": Infinity,
      "learning_rate": 2.0364123136598035e-05,
      "loss": 3.4464,
      "step": 1224
    },
    {
      "epoch": 0.80026130981545,
      "grad_norm": Infinity,
      "learning_rate": 2.0235916680287015e-05,
      "loss": 3.9101,
      "step": 1225
    },
    {
      "epoch": 0.8009145843540748,
      "grad_norm": Infinity,
      "learning_rate": 2.010806962849219e-05,
      "loss": 1.3025,
      "step": 1226
    },
    {
      "epoch": 0.8015678588926997,
      "grad_norm": Infinity,
      "learning_rate": 1.998058255727234e-05,
      "loss": 1.2785,
      "step": 1227
    },
    {
      "epoch": 0.8022211334313245,
      "grad_norm": Infinity,
      "learning_rate": 1.985345604106439e-05,
      "loss": 1.397,
      "step": 1228
    },
    {
      "epoch": 0.8028744079699494,
      "grad_norm": Infinity,
      "learning_rate": 1.9726690652680578e-05,
      "loss": 1.4421,
      "step": 1229
    },
    {
      "epoch": 0.8035276825085742,
      "grad_norm": Infinity,
      "learning_rate": 1.9600286963305957e-05,
      "loss": 1.5002,
      "step": 1230
    },
    {
      "epoch": 0.804180957047199,
      "grad_norm": Infinity,
      "learning_rate": 1.9474245542495807e-05,
      "loss": 1.5233,
      "step": 1231
    },
    {
      "epoch": 0.8048342315858239,
      "grad_norm": Infinity,
      "learning_rate": 1.9348566958173197e-05,
      "loss": 1.5117,
      "step": 1232
    },
    {
      "epoch": 0.8054875061244487,
      "grad_norm": Infinity,
      "learning_rate": 1.9223251776626107e-05,
      "loss": 1.597,
      "step": 1233
    },
    {
      "epoch": 0.8061407806630737,
      "grad_norm": Infinity,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 1.5572,
      "step": 1234
    },
    {
      "epoch": 0.8067940552016986,
      "grad_norm": Infinity,
      "learning_rate": 1.897371387882134e-05,
      "loss": 1.7387,
      "step": 1235
    },
    {
      "epoch": 0.8074473297403234,
      "grad_norm": Infinity,
      "learning_rate": 1.884949228694246e-05,
      "loss": 1.738,
      "step": 1236
    },
    {
      "epoch": 0.8081006042789483,
      "grad_norm": Infinity,
      "learning_rate": 1.8725636346591822e-05,
      "loss": 1.7022,
      "step": 1237
    },
    {
      "epoch": 0.8087538788175731,
      "grad_norm": Infinity,
      "learning_rate": 1.860214661584486e-05,
      "loss": 1.8925,
      "step": 1238
    },
    {
      "epoch": 0.809407153356198,
      "grad_norm": Infinity,
      "learning_rate": 1.8479023651127115e-05,
      "loss": 2.0958,
      "step": 1239
    },
    {
      "epoch": 0.8100604278948228,
      "grad_norm": Infinity,
      "learning_rate": 1.835626800721144e-05,
      "loss": 1.893,
      "step": 1240
    },
    {
      "epoch": 0.8107137024334476,
      "grad_norm": Infinity,
      "learning_rate": 1.8233880237215585e-05,
      "loss": 2.0566,
      "step": 1241
    },
    {
      "epoch": 0.8113669769720725,
      "grad_norm": Infinity,
      "learning_rate": 1.811186089259983e-05,
      "loss": 2.0772,
      "step": 1242
    },
    {
      "epoch": 0.8120202515106973,
      "grad_norm": Infinity,
      "learning_rate": 1.7990210523164198e-05,
      "loss": 2.3327,
      "step": 1243
    },
    {
      "epoch": 0.8126735260493222,
      "grad_norm": Infinity,
      "learning_rate": 1.7868929677046364e-05,
      "loss": 2.2319,
      "step": 1244
    },
    {
      "epoch": 0.813326800587947,
      "grad_norm": Infinity,
      "learning_rate": 1.7748018900718854e-05,
      "loss": 2.6381,
      "step": 1245
    },
    {
      "epoch": 0.8139800751265719,
      "grad_norm": Infinity,
      "learning_rate": 1.762747873898679e-05,
      "loss": 2.5717,
      "step": 1246
    },
    {
      "epoch": 0.8146333496651968,
      "grad_norm": Infinity,
      "learning_rate": 1.750730973498529e-05,
      "loss": 3.0298,
      "step": 1247
    },
    {
      "epoch": 0.8152866242038217,
      "grad_norm": Infinity,
      "learning_rate": 1.7387512430177234e-05,
      "loss": 3.1643,
      "step": 1248
    },
    {
      "epoch": 0.8159398987424465,
      "grad_norm": Infinity,
      "learning_rate": 1.726808736435046e-05,
      "loss": 3.1612,
      "step": 1249
    },
    {
      "epoch": 0.8165931732810714,
      "grad_norm": Infinity,
      "learning_rate": 1.7149035075615794e-05,
      "loss": 3.9105,
      "step": 1250
    },
    {
      "epoch": 0.8172464478196962,
      "grad_norm": Infinity,
      "learning_rate": 1.703035610040423e-05,
      "loss": 1.1581,
      "step": 1251
    },
    {
      "epoch": 0.8178997223583211,
      "grad_norm": Infinity,
      "learning_rate": 1.69120509734647e-05,
      "loss": 1.3175,
      "step": 1252
    },
    {
      "epoch": 0.8185529968969459,
      "grad_norm": Infinity,
      "learning_rate": 1.679412022786172e-05,
      "loss": 1.3558,
      "step": 1253
    },
    {
      "epoch": 0.8192062714355708,
      "grad_norm": Infinity,
      "learning_rate": 1.6676564394972727e-05,
      "loss": 1.48,
      "step": 1254
    },
    {
      "epoch": 0.8198595459741956,
      "grad_norm": Infinity,
      "learning_rate": 1.6559384004486055e-05,
      "loss": 1.5116,
      "step": 1255
    },
    {
      "epoch": 0.8205128205128205,
      "grad_norm": Infinity,
      "learning_rate": 1.6442579584398232e-05,
      "loss": 1.4282,
      "step": 1256
    },
    {
      "epoch": 0.8211660950514453,
      "grad_norm": Infinity,
      "learning_rate": 1.6326151661011724e-05,
      "loss": 1.4939,
      "step": 1257
    },
    {
      "epoch": 0.8218193695900702,
      "grad_norm": Infinity,
      "learning_rate": 1.6210100758932667e-05,
      "loss": 1.4974,
      "step": 1258
    },
    {
      "epoch": 0.822472644128695,
      "grad_norm": Infinity,
      "learning_rate": 1.6094427401068224e-05,
      "loss": 1.4659,
      "step": 1259
    },
    {
      "epoch": 0.82312591866732,
      "grad_norm": Infinity,
      "learning_rate": 1.5979132108624574e-05,
      "loss": 1.5381,
      "step": 1260
    },
    {
      "epoch": 0.8237791932059448,
      "grad_norm": Infinity,
      "learning_rate": 1.5864215401104287e-05,
      "loss": 1.7296,
      "step": 1261
    },
    {
      "epoch": 0.8244324677445697,
      "grad_norm": Infinity,
      "learning_rate": 1.574967779630414e-05,
      "loss": 1.7319,
      "step": 1262
    },
    {
      "epoch": 0.8250857422831945,
      "grad_norm": Infinity,
      "learning_rate": 1.563551981031267e-05,
      "loss": 1.7433,
      "step": 1263
    },
    {
      "epoch": 0.8257390168218194,
      "grad_norm": Infinity,
      "learning_rate": 1.552174195750803e-05,
      "loss": 1.8011,
      "step": 1264
    },
    {
      "epoch": 0.8263922913604442,
      "grad_norm": Infinity,
      "learning_rate": 1.5408344750555383e-05,
      "loss": 2.0143,
      "step": 1265
    },
    {
      "epoch": 0.8270455658990691,
      "grad_norm": Infinity,
      "learning_rate": 1.529532870040492e-05,
      "loss": 2.0743,
      "step": 1266
    },
    {
      "epoch": 0.8276988404376939,
      "grad_norm": Infinity,
      "learning_rate": 1.5182694316289314e-05,
      "loss": 2.1506,
      "step": 1267
    },
    {
      "epoch": 0.8283521149763188,
      "grad_norm": Infinity,
      "learning_rate": 1.5070442105721495e-05,
      "loss": 2.1783,
      "step": 1268
    },
    {
      "epoch": 0.8290053895149436,
      "grad_norm": Infinity,
      "learning_rate": 1.4958572574492501e-05,
      "loss": 2.5343,
      "step": 1269
    },
    {
      "epoch": 0.8296586640535685,
      "grad_norm": Infinity,
      "learning_rate": 1.4847086226668872e-05,
      "loss": 2.6854,
      "step": 1270
    },
    {
      "epoch": 0.8303119385921933,
      "grad_norm": Infinity,
      "learning_rate": 1.4735983564590783e-05,
      "loss": 2.8695,
      "step": 1271
    },
    {
      "epoch": 0.8309652131308183,
      "grad_norm": Infinity,
      "learning_rate": 1.4625265088869477e-05,
      "loss": 2.959,
      "step": 1272
    },
    {
      "epoch": 0.8316184876694431,
      "grad_norm": Infinity,
      "learning_rate": 1.4514931298385093e-05,
      "loss": 3.3266,
      "step": 1273
    },
    {
      "epoch": 0.832271762208068,
      "grad_norm": Infinity,
      "learning_rate": 1.4404982690284575e-05,
      "loss": 3.8966,
      "step": 1274
    },
    {
      "epoch": 0.8329250367466928,
      "grad_norm": Infinity,
      "learning_rate": 1.429541975997908e-05,
      "loss": 4.5467,
      "step": 1275
    },
    {
      "epoch": 0.8335783112853177,
      "grad_norm": Infinity,
      "learning_rate": 1.4186243001142164e-05,
      "loss": 1.2431,
      "step": 1276
    },
    {
      "epoch": 0.8342315858239425,
      "grad_norm": Infinity,
      "learning_rate": 1.4077452905707234e-05,
      "loss": 1.424,
      "step": 1277
    },
    {
      "epoch": 0.8348848603625674,
      "grad_norm": Infinity,
      "learning_rate": 1.396904996386551e-05,
      "loss": 1.4275,
      "step": 1278
    },
    {
      "epoch": 0.8355381349011922,
      "grad_norm": Infinity,
      "learning_rate": 1.38610346640637e-05,
      "loss": 1.3774,
      "step": 1279
    },
    {
      "epoch": 0.8361914094398171,
      "grad_norm": Infinity,
      "learning_rate": 1.3753407493001968e-05,
      "loss": 1.4958,
      "step": 1280
    },
    {
      "epoch": 0.8368446839784419,
      "grad_norm": Infinity,
      "learning_rate": 1.3646168935631464e-05,
      "loss": 1.4944,
      "step": 1281
    },
    {
      "epoch": 0.8374979585170668,
      "grad_norm": Infinity,
      "learning_rate": 1.3539319475152457e-05,
      "loss": 1.5639,
      "step": 1282
    },
    {
      "epoch": 0.8381512330556916,
      "grad_norm": Infinity,
      "learning_rate": 1.3432859593011948e-05,
      "loss": 1.6105,
      "step": 1283
    },
    {
      "epoch": 0.8388045075943165,
      "grad_norm": Infinity,
      "learning_rate": 1.3326789768901504e-05,
      "loss": 1.7179,
      "step": 1284
    },
    {
      "epoch": 0.8394577821329414,
      "grad_norm": Infinity,
      "learning_rate": 1.3221110480755305e-05,
      "loss": 1.5552,
      "step": 1285
    },
    {
      "epoch": 0.8401110566715663,
      "grad_norm": Infinity,
      "learning_rate": 1.3115822204747619e-05,
      "loss": 1.6532,
      "step": 1286
    },
    {
      "epoch": 0.8407643312101911,
      "grad_norm": Infinity,
      "learning_rate": 1.3010925415291075e-05,
      "loss": 1.7276,
      "step": 1287
    },
    {
      "epoch": 0.841417605748816,
      "grad_norm": Infinity,
      "learning_rate": 1.2906420585034229e-05,
      "loss": 1.731,
      "step": 1288
    },
    {
      "epoch": 0.8420708802874408,
      "grad_norm": Infinity,
      "learning_rate": 1.2802308184859502e-05,
      "loss": 1.8079,
      "step": 1289
    },
    {
      "epoch": 0.8427241548260657,
      "grad_norm": Infinity,
      "learning_rate": 1.2698588683881186e-05,
      "loss": 1.8644,
      "step": 1290
    },
    {
      "epoch": 0.8433774293646905,
      "grad_norm": Infinity,
      "learning_rate": 1.2595262549443133e-05,
      "loss": 2.2312,
      "step": 1291
    },
    {
      "epoch": 0.8440307039033154,
      "grad_norm": Infinity,
      "learning_rate": 1.2492330247116802e-05,
      "loss": 2.2468,
      "step": 1292
    },
    {
      "epoch": 0.8446839784419402,
      "grad_norm": Infinity,
      "learning_rate": 1.2389792240699084e-05,
      "loss": 2.4955,
      "step": 1293
    },
    {
      "epoch": 0.8453372529805651,
      "grad_norm": Infinity,
      "learning_rate": 1.2287648992210243e-05,
      "loss": 2.4229,
      "step": 1294
    },
    {
      "epoch": 0.8459905275191899,
      "grad_norm": Infinity,
      "learning_rate": 1.2185900961891794e-05,
      "loss": 2.4597,
      "step": 1295
    },
    {
      "epoch": 0.8466438020578148,
      "grad_norm": Infinity,
      "learning_rate": 1.208454860820456e-05,
      "loss": 3.0822,
      "step": 1296
    },
    {
      "epoch": 0.8472970765964396,
      "grad_norm": Infinity,
      "learning_rate": 1.1983592387826347e-05,
      "loss": 3.045,
      "step": 1297
    },
    {
      "epoch": 0.8479503511350646,
      "grad_norm": Infinity,
      "learning_rate": 1.1883032755650204e-05,
      "loss": 3.0758,
      "step": 1298
    },
    {
      "epoch": 0.8486036256736894,
      "grad_norm": Infinity,
      "learning_rate": 1.1782870164782111e-05,
      "loss": 3.4634,
      "step": 1299
    },
    {
      "epoch": 0.8492569002123143,
      "grad_norm": Infinity,
      "learning_rate": 1.1683105066539068e-05,
      "loss": 4.4563,
      "step": 1300
    },
    {
      "epoch": 0.8499101747509391,
      "grad_norm": Infinity,
      "learning_rate": 1.15837379104471e-05,
      "loss": 1.1769,
      "step": 1301
    },
    {
      "epoch": 0.850563449289564,
      "grad_norm": Infinity,
      "learning_rate": 1.1484769144239037e-05,
      "loss": 1.3331,
      "step": 1302
    },
    {
      "epoch": 0.8512167238281888,
      "grad_norm": Infinity,
      "learning_rate": 1.1386199213852755e-05,
      "loss": 1.361,
      "step": 1303
    },
    {
      "epoch": 0.8518699983668137,
      "grad_norm": Infinity,
      "learning_rate": 1.1288028563428965e-05,
      "loss": 1.4695,
      "step": 1304
    },
    {
      "epoch": 0.8525232729054385,
      "grad_norm": Infinity,
      "learning_rate": 1.1190257635309275e-05,
      "loss": 1.5785,
      "step": 1305
    },
    {
      "epoch": 0.8531765474440633,
      "grad_norm": Infinity,
      "learning_rate": 1.1092886870034291e-05,
      "loss": 1.5306,
      "step": 1306
    },
    {
      "epoch": 0.8538298219826882,
      "grad_norm": Infinity,
      "learning_rate": 1.0995916706341459e-05,
      "loss": 1.4277,
      "step": 1307
    },
    {
      "epoch": 0.854483096521313,
      "grad_norm": Infinity,
      "learning_rate": 1.0899347581163221e-05,
      "loss": 1.6138,
      "step": 1308
    },
    {
      "epoch": 0.8551363710599379,
      "grad_norm": Infinity,
      "learning_rate": 1.0803179929624973e-05,
      "loss": 1.6152,
      "step": 1309
    },
    {
      "epoch": 0.8557896455985627,
      "grad_norm": Infinity,
      "learning_rate": 1.0707414185043163e-05,
      "loss": 1.6077,
      "step": 1310
    },
    {
      "epoch": 0.8564429201371877,
      "grad_norm": Infinity,
      "learning_rate": 1.0612050778923276e-05,
      "loss": 1.6382,
      "step": 1311
    },
    {
      "epoch": 0.8570961946758126,
      "grad_norm": Infinity,
      "learning_rate": 1.0517090140957998e-05,
      "loss": 1.7534,
      "step": 1312
    },
    {
      "epoch": 0.8577494692144374,
      "grad_norm": Infinity,
      "learning_rate": 1.0422532699025068e-05,
      "loss": 1.8991,
      "step": 1313
    },
    {
      "epoch": 0.8584027437530622,
      "grad_norm": Infinity,
      "learning_rate": 1.0328378879185641e-05,
      "loss": 1.9086,
      "step": 1314
    },
    {
      "epoch": 0.8590560182916871,
      "grad_norm": Infinity,
      "learning_rate": 1.0234629105682103e-05,
      "loss": 2.1443,
      "step": 1315
    },
    {
      "epoch": 0.8597092928303119,
      "grad_norm": Infinity,
      "learning_rate": 1.0141283800936297e-05,
      "loss": 2.0489,
      "step": 1316
    },
    {
      "epoch": 0.8603625673689368,
      "grad_norm": Infinity,
      "learning_rate": 1.0048343385547676e-05,
      "loss": 2.0992,
      "step": 1317
    },
    {
      "epoch": 0.8610158419075616,
      "grad_norm": Infinity,
      "learning_rate": 9.955808278291156e-06,
      "loss": 2.5535,
      "step": 1318
    },
    {
      "epoch": 0.8616691164461865,
      "grad_norm": Infinity,
      "learning_rate": 9.863678896115559e-06,
      "loss": 2.5644,
      "step": 1319
    },
    {
      "epoch": 0.8623223909848113,
      "grad_norm": Infinity,
      "learning_rate": 9.771955654141496e-06,
      "loss": 2.6898,
      "step": 1320
    },
    {
      "epoch": 0.8629756655234362,
      "grad_norm": Infinity,
      "learning_rate": 9.68063896565955e-06,
      "loss": 2.8441,
      "step": 1321
    },
    {
      "epoch": 0.863628940062061,
      "grad_norm": Infinity,
      "learning_rate": 9.589729242128532e-06,
      "loss": 2.7607,
      "step": 1322
    },
    {
      "epoch": 0.8642822146006859,
      "grad_norm": Infinity,
      "learning_rate": 9.499226893173453e-06,
      "loss": 3.2205,
      "step": 1323
    },
    {
      "epoch": 0.8649354891393108,
      "grad_norm": Infinity,
      "learning_rate": 9.409132326583758e-06,
      "loss": 3.5218,
      "step": 1324
    },
    {
      "epoch": 0.8655887636779357,
      "grad_norm": Infinity,
      "learning_rate": 9.319445948311534e-06,
      "loss": 3.9074,
      "step": 1325
    },
    {
      "epoch": 0.8662420382165605,
      "grad_norm": Infinity,
      "learning_rate": 9.230168162469599e-06,
      "loss": 1.3831,
      "step": 1326
    },
    {
      "epoch": 0.8668953127551854,
      "grad_norm": Infinity,
      "learning_rate": 9.141299371329704e-06,
      "loss": 1.446,
      "step": 1327
    },
    {
      "epoch": 0.8675485872938102,
      "grad_norm": Infinity,
      "learning_rate": 9.052839975320836e-06,
      "loss": 1.3385,
      "step": 1328
    },
    {
      "epoch": 0.8682018618324351,
      "grad_norm": Infinity,
      "learning_rate": 8.964790373027132e-06,
      "loss": 1.4632,
      "step": 1329
    },
    {
      "epoch": 0.8688551363710599,
      "grad_norm": Infinity,
      "learning_rate": 8.87715096118642e-06,
      "loss": 1.5529,
      "step": 1330
    },
    {
      "epoch": 0.8695084109096848,
      "grad_norm": Infinity,
      "learning_rate": 8.789922134688244e-06,
      "loss": 1.572,
      "step": 1331
    },
    {
      "epoch": 0.8701616854483096,
      "grad_norm": Infinity,
      "learning_rate": 8.703104286572028e-06,
      "loss": 1.5595,
      "step": 1332
    },
    {
      "epoch": 0.8708149599869345,
      "grad_norm": Infinity,
      "learning_rate": 8.616697808025486e-06,
      "loss": 1.5692,
      "step": 1333
    },
    {
      "epoch": 0.8714682345255593,
      "grad_norm": Infinity,
      "learning_rate": 8.53070308838273e-06,
      "loss": 1.535,
      "step": 1334
    },
    {
      "epoch": 0.8721215090641842,
      "grad_norm": Infinity,
      "learning_rate": 8.445120515122551e-06,
      "loss": 1.5121,
      "step": 1335
    },
    {
      "epoch": 0.8727747836028091,
      "grad_norm": Infinity,
      "learning_rate": 8.359950473866663e-06,
      "loss": 1.6005,
      "step": 1336
    },
    {
      "epoch": 0.873428058141434,
      "grad_norm": Infinity,
      "learning_rate": 8.275193348377974e-06,
      "loss": 1.7782,
      "step": 1337
    },
    {
      "epoch": 0.8740813326800588,
      "grad_norm": Infinity,
      "learning_rate": 8.190849520558908e-06,
      "loss": 2.003,
      "step": 1338
    },
    {
      "epoch": 0.8747346072186837,
      "grad_norm": Infinity,
      "learning_rate": 8.106919370449572e-06,
      "loss": 1.9703,
      "step": 1339
    },
    {
      "epoch": 0.8753878817573085,
      "grad_norm": Infinity,
      "learning_rate": 8.023403276226126e-06,
      "loss": 2.0316,
      "step": 1340
    },
    {
      "epoch": 0.8760411562959334,
      "grad_norm": Infinity,
      "learning_rate": 7.940301614199074e-06,
      "loss": 1.9369,
      "step": 1341
    },
    {
      "epoch": 0.8766944308345582,
      "grad_norm": Infinity,
      "learning_rate": 7.857614758811527e-06,
      "loss": 2.2225,
      "step": 1342
    },
    {
      "epoch": 0.8773477053731831,
      "grad_norm": Infinity,
      "learning_rate": 7.775343082637553e-06,
      "loss": 2.26,
      "step": 1343
    },
    {
      "epoch": 0.8780009799118079,
      "grad_norm": Infinity,
      "learning_rate": 7.693486956380525e-06,
      "loss": 2.3674,
      "step": 1344
    },
    {
      "epoch": 0.8786542544504328,
      "grad_norm": Infinity,
      "learning_rate": 7.612046748871327e-06,
      "loss": 2.9854,
      "step": 1345
    },
    {
      "epoch": 0.8793075289890576,
      "grad_norm": Infinity,
      "learning_rate": 7.531022827066858e-06,
      "loss": 2.6,
      "step": 1346
    },
    {
      "epoch": 0.8799608035276825,
      "grad_norm": Infinity,
      "learning_rate": 7.450415556048296e-06,
      "loss": 3.0325,
      "step": 1347
    },
    {
      "epoch": 0.8806140780663073,
      "grad_norm": Infinity,
      "learning_rate": 7.370225299019362e-06,
      "loss": 3.0668,
      "step": 1348
    },
    {
      "epoch": 0.8812673526049323,
      "grad_norm": Infinity,
      "learning_rate": 7.290452417304916e-06,
      "loss": 3.2571,
      "step": 1349
    },
    {
      "epoch": 0.8819206271435571,
      "grad_norm": Infinity,
      "learning_rate": 7.211097270349066e-06,
      "loss": 3.7934,
      "step": 1350
    },
    {
      "epoch": 0.882573901682182,
      "grad_norm": Infinity,
      "learning_rate": 7.132160215713757e-06,
      "loss": 1.2629,
      "step": 1351
    },
    {
      "epoch": 0.8832271762208068,
      "grad_norm": Infinity,
      "learning_rate": 7.053641609077033e-06,
      "loss": 1.4117,
      "step": 1352
    },
    {
      "epoch": 0.8838804507594317,
      "grad_norm": Infinity,
      "learning_rate": 6.975541804231478e-06,
      "loss": 1.3355,
      "step": 1353
    },
    {
      "epoch": 0.8845337252980565,
      "grad_norm": Infinity,
      "learning_rate": 6.897861153082618e-06,
      "loss": 1.4311,
      "step": 1354
    },
    {
      "epoch": 0.8851869998366814,
      "grad_norm": Infinity,
      "learning_rate": 6.820600005647382e-06,
      "loss": 1.5243,
      "step": 1355
    },
    {
      "epoch": 0.8858402743753062,
      "grad_norm": Infinity,
      "learning_rate": 6.743758710052439e-06,
      "loss": 1.4378,
      "step": 1356
    },
    {
      "epoch": 0.8864935489139311,
      "grad_norm": Infinity,
      "learning_rate": 6.6673376125326645e-06,
      "loss": 1.4713,
      "step": 1357
    },
    {
      "epoch": 0.8871468234525559,
      "grad_norm": Infinity,
      "learning_rate": 6.591337057429603e-06,
      "loss": 1.5855,
      "step": 1358
    },
    {
      "epoch": 0.8878000979911808,
      "grad_norm": Infinity,
      "learning_rate": 6.515757387189902e-06,
      "loss": 1.5555,
      "step": 1359
    },
    {
      "epoch": 0.8884533725298056,
      "grad_norm": Infinity,
      "learning_rate": 6.440598942363796e-06,
      "loss": 1.573,
      "step": 1360
    },
    {
      "epoch": 0.8891066470684305,
      "grad_norm": Infinity,
      "learning_rate": 6.365862061603456e-06,
      "loss": 1.7148,
      "step": 1361
    },
    {
      "epoch": 0.8897599216070554,
      "grad_norm": Infinity,
      "learning_rate": 6.291547081661631e-06,
      "loss": 1.6704,
      "step": 1362
    },
    {
      "epoch": 0.8904131961456803,
      "grad_norm": Infinity,
      "learning_rate": 6.217654337390078e-06,
      "loss": 1.8345,
      "step": 1363
    },
    {
      "epoch": 0.8910664706843051,
      "grad_norm": Infinity,
      "learning_rate": 6.144184161737887e-06,
      "loss": 1.8895,
      "step": 1364
    },
    {
      "epoch": 0.89171974522293,
      "grad_norm": Infinity,
      "learning_rate": 6.071136885750272e-06,
      "loss": 1.9469,
      "step": 1365
    },
    {
      "epoch": 0.8923730197615548,
      "grad_norm": Infinity,
      "learning_rate": 5.99851283856685e-06,
      "loss": 2.2034,
      "step": 1366
    },
    {
      "epoch": 0.8930262943001797,
      "grad_norm": Infinity,
      "learning_rate": 5.926312347420238e-06,
      "loss": 2.2942,
      "step": 1367
    },
    {
      "epoch": 0.8936795688388045,
      "grad_norm": Infinity,
      "learning_rate": 5.854535737634581e-06,
      "loss": 2.514,
      "step": 1368
    },
    {
      "epoch": 0.8943328433774294,
      "grad_norm": Infinity,
      "learning_rate": 5.783183332624098e-06,
      "loss": 2.5924,
      "step": 1369
    },
    {
      "epoch": 0.8949861179160542,
      "grad_norm": Infinity,
      "learning_rate": 5.71225545389158e-06,
      "loss": 2.6971,
      "step": 1370
    },
    {
      "epoch": 0.895639392454679,
      "grad_norm": Infinity,
      "learning_rate": 5.641752421027014e-06,
      "loss": 2.4167,
      "step": 1371
    },
    {
      "epoch": 0.8962926669933039,
      "grad_norm": Infinity,
      "learning_rate": 5.571674551706041e-06,
      "loss": 3.0307,
      "step": 1372
    },
    {
      "epoch": 0.8969459415319287,
      "grad_norm": Infinity,
      "learning_rate": 5.5020221616886025e-06,
      "loss": 3.0822,
      "step": 1373
    },
    {
      "epoch": 0.8975992160705536,
      "grad_norm": Infinity,
      "learning_rate": 5.432795564817539e-06,
      "loss": 3.6224,
      "step": 1374
    },
    {
      "epoch": 0.8982524906091786,
      "grad_norm": Infinity,
      "learning_rate": 5.363995073017047e-06,
      "loss": 3.9485,
      "step": 1375
    },
    {
      "epoch": 0.8989057651478034,
      "grad_norm": Infinity,
      "learning_rate": 5.295620996291451e-06,
      "loss": 1.369,
      "step": 1376
    },
    {
      "epoch": 0.8995590396864283,
      "grad_norm": Infinity,
      "learning_rate": 5.227673642723651e-06,
      "loss": 1.3912,
      "step": 1377
    },
    {
      "epoch": 0.9002123142250531,
      "grad_norm": Infinity,
      "learning_rate": 5.160153318473815e-06,
      "loss": 1.4591,
      "step": 1378
    },
    {
      "epoch": 0.900865588763678,
      "grad_norm": Infinity,
      "learning_rate": 5.093060327778043e-06,
      "loss": 1.491,
      "step": 1379
    },
    {
      "epoch": 0.9015188633023028,
      "grad_norm": Infinity,
      "learning_rate": 5.026394972946813e-06,
      "loss": 1.472,
      "step": 1380
    },
    {
      "epoch": 0.9021721378409276,
      "grad_norm": Infinity,
      "learning_rate": 4.9601575543638535e-06,
      "loss": 1.4614,
      "step": 1381
    },
    {
      "epoch": 0.9028254123795525,
      "grad_norm": Infinity,
      "learning_rate": 4.8943483704846475e-06,
      "loss": 1.5487,
      "step": 1382
    },
    {
      "epoch": 0.9034786869181773,
      "grad_norm": Infinity,
      "learning_rate": 4.828967717835087e-06,
      "loss": 1.546,
      "step": 1383
    },
    {
      "epoch": 0.9041319614568022,
      "grad_norm": Infinity,
      "learning_rate": 4.76401589101021e-06,
      "loss": 1.4852,
      "step": 1384
    },
    {
      "epoch": 0.904785235995427,
      "grad_norm": Infinity,
      "learning_rate": 4.69949318267281e-06,
      "loss": 1.742,
      "step": 1385
    },
    {
      "epoch": 0.9054385105340519,
      "grad_norm": Infinity,
      "learning_rate": 4.635399883552128e-06,
      "loss": 1.6791,
      "step": 1386
    },
    {
      "epoch": 0.9060917850726768,
      "grad_norm": Infinity,
      "learning_rate": 4.571736282442607e-06,
      "loss": 1.7485,
      "step": 1387
    },
    {
      "epoch": 0.9067450596113017,
      "grad_norm": Infinity,
      "learning_rate": 4.508502666202474e-06,
      "loss": 1.9477,
      "step": 1388
    },
    {
      "epoch": 0.9073983341499265,
      "grad_norm": Infinity,
      "learning_rate": 4.445699319752539e-06,
      "loss": 1.8895,
      "step": 1389
    },
    {
      "epoch": 0.9080516086885514,
      "grad_norm": Infinity,
      "learning_rate": 4.383326526074916e-06,
      "loss": 1.9285,
      "step": 1390
    },
    {
      "epoch": 0.9087048832271762,
      "grad_norm": Infinity,
      "learning_rate": 4.32138456621164e-06,
      "loss": 2.2509,
      "step": 1391
    },
    {
      "epoch": 0.9093581577658011,
      "grad_norm": Infinity,
      "learning_rate": 4.2598737192635405e-06,
      "loss": 2.3064,
      "step": 1392
    },
    {
      "epoch": 0.9100114323044259,
      "grad_norm": Infinity,
      "learning_rate": 4.198794262388905e-06,
      "loss": 2.6755,
      "step": 1393
    },
    {
      "epoch": 0.9106647068430508,
      "grad_norm": Infinity,
      "learning_rate": 4.138146470802218e-06,
      "loss": 2.3408,
      "step": 1394
    },
    {
      "epoch": 0.9113179813816756,
      "grad_norm": Infinity,
      "learning_rate": 4.077930617773007e-06,
      "loss": 2.8286,
      "step": 1395
    },
    {
      "epoch": 0.9119712559203005,
      "grad_norm": Infinity,
      "learning_rate": 4.018146974624448e-06,
      "loss": 2.7035,
      "step": 1396
    },
    {
      "epoch": 0.9126245304589253,
      "grad_norm": Infinity,
      "learning_rate": 3.958795810732363e-06,
      "loss": 2.9971,
      "step": 1397
    },
    {
      "epoch": 0.9132778049975502,
      "grad_norm": Infinity,
      "learning_rate": 3.899877393523821e-06,
      "loss": 3.2009,
      "step": 1398
    },
    {
      "epoch": 0.913931079536175,
      "grad_norm": Infinity,
      "learning_rate": 3.841391988476018e-06,
      "loss": 3.1459,
      "step": 1399
    },
    {
      "epoch": 0.9145843540748,
      "grad_norm": Infinity,
      "learning_rate": 3.783339859115065e-06,
      "loss": 3.91,
      "step": 1400
    },
    {
      "epoch": 0.9152376286134248,
      "grad_norm": Infinity,
      "learning_rate": 3.7257212670148035e-06,
      "loss": 1.3089,
      "step": 1401
    },
    {
      "epoch": 0.9158909031520497,
      "grad_norm": Infinity,
      "learning_rate": 3.6685364717956138e-06,
      "loss": 1.3872,
      "step": 1402
    },
    {
      "epoch": 0.9165441776906745,
      "grad_norm": Infinity,
      "learning_rate": 3.611785731123274e-06,
      "loss": 1.3304,
      "step": 1403
    },
    {
      "epoch": 0.9171974522292994,
      "grad_norm": Infinity,
      "learning_rate": 3.555469300707759e-06,
      "loss": 1.3995,
      "step": 1404
    },
    {
      "epoch": 0.9178507267679242,
      "grad_norm": Infinity,
      "learning_rate": 3.4995874343021094e-06,
      "loss": 1.4048,
      "step": 1405
    },
    {
      "epoch": 0.9185040013065491,
      "grad_norm": Infinity,
      "learning_rate": 3.444140383701333e-06,
      "loss": 1.4064,
      "step": 1406
    },
    {
      "epoch": 0.9191572758451739,
      "grad_norm": Infinity,
      "learning_rate": 3.389128398741148e-06,
      "loss": 1.596,
      "step": 1407
    },
    {
      "epoch": 0.9198105503837988,
      "grad_norm": Infinity,
      "learning_rate": 3.3345517272970084e-06,
      "loss": 1.5013,
      "step": 1408
    },
    {
      "epoch": 0.9204638249224236,
      "grad_norm": Infinity,
      "learning_rate": 3.2804106152828582e-06,
      "loss": 1.7069,
      "step": 1409
    },
    {
      "epoch": 0.9211170994610485,
      "grad_norm": Infinity,
      "learning_rate": 3.226705306650113e-06,
      "loss": 1.6444,
      "step": 1410
    },
    {
      "epoch": 0.9217703739996733,
      "grad_norm": Infinity,
      "learning_rate": 3.173436043386535e-06,
      "loss": 1.5149,
      "step": 1411
    },
    {
      "epoch": 0.9224236485382982,
      "grad_norm": Infinity,
      "learning_rate": 3.120603065515071e-06,
      "loss": 1.7224,
      "step": 1412
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": Infinity,
      "learning_rate": 3.068206611092905e-06,
      "loss": 1.7857,
      "step": 1413
    },
    {
      "epoch": 0.923730197615548,
      "grad_norm": Infinity,
      "learning_rate": 3.016246916210297e-06,
      "loss": 1.8834,
      "step": 1414
    },
    {
      "epoch": 0.9243834721541728,
      "grad_norm": Infinity,
      "learning_rate": 2.9647242149895006e-06,
      "loss": 1.9377,
      "step": 1415
    },
    {
      "epoch": 0.9250367466927977,
      "grad_norm": Infinity,
      "learning_rate": 2.9136387395837683e-06,
      "loss": 2.1515,
      "step": 1416
    },
    {
      "epoch": 0.9256900212314225,
      "grad_norm": Infinity,
      "learning_rate": 2.8629907201763283e-06,
      "loss": 2.3141,
      "step": 1417
    },
    {
      "epoch": 0.9263432957700474,
      "grad_norm": Infinity,
      "learning_rate": 2.8127803849791968e-06,
      "loss": 2.3314,
      "step": 1418
    },
    {
      "epoch": 0.9269965703086722,
      "grad_norm": Infinity,
      "learning_rate": 2.7630079602323442e-06,
      "loss": 2.3136,
      "step": 1419
    },
    {
      "epoch": 0.9276498448472971,
      "grad_norm": Infinity,
      "learning_rate": 2.7136736702025433e-06,
      "loss": 2.6563,
      "step": 1420
    },
    {
      "epoch": 0.9283031193859219,
      "grad_norm": Infinity,
      "learning_rate": 2.6647777371824e-06,
      "loss": 2.7181,
      "step": 1421
    },
    {
      "epoch": 0.9289563939245468,
      "grad_norm": Infinity,
      "learning_rate": 2.616320381489401e-06,
      "loss": 3.1091,
      "step": 1422
    },
    {
      "epoch": 0.9296096684631716,
      "grad_norm": Infinity,
      "learning_rate": 2.5683018214647693e-06,
      "loss": 3.1406,
      "step": 1423
    },
    {
      "epoch": 0.9302629430017965,
      "grad_norm": Infinity,
      "learning_rate": 2.520722273472698e-06,
      "loss": 3.4426,
      "step": 1424
    },
    {
      "epoch": 0.9309162175404213,
      "grad_norm": Infinity,
      "learning_rate": 2.473581951899184e-06,
      "loss": 4.6078,
      "step": 1425
    },
    {
      "epoch": 0.9315694920790463,
      "grad_norm": Infinity,
      "learning_rate": 2.4268810691511546e-06,
      "loss": 1.3089,
      "step": 1426
    },
    {
      "epoch": 0.9322227666176711,
      "grad_norm": Infinity,
      "learning_rate": 2.3806198356555287e-06,
      "loss": 1.4224,
      "step": 1427
    },
    {
      "epoch": 0.932876041156296,
      "grad_norm": Infinity,
      "learning_rate": 2.3347984598581783e-06,
      "loss": 1.4428,
      "step": 1428
    },
    {
      "epoch": 0.9335293156949208,
      "grad_norm": Infinity,
      "learning_rate": 2.289417148223094e-06,
      "loss": 1.4401,
      "step": 1429
    },
    {
      "epoch": 0.9341825902335457,
      "grad_norm": Infinity,
      "learning_rate": 2.2444761052313856e-06,
      "loss": 1.4837,
      "step": 1430
    },
    {
      "epoch": 0.9348358647721705,
      "grad_norm": Infinity,
      "learning_rate": 2.1999755333803718e-06,
      "loss": 1.5233,
      "step": 1431
    },
    {
      "epoch": 0.9354891393107954,
      "grad_norm": Infinity,
      "learning_rate": 2.1559156331826926e-06,
      "loss": 1.5427,
      "step": 1432
    },
    {
      "epoch": 0.9361424138494202,
      "grad_norm": Infinity,
      "learning_rate": 2.1122966031653977e-06,
      "loss": 1.5739,
      "step": 1433
    },
    {
      "epoch": 0.9367956883880451,
      "grad_norm": Infinity,
      "learning_rate": 2.069118639868994e-06,
      "loss": 1.525,
      "step": 1434
    },
    {
      "epoch": 0.9374489629266699,
      "grad_norm": Infinity,
      "learning_rate": 2.0263819378466884e-06,
      "loss": 1.7342,
      "step": 1435
    },
    {
      "epoch": 0.9381022374652948,
      "grad_norm": Infinity,
      "learning_rate": 1.984086689663378e-06,
      "loss": 1.9,
      "step": 1436
    },
    {
      "epoch": 0.9387555120039196,
      "grad_norm": Infinity,
      "learning_rate": 1.942233085894851e-06,
      "loss": 1.7909,
      "step": 1437
    },
    {
      "epoch": 0.9394087865425444,
      "grad_norm": Infinity,
      "learning_rate": 1.9008213151269328e-06,
      "loss": 1.8013,
      "step": 1438
    },
    {
      "epoch": 0.9400620610811694,
      "grad_norm": Infinity,
      "learning_rate": 1.8598515639545622e-06,
      "loss": 1.8483,
      "step": 1439
    },
    {
      "epoch": 0.9407153356197943,
      "grad_norm": Infinity,
      "learning_rate": 1.8193240169810943e-06,
      "loss": 2.0898,
      "step": 1440
    },
    {
      "epoch": 0.9413686101584191,
      "grad_norm": Infinity,
      "learning_rate": 1.7792388568173002e-06,
      "loss": 2.1185,
      "step": 1441
    },
    {
      "epoch": 0.942021884697044,
      "grad_norm": Infinity,
      "learning_rate": 1.7395962640806674e-06,
      "loss": 2.164,
      "step": 1442
    },
    {
      "epoch": 0.9426751592356688,
      "grad_norm": Infinity,
      "learning_rate": 1.7003964173945563e-06,
      "loss": 2.3025,
      "step": 1443
    },
    {
      "epoch": 0.9433284337742937,
      "grad_norm": Infinity,
      "learning_rate": 1.6616394933873235e-06,
      "loss": 2.4831,
      "step": 1444
    },
    {
      "epoch": 0.9439817083129185,
      "grad_norm": Infinity,
      "learning_rate": 1.623325666691644e-06,
      "loss": 2.6088,
      "step": 1445
    },
    {
      "epoch": 0.9446349828515433,
      "grad_norm": Infinity,
      "learning_rate": 1.5854551099436455e-06,
      "loss": 2.8619,
      "step": 1446
    },
    {
      "epoch": 0.9452882573901682,
      "grad_norm": Infinity,
      "learning_rate": 1.5480279937821418e-06,
      "loss": 2.9166,
      "step": 1447
    },
    {
      "epoch": 0.945941531928793,
      "grad_norm": Infinity,
      "learning_rate": 1.511044486847879e-06,
      "loss": 3.3526,
      "step": 1448
    },
    {
      "epoch": 0.9465948064674179,
      "grad_norm": Infinity,
      "learning_rate": 1.4745047557827796e-06,
      "loss": 3.6262,
      "step": 1449
    },
    {
      "epoch": 0.9472480810060427,
      "grad_norm": Infinity,
      "learning_rate": 1.4384089652291543e-06,
      "loss": 4.5926,
      "step": 1450
    },
    {
      "epoch": 0.9479013555446677,
      "grad_norm": Infinity,
      "learning_rate": 1.4027572778290255e-06,
      "loss": 1.2628,
      "step": 1451
    },
    {
      "epoch": 0.9485546300832925,
      "grad_norm": Infinity,
      "learning_rate": 1.3675498542233156e-06,
      "loss": 1.3955,
      "step": 1452
    },
    {
      "epoch": 0.9492079046219174,
      "grad_norm": Infinity,
      "learning_rate": 1.3327868530511934e-06,
      "loss": 1.3785,
      "step": 1453
    },
    {
      "epoch": 0.9498611791605422,
      "grad_norm": Infinity,
      "learning_rate": 1.2984684309493179e-06,
      "loss": 1.4708,
      "step": 1454
    },
    {
      "epoch": 0.9505144536991671,
      "grad_norm": Infinity,
      "learning_rate": 1.2645947425511395e-06,
      "loss": 1.4621,
      "step": 1455
    },
    {
      "epoch": 0.9511677282377919,
      "grad_norm": Infinity,
      "learning_rate": 1.231165940486234e-06,
      "loss": 1.567,
      "step": 1456
    },
    {
      "epoch": 0.9518210027764168,
      "grad_norm": Infinity,
      "learning_rate": 1.1981821753795365e-06,
      "loss": 1.6199,
      "step": 1457
    },
    {
      "epoch": 0.9524742773150416,
      "grad_norm": Infinity,
      "learning_rate": 1.1656435958507738e-06,
      "loss": 1.5783,
      "step": 1458
    },
    {
      "epoch": 0.9531275518536665,
      "grad_norm": Infinity,
      "learning_rate": 1.133550348513701e-06,
      "loss": 1.5525,
      "step": 1459
    },
    {
      "epoch": 0.9537808263922913,
      "grad_norm": Infinity,
      "learning_rate": 1.1019025779754666e-06,
      "loss": 1.6696,
      "step": 1460
    },
    {
      "epoch": 0.9544341009309162,
      "grad_norm": Infinity,
      "learning_rate": 1.0707004268360243e-06,
      "loss": 1.7598,
      "step": 1461
    },
    {
      "epoch": 0.955087375469541,
      "grad_norm": Infinity,
      "learning_rate": 1.039944035687368e-06,
      "loss": 1.8395,
      "step": 1462
    },
    {
      "epoch": 0.9557406500081659,
      "grad_norm": Infinity,
      "learning_rate": 1.0096335431130089e-06,
      "loss": 2.0216,
      "step": 1463
    },
    {
      "epoch": 0.9563939245467908,
      "grad_norm": Infinity,
      "learning_rate": 9.797690856872987e-07,
      "loss": 1.8501,
      "step": 1464
    },
    {
      "epoch": 0.9570471990854157,
      "grad_norm": Infinity,
      "learning_rate": 9.503507979748305e-07,
      "loss": 1.9238,
      "step": 1465
    },
    {
      "epoch": 0.9577004736240405,
      "grad_norm": Infinity,
      "learning_rate": 9.21378812529794e-07,
      "loss": 2.2574,
      "step": 1466
    },
    {
      "epoch": 0.9583537481626654,
      "grad_norm": Infinity,
      "learning_rate": 8.928532598954654e-07,
      "loss": 2.3334,
      "step": 1467
    },
    {
      "epoch": 0.9590070227012902,
      "grad_norm": Infinity,
      "learning_rate": 8.647742686035298e-07,
      "loss": 2.5487,
      "step": 1468
    },
    {
      "epoch": 0.9596602972399151,
      "grad_norm": Infinity,
      "learning_rate": 8.371419651735268e-07,
      "loss": 2.7241,
      "step": 1469
    },
    {
      "epoch": 0.9603135717785399,
      "grad_norm": Infinity,
      "learning_rate": 8.099564741123166e-07,
      "loss": 2.6671,
      "step": 1470
    },
    {
      "epoch": 0.9609668463171648,
      "grad_norm": Infinity,
      "learning_rate": 7.832179179134703e-07,
      "loss": 2.8966,
      "step": 1471
    },
    {
      "epoch": 0.9616201208557896,
      "grad_norm": Infinity,
      "learning_rate": 7.569264170567691e-07,
      "loss": 2.9527,
      "step": 1472
    },
    {
      "epoch": 0.9622733953944145,
      "grad_norm": Infinity,
      "learning_rate": 7.310820900075955e-07,
      "loss": 3.0179,
      "step": 1473
    },
    {
      "epoch": 0.9629266699330393,
      "grad_norm": Infinity,
      "learning_rate": 7.05685053216465e-07,
      "loss": 3.6077,
      "step": 1474
    },
    {
      "epoch": 0.9635799444716642,
      "grad_norm": Infinity,
      "learning_rate": 6.807354211184613e-07,
      "loss": 4.2041,
      "step": 1475
    },
    {
      "epoch": 0.964233219010289,
      "grad_norm": Infinity,
      "learning_rate": 6.562333061327364e-07,
      "loss": 1.3957,
      "step": 1476
    },
    {
      "epoch": 0.964886493548914,
      "grad_norm": Infinity,
      "learning_rate": 6.321788186619992e-07,
      "loss": 1.4075,
      "step": 1477
    },
    {
      "epoch": 0.9655397680875388,
      "grad_norm": Infinity,
      "learning_rate": 6.08572067092017e-07,
      "loss": 1.4448,
      "step": 1478
    },
    {
      "epoch": 0.9661930426261637,
      "grad_norm": Infinity,
      "learning_rate": 5.854131577911259e-07,
      "loss": 1.4893,
      "step": 1479
    },
    {
      "epoch": 0.9668463171647885,
      "grad_norm": Infinity,
      "learning_rate": 5.627021951097545e-07,
      "loss": 1.4216,
      "step": 1480
    },
    {
      "epoch": 0.9674995917034134,
      "grad_norm": Infinity,
      "learning_rate": 5.404392813799675e-07,
      "loss": 1.5241,
      "step": 1481
    },
    {
      "epoch": 0.9681528662420382,
      "grad_norm": Infinity,
      "learning_rate": 5.186245169149784e-07,
      "loss": 1.4522,
      "step": 1482
    },
    {
      "epoch": 0.9688061407806631,
      "grad_norm": Infinity,
      "learning_rate": 4.972580000087046e-07,
      "loss": 1.5146,
      "step": 1483
    },
    {
      "epoch": 0.9694594153192879,
      "grad_norm": Infinity,
      "learning_rate": 4.7633982693535693e-07,
      "loss": 1.6367,
      "step": 1484
    },
    {
      "epoch": 0.9701126898579128,
      "grad_norm": Infinity,
      "learning_rate": 4.5587009194894004e-07,
      "loss": 1.6609,
      "step": 1485
    },
    {
      "epoch": 0.9707659643965376,
      "grad_norm": Infinity,
      "learning_rate": 4.3584888728289695e-07,
      "loss": 1.6756,
      "step": 1486
    },
    {
      "epoch": 0.9714192389351625,
      "grad_norm": Infinity,
      "learning_rate": 4.1627630314965415e-07,
      "loss": 1.6661,
      "step": 1487
    },
    {
      "epoch": 0.9720725134737873,
      "grad_norm": Infinity,
      "learning_rate": 3.9715242774021055e-07,
      "loss": 1.8335,
      "step": 1488
    },
    {
      "epoch": 0.9727257880124122,
      "grad_norm": Infinity,
      "learning_rate": 3.7847734722378234e-07,
      "loss": 1.9859,
      "step": 1489
    },
    {
      "epoch": 0.9733790625510371,
      "grad_norm": Infinity,
      "learning_rate": 3.6025114574734785e-07,
      "loss": 2.034,
      "step": 1490
    },
    {
      "epoch": 0.974032337089662,
      "grad_norm": Infinity,
      "learning_rate": 3.424739054353476e-07,
      "loss": 1.9654,
      "step": 1491
    },
    {
      "epoch": 0.9746856116282868,
      "grad_norm": Infinity,
      "learning_rate": 3.2514570638925157e-07,
      "loss": 2.3266,
      "step": 1492
    },
    {
      "epoch": 0.9753388861669117,
      "grad_norm": Infinity,
      "learning_rate": 3.0826662668720364e-07,
      "loss": 2.358,
      "step": 1493
    },
    {
      "epoch": 0.9759921607055365,
      "grad_norm": Infinity,
      "learning_rate": 2.9183674238372206e-07,
      "loss": 2.384,
      "step": 1494
    },
    {
      "epoch": 0.9766454352441614,
      "grad_norm": Infinity,
      "learning_rate": 2.758561275092886e-07,
      "loss": 2.6566,
      "step": 1495
    },
    {
      "epoch": 0.9772987097827862,
      "grad_norm": Infinity,
      "learning_rate": 2.603248540700709e-07,
      "loss": 3.0257,
      "step": 1496
    },
    {
      "epoch": 0.9779519843214111,
      "grad_norm": Infinity,
      "learning_rate": 2.452429920475674e-07,
      "loss": 2.866,
      "step": 1497
    },
    {
      "epoch": 0.9786052588600359,
      "grad_norm": Infinity,
      "learning_rate": 2.3061060939828517e-07,
      "loss": 2.8559,
      "step": 1498
    },
    {
      "epoch": 0.9792585333986608,
      "grad_norm": Infinity,
      "learning_rate": 2.1642777205346242e-07,
      "loss": 3.8677,
      "step": 1499
    },
    {
      "epoch": 0.9799118079372856,
      "grad_norm": Infinity,
      "learning_rate": 2.0269454391874666e-07,
      "loss": 3.9609,
      "step": 1500
    },
    {
      "epoch": 0.9805650824759105,
      "grad_norm": Infinity,
      "learning_rate": 1.89410986873928e-07,
      "loss": 1.2609,
      "step": 1501
    },
    {
      "epoch": 0.9812183570145354,
      "grad_norm": Infinity,
      "learning_rate": 1.7657716077265073e-07,
      "loss": 1.307,
      "step": 1502
    },
    {
      "epoch": 0.9818716315531603,
      "grad_norm": Infinity,
      "learning_rate": 1.6419312344211347e-07,
      "loss": 1.3644,
      "step": 1503
    },
    {
      "epoch": 0.9825249060917851,
      "grad_norm": Infinity,
      "learning_rate": 1.5225893068286922e-07,
      "loss": 1.4293,
      "step": 1504
    },
    {
      "epoch": 0.98317818063041,
      "grad_norm": Infinity,
      "learning_rate": 1.4077463626852582e-07,
      "loss": 1.3799,
      "step": 1505
    },
    {
      "epoch": 0.9838314551690348,
      "grad_norm": Infinity,
      "learning_rate": 1.2974029194551262e-07,
      "loss": 1.4381,
      "step": 1506
    },
    {
      "epoch": 0.9844847297076597,
      "grad_norm": Infinity,
      "learning_rate": 1.1915594743288072e-07,
      "loss": 1.5463,
      "step": 1507
    },
    {
      "epoch": 0.9851380042462845,
      "grad_norm": Infinity,
      "learning_rate": 1.0902165042202539e-07,
      "loss": 1.5154,
      "step": 1508
    },
    {
      "epoch": 0.9857912787849094,
      "grad_norm": Infinity,
      "learning_rate": 9.933744657651956e-08,
      "loss": 1.6106,
      "step": 1509
    },
    {
      "epoch": 0.9864445533235342,
      "grad_norm": Infinity,
      "learning_rate": 9.010337953185843e-08,
      "loss": 1.6065,
      "step": 1510
    },
    {
      "epoch": 0.987097827862159,
      "grad_norm": Infinity,
      "learning_rate": 8.131949089531521e-08,
      "loss": 1.7721,
      "step": 1511
    },
    {
      "epoch": 0.9877511024007839,
      "grad_norm": Infinity,
      "learning_rate": 7.298582024571899e-08,
      "loss": 1.8281,
      "step": 1512
    },
    {
      "epoch": 0.9884043769394087,
      "grad_norm": Infinity,
      "learning_rate": 6.510240513328824e-08,
      "loss": 1.8281,
      "step": 1513
    },
    {
      "epoch": 0.9890576514780336,
      "grad_norm": Infinity,
      "learning_rate": 5.7669281079475446e-08,
      "loss": 1.8849,
      "step": 1514
    },
    {
      "epoch": 0.9897109260166586,
      "grad_norm": Infinity,
      "learning_rate": 5.068648157675604e-08,
      "loss": 1.9238,
      "step": 1515
    },
    {
      "epoch": 0.9903642005552834,
      "grad_norm": Infinity,
      "learning_rate": 4.4154038088561886e-08,
      "loss": 2.0281,
      "step": 1516
    },
    {
      "epoch": 0.9910174750939083,
      "grad_norm": Infinity,
      "learning_rate": 3.8071980049037006e-08,
      "loss": 2.3601,
      "step": 1517
    },
    {
      "epoch": 0.9916707496325331,
      "grad_norm": Infinity,
      "learning_rate": 3.244033486300424e-08,
      "loss": 2.4496,
      "step": 1518
    },
    {
      "epoch": 0.992324024171158,
      "grad_norm": Infinity,
      "learning_rate": 2.7259127905776562e-08,
      "loss": 2.399,
      "step": 1519
    },
    {
      "epoch": 0.9929772987097828,
      "grad_norm": Infinity,
      "learning_rate": 2.2528382523057113e-08,
      "loss": 2.5236,
      "step": 1520
    },
    {
      "epoch": 0.9936305732484076,
      "grad_norm": Infinity,
      "learning_rate": 1.8248120030850413e-08,
      "loss": 2.7336,
      "step": 1521
    },
    {
      "epoch": 0.9942838477870325,
      "grad_norm": Infinity,
      "learning_rate": 1.4418359715351327e-08,
      "loss": 2.898,
      "step": 1522
    },
    {
      "epoch": 0.9949371223256573,
      "grad_norm": Infinity,
      "learning_rate": 1.1039118832867345e-08,
      "loss": 3.4187,
      "step": 1523
    },
    {
      "epoch": 0.9955903968642822,
      "grad_norm": Infinity,
      "learning_rate": 8.110412609718676e-09,
      "loss": 3.3366,
      "step": 1524
    },
    {
      "epoch": 0.996243671402907,
      "grad_norm": Infinity,
      "learning_rate": 5.6322542422049266e-09,
      "loss": 4.1065,
      "step": 1525
    },
    {
      "epoch": 0.9968969459415319,
      "grad_norm": Infinity,
      "learning_rate": 3.604654896527393e-09,
      "loss": 1.3718,
      "step": 1526
    },
    {
      "epoch": 0.9975502204801567,
      "grad_norm": Infinity,
      "learning_rate": 2.027623708722448e-09,
      "loss": 1.5273,
      "step": 1527
    },
    {
      "epoch": 0.9982034950187817,
      "grad_norm": Infinity,
      "learning_rate": 9.011677846504363e-10,
      "loss": 1.8061,
      "step": 1528
    },
    {
      "epoch": 0.9988567695574065,
      "grad_norm": Infinity,
      "learning_rate": 2.252921999401636e-10,
      "loss": 2.0641,
      "step": 1529
    },
    {
      "epoch": 0.9995100440960314,
      "grad_norm": Infinity,
      "learning_rate": 0.0,
      "loss": 2.8842,
      "step": 1530
    }
  ],
  "logging_steps": 1,
  "max_steps": 1530,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 383,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.299736884851507e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}