{
  "best_metric": 1.2393983602523804,
  "best_model_checkpoint": "/export/data/salmasia/tradutor/checkpoints/hf_llama3_fft/checkpoint-3000",
  "epoch": 0.999869692287645,
  "eval_steps": 1000,
  "global_step": 3357,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0002978461996686461,
      "grad_norm": 34.83498001098633,
      "learning_rate": 2e-08,
      "loss": 2.5092,
      "step": 1
    },
    {
      "epoch": 0.0005956923993372922,
      "grad_norm": 34.53335189819336,
      "learning_rate": 4e-08,
      "loss": 2.5318,
      "step": 2
    },
    {
      "epoch": 0.0008935385990059383,
      "grad_norm": 34.28140640258789,
      "learning_rate": 6.000000000000001e-08,
      "loss": 2.5012,
      "step": 3
    },
    {
      "epoch": 0.0011913847986745843,
      "grad_norm": 33.16825866699219,
      "learning_rate": 8e-08,
      "loss": 2.4616,
      "step": 4
    },
    {
      "epoch": 0.0014892309983432306,
      "grad_norm": 33.8007926940918,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 2.5052,
      "step": 5
    },
    {
      "epoch": 0.0017870771980118766,
      "grad_norm": 34.03289031982422,
      "learning_rate": 1.2000000000000002e-07,
      "loss": 2.5146,
      "step": 6
    },
    {
      "epoch": 0.002084923397680523,
      "grad_norm": 32.96837615966797,
      "learning_rate": 1.4e-07,
      "loss": 2.5008,
      "step": 7
    },
    {
      "epoch": 0.0023827695973491687,
      "grad_norm": 32.52588653564453,
      "learning_rate": 1.6e-07,
      "loss": 2.4525,
      "step": 8
    },
    {
      "epoch": 0.002680615797017815,
      "grad_norm": 31.8045711517334,
      "learning_rate": 1.8e-07,
      "loss": 2.4804,
      "step": 9
    },
    {
      "epoch": 0.002978461996686461,
      "grad_norm": 30.462175369262695,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 2.4635,
      "step": 10
    },
    {
      "epoch": 0.003276308196355107,
      "grad_norm": 31.271331787109375,
      "learning_rate": 2.2e-07,
      "loss": 2.4849,
      "step": 11
    },
    {
      "epoch": 0.0035741543960237532,
      "grad_norm": 29.498044967651367,
      "learning_rate": 2.4000000000000003e-07,
      "loss": 2.495,
      "step": 12
    },
    {
      "epoch": 0.0038720005956923995,
      "grad_norm": 29.363866806030273,
      "learning_rate": 2.6e-07,
      "loss": 2.4967,
      "step": 13
    },
    {
      "epoch": 0.004169846795361046,
      "grad_norm": 28.964969635009766,
      "learning_rate": 2.8e-07,
      "loss": 2.4625,
      "step": 14
    },
    {
      "epoch": 0.004467692995029691,
      "grad_norm": 28.562702178955078,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 2.4551,
      "step": 15
    },
    {
      "epoch": 0.004765539194698337,
      "grad_norm": 27.29281997680664,
      "learning_rate": 3.2e-07,
      "loss": 2.4143,
      "step": 16
    },
    {
      "epoch": 0.005063385394366984,
      "grad_norm": 26.652280807495117,
      "learning_rate": 3.4000000000000003e-07,
      "loss": 2.3603,
      "step": 17
    },
    {
      "epoch": 0.00536123159403563,
      "grad_norm": 26.599451065063477,
      "learning_rate": 3.6e-07,
      "loss": 2.3918,
      "step": 18
    },
    {
      "epoch": 0.005659077793704276,
      "grad_norm": 25.272802352905273,
      "learning_rate": 3.8e-07,
      "loss": 2.3206,
      "step": 19
    },
    {
      "epoch": 0.005956923993372922,
      "grad_norm": 25.93230438232422,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 2.3542,
      "step": 20
    },
    {
      "epoch": 0.006254770193041568,
      "grad_norm": 23.03489875793457,
      "learning_rate": 4.2000000000000006e-07,
      "loss": 2.2338,
      "step": 21
    },
    {
      "epoch": 0.006552616392710214,
      "grad_norm": 19.766443252563477,
      "learning_rate": 4.4e-07,
      "loss": 2.1051,
      "step": 22
    },
    {
      "epoch": 0.00685046259237886,
      "grad_norm": 20.197357177734375,
      "learning_rate": 4.6000000000000004e-07,
      "loss": 2.14,
      "step": 23
    },
    {
      "epoch": 0.0071483087920475065,
      "grad_norm": 19.19549560546875,
      "learning_rate": 4.800000000000001e-07,
      "loss": 2.1115,
      "step": 24
    },
    {
      "epoch": 0.007446154991716153,
      "grad_norm": 18.26253318786621,
      "learning_rate": 5.000000000000001e-07,
      "loss": 2.0671,
      "step": 25
    },
    {
      "epoch": 0.007744001191384799,
      "grad_norm": 17.11869239807129,
      "learning_rate": 5.2e-07,
      "loss": 2.0335,
      "step": 26
    },
    {
      "epoch": 0.008041847391053444,
      "grad_norm": 17.106002807617188,
      "learning_rate": 5.4e-07,
      "loss": 2.0321,
      "step": 27
    },
    {
      "epoch": 0.008339693590722091,
      "grad_norm": 15.460068702697754,
      "learning_rate": 5.6e-07,
      "loss": 1.9684,
      "step": 28
    },
    {
      "epoch": 0.008637539790390737,
      "grad_norm": 11.006105422973633,
      "learning_rate": 5.800000000000001e-07,
      "loss": 1.8487,
      "step": 29
    },
    {
      "epoch": 0.008935385990059382,
      "grad_norm": 12.413909912109375,
      "learning_rate": 6.000000000000001e-07,
      "loss": 1.7969,
      "step": 30
    },
    {
      "epoch": 0.00923323218972803,
      "grad_norm": 11.074830055236816,
      "learning_rate": 6.200000000000001e-07,
      "loss": 1.7458,
      "step": 31
    },
    {
      "epoch": 0.009531078389396675,
      "grad_norm": 10.06778335571289,
      "learning_rate": 6.4e-07,
      "loss": 1.756,
      "step": 32
    },
    {
      "epoch": 0.009828924589065322,
      "grad_norm": 8.675497055053711,
      "learning_rate": 6.6e-07,
      "loss": 1.7158,
      "step": 33
    },
    {
      "epoch": 0.010126770788733967,
      "grad_norm": 7.968958854675293,
      "learning_rate": 6.800000000000001e-07,
      "loss": 1.7047,
      "step": 34
    },
    {
      "epoch": 0.010424616988402614,
      "grad_norm": 7.71165132522583,
      "learning_rate": 7.000000000000001e-07,
      "loss": 1.6919,
      "step": 35
    },
    {
      "epoch": 0.01072246318807126,
      "grad_norm": 7.4013543128967285,
      "learning_rate": 7.2e-07,
      "loss": 1.6758,
      "step": 36
    },
    {
      "epoch": 0.011020309387739905,
      "grad_norm": 6.465404033660889,
      "learning_rate": 7.4e-07,
      "loss": 1.6325,
      "step": 37
    },
    {
      "epoch": 0.011318155587408552,
      "grad_norm": 5.695428371429443,
      "learning_rate": 7.6e-07,
      "loss": 1.641,
      "step": 38
    },
    {
      "epoch": 0.011616001787077198,
      "grad_norm": 8.826560020446777,
      "learning_rate": 7.8e-07,
      "loss": 1.6209,
      "step": 39
    },
    {
      "epoch": 0.011913847986745845,
      "grad_norm": 8.145610809326172,
      "learning_rate": 8.000000000000001e-07,
      "loss": 1.6045,
      "step": 40
    },
    {
      "epoch": 0.01221169418641449,
      "grad_norm": 6.528449058532715,
      "learning_rate": 8.200000000000001e-07,
      "loss": 1.6041,
      "step": 41
    },
    {
      "epoch": 0.012509540386083135,
      "grad_norm": 5.022379398345947,
      "learning_rate": 8.400000000000001e-07,
      "loss": 1.6023,
      "step": 42
    },
    {
      "epoch": 0.012807386585751783,
      "grad_norm": 4.4542412757873535,
      "learning_rate": 8.6e-07,
      "loss": 1.5865,
      "step": 43
    },
    {
      "epoch": 0.013105232785420428,
      "grad_norm": 4.391264915466309,
      "learning_rate": 8.8e-07,
      "loss": 1.5966,
      "step": 44
    },
    {
      "epoch": 0.013403078985089075,
      "grad_norm": 4.17359733581543,
      "learning_rate": 9.000000000000001e-07,
      "loss": 1.5522,
      "step": 45
    },
    {
      "epoch": 0.01370092518475772,
      "grad_norm": 4.19918966293335,
      "learning_rate": 9.200000000000001e-07,
      "loss": 1.5895,
      "step": 46
    },
    {
      "epoch": 0.013998771384426368,
      "grad_norm": 4.2506818771362305,
      "learning_rate": 9.400000000000001e-07,
      "loss": 1.5687,
      "step": 47
    },
    {
      "epoch": 0.014296617584095013,
      "grad_norm": 4.17303991317749,
      "learning_rate": 9.600000000000001e-07,
      "loss": 1.5583,
      "step": 48
    },
    {
      "epoch": 0.014594463783763658,
      "grad_norm": 4.0772480964660645,
      "learning_rate": 9.800000000000001e-07,
      "loss": 1.5568,
      "step": 49
    },
    {
      "epoch": 0.014892309983432305,
      "grad_norm": 4.025362014770508,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.5453,
      "step": 50
    },
    {
      "epoch": 0.01519015618310095,
      "grad_norm": 3.906654119491577,
      "learning_rate": 1.02e-06,
      "loss": 1.5505,
      "step": 51
    },
    {
      "epoch": 0.015488002382769598,
      "grad_norm": 3.777599573135376,
      "learning_rate": 1.04e-06,
      "loss": 1.5388,
      "step": 52
    },
    {
      "epoch": 0.015785848582438245,
      "grad_norm": 3.7182812690734863,
      "learning_rate": 1.06e-06,
      "loss": 1.5208,
      "step": 53
    },
    {
      "epoch": 0.01608369478210689,
      "grad_norm": 3.8009390830993652,
      "learning_rate": 1.08e-06,
      "loss": 1.518,
      "step": 54
    },
    {
      "epoch": 0.016381540981775536,
      "grad_norm": 3.787339448928833,
      "learning_rate": 1.1e-06,
      "loss": 1.5215,
      "step": 55
    },
    {
      "epoch": 0.016679387181444183,
      "grad_norm": 3.7528140544891357,
      "learning_rate": 1.12e-06,
      "loss": 1.4972,
      "step": 56
    },
    {
      "epoch": 0.016977233381112827,
      "grad_norm": 3.8257265090942383,
      "learning_rate": 1.14e-06,
      "loss": 1.5345,
      "step": 57
    },
    {
      "epoch": 0.017275079580781474,
      "grad_norm": 3.6513845920562744,
      "learning_rate": 1.1600000000000001e-06,
      "loss": 1.5082,
      "step": 58
    },
    {
      "epoch": 0.01757292578045012,
      "grad_norm": 3.6084976196289062,
      "learning_rate": 1.1800000000000001e-06,
      "loss": 1.5102,
      "step": 59
    },
    {
      "epoch": 0.017870771980118764,
      "grad_norm": 3.7050273418426514,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 1.4969,
      "step": 60
    },
    {
      "epoch": 0.01816861817978741,
      "grad_norm": 3.416550397872925,
      "learning_rate": 1.2200000000000002e-06,
      "loss": 1.5007,
      "step": 61
    },
    {
      "epoch": 0.01846646437945606,
      "grad_norm": 3.314201593399048,
      "learning_rate": 1.2400000000000002e-06,
      "loss": 1.5204,
      "step": 62
    },
    {
      "epoch": 0.018764310579124706,
      "grad_norm": 3.056121826171875,
      "learning_rate": 1.26e-06,
      "loss": 1.4937,
      "step": 63
    },
    {
      "epoch": 0.01906215677879335,
      "grad_norm": 3.09470272064209,
      "learning_rate": 1.28e-06,
      "loss": 1.4942,
      "step": 64
    },
    {
      "epoch": 0.019360002978461997,
      "grad_norm": 2.9059770107269287,
      "learning_rate": 1.3e-06,
      "loss": 1.4962,
      "step": 65
    },
    {
      "epoch": 0.019657849178130644,
      "grad_norm": 2.898501396179199,
      "learning_rate": 1.32e-06,
      "loss": 1.5069,
      "step": 66
    },
    {
      "epoch": 0.019955695377799287,
      "grad_norm": 2.633026599884033,
      "learning_rate": 1.34e-06,
      "loss": 1.4815,
      "step": 67
    },
    {
      "epoch": 0.020253541577467934,
      "grad_norm": 2.517204761505127,
      "learning_rate": 1.3600000000000001e-06,
      "loss": 1.4605,
      "step": 68
    },
    {
      "epoch": 0.02055138777713658,
      "grad_norm": 2.39905047416687,
      "learning_rate": 1.3800000000000001e-06,
      "loss": 1.4682,
      "step": 69
    },
    {
      "epoch": 0.02084923397680523,
      "grad_norm": 2.328273057937622,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 1.4611,
      "step": 70
    },
    {
      "epoch": 0.021147080176473872,
      "grad_norm": 2.2634503841400146,
      "learning_rate": 1.42e-06,
      "loss": 1.4766,
      "step": 71
    },
    {
      "epoch": 0.02144492637614252,
      "grad_norm": 2.214564085006714,
      "learning_rate": 1.44e-06,
      "loss": 1.4643,
      "step": 72
    },
    {
      "epoch": 0.021742772575811167,
      "grad_norm": 2.2107338905334473,
      "learning_rate": 1.46e-06,
      "loss": 1.4523,
      "step": 73
    },
    {
      "epoch": 0.02204061877547981,
      "grad_norm": 2.266869306564331,
      "learning_rate": 1.48e-06,
      "loss": 1.4731,
      "step": 74
    },
    {
      "epoch": 0.022338464975148457,
      "grad_norm": 2.2619035243988037,
      "learning_rate": 1.5e-06,
      "loss": 1.4511,
      "step": 75
    },
    {
      "epoch": 0.022636311174817104,
      "grad_norm": 2.4727818965911865,
      "learning_rate": 1.52e-06,
      "loss": 1.4563,
      "step": 76
    },
    {
      "epoch": 0.02293415737448575,
      "grad_norm": 2.442145586013794,
      "learning_rate": 1.54e-06,
      "loss": 1.4299,
      "step": 77
    },
    {
      "epoch": 0.023232003574154395,
      "grad_norm": 2.7427873611450195,
      "learning_rate": 1.56e-06,
      "loss": 1.4515,
      "step": 78
    },
    {
      "epoch": 0.023529849773823042,
      "grad_norm": 2.7319259643554688,
      "learning_rate": 1.5800000000000001e-06,
      "loss": 1.4177,
      "step": 79
    },
    {
      "epoch": 0.02382769597349169,
      "grad_norm": 2.93934965133667,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 1.4528,
      "step": 80
    },
    {
      "epoch": 0.024125542173160333,
      "grad_norm": 3.001314163208008,
      "learning_rate": 1.6200000000000002e-06,
      "loss": 1.4286,
      "step": 81
    },
    {
      "epoch": 0.02442338837282898,
      "grad_norm": 3.327662944793701,
      "learning_rate": 1.6400000000000002e-06,
      "loss": 1.428,
      "step": 82
    },
    {
      "epoch": 0.024721234572497627,
      "grad_norm": 3.557964324951172,
      "learning_rate": 1.6600000000000002e-06,
      "loss": 1.4214,
      "step": 83
    },
    {
      "epoch": 0.02501908077216627,
      "grad_norm": 3.842317819595337,
      "learning_rate": 1.6800000000000002e-06,
      "loss": 1.4391,
      "step": 84
    },
    {
      "epoch": 0.025316926971834918,
      "grad_norm": 3.9461281299591064,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 1.4291,
      "step": 85
    },
    {
      "epoch": 0.025614773171503565,
      "grad_norm": 4.395602226257324,
      "learning_rate": 1.72e-06,
      "loss": 1.4369,
      "step": 86
    },
    {
      "epoch": 0.025912619371172212,
      "grad_norm": 4.573873043060303,
      "learning_rate": 1.74e-06,
      "loss": 1.4069,
      "step": 87
    },
    {
      "epoch": 0.026210465570840856,
      "grad_norm": 4.9001641273498535,
      "learning_rate": 1.76e-06,
      "loss": 1.4042,
      "step": 88
    },
    {
      "epoch": 0.026508311770509503,
      "grad_norm": 5.003119468688965,
      "learning_rate": 1.7800000000000001e-06,
      "loss": 1.3993,
      "step": 89
    },
    {
      "epoch": 0.02680615797017815,
      "grad_norm": 4.907381057739258,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 1.4033,
      "step": 90
    },
    {
      "epoch": 0.027104004169846794,
      "grad_norm": 4.829253196716309,
      "learning_rate": 1.8200000000000002e-06,
      "loss": 1.4113,
      "step": 91
    },
    {
      "epoch": 0.02740185036951544,
      "grad_norm": 4.23521089553833,
      "learning_rate": 1.8400000000000002e-06,
      "loss": 1.4007,
      "step": 92
    },
    {
      "epoch": 0.027699696569184088,
      "grad_norm": 3.7847466468811035,
      "learning_rate": 1.8600000000000002e-06,
      "loss": 1.3984,
      "step": 93
    },
    {
      "epoch": 0.027997542768852735,
      "grad_norm": 3.814786911010742,
      "learning_rate": 1.8800000000000002e-06,
      "loss": 1.3977,
      "step": 94
    },
    {
      "epoch": 0.02829538896852138,
      "grad_norm": 3.6349170207977295,
      "learning_rate": 1.9000000000000002e-06,
      "loss": 1.4038,
      "step": 95
    },
    {
      "epoch": 0.028593235168190026,
      "grad_norm": 3.8696351051330566,
      "learning_rate": 1.9200000000000003e-06,
      "loss": 1.3747,
      "step": 96
    },
    {
      "epoch": 0.028891081367858673,
      "grad_norm": 4.498185634613037,
      "learning_rate": 1.94e-06,
      "loss": 1.4039,
      "step": 97
    },
    {
      "epoch": 0.029188927567527317,
      "grad_norm": 4.332417011260986,
      "learning_rate": 1.9600000000000003e-06,
      "loss": 1.3736,
      "step": 98
    },
    {
      "epoch": 0.029486773767195964,
      "grad_norm": 4.389076232910156,
      "learning_rate": 1.98e-06,
      "loss": 1.3849,
      "step": 99
    },
    {
      "epoch": 0.02978461996686461,
      "grad_norm": 3.8623921871185303,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.3746,
      "step": 100
    },
    {
      "epoch": 0.030082466166533258,
      "grad_norm": 2.931136131286621,
      "learning_rate": 2.02e-06,
      "loss": 1.3989,
      "step": 101
    },
    {
      "epoch": 0.0303803123662019,
      "grad_norm": 1.8142356872558594,
      "learning_rate": 2.04e-06,
      "loss": 1.3828,
      "step": 102
    },
    {
      "epoch": 0.03067815856587055,
      "grad_norm": 1.2492481470108032,
      "learning_rate": 2.06e-06,
      "loss": 1.3817,
      "step": 103
    },
    {
      "epoch": 0.030976004765539196,
      "grad_norm": 1.0658767223358154,
      "learning_rate": 2.08e-06,
      "loss": 1.3654,
      "step": 104
    },
    {
      "epoch": 0.03127385096520784,
      "grad_norm": 1.1100436449050903,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 1.3901,
      "step": 105
    },
    {
      "epoch": 0.03157169716487649,
      "grad_norm": 1.0781537294387817,
      "learning_rate": 2.12e-06,
      "loss": 1.373,
      "step": 106
    },
    {
      "epoch": 0.03186954336454513,
      "grad_norm": 1.005652666091919,
      "learning_rate": 2.1400000000000003e-06,
      "loss": 1.3828,
      "step": 107
    },
    {
      "epoch": 0.03216738956421378,
      "grad_norm": 1.0426665544509888,
      "learning_rate": 2.16e-06,
      "loss": 1.3703,
      "step": 108
    },
    {
      "epoch": 0.032465235763882425,
      "grad_norm": 1.06223464012146,
      "learning_rate": 2.1800000000000003e-06,
      "loss": 1.3591,
      "step": 109
    },
    {
      "epoch": 0.03276308196355107,
      "grad_norm": 1.1222498416900635,
      "learning_rate": 2.2e-06,
      "loss": 1.3498,
      "step": 110
    },
    {
      "epoch": 0.03306092816321972,
      "grad_norm": 1.079365611076355,
      "learning_rate": 2.2200000000000003e-06,
      "loss": 1.3731,
      "step": 111
    },
    {
      "epoch": 0.033358774362888366,
      "grad_norm": 1.1567610502243042,
      "learning_rate": 2.24e-06,
      "loss": 1.3774,
      "step": 112
    },
    {
      "epoch": 0.03365662056255701,
      "grad_norm": 1.0220420360565186,
      "learning_rate": 2.2600000000000004e-06,
      "loss": 1.3572,
      "step": 113
    },
    {
      "epoch": 0.03395446676222565,
      "grad_norm": 1.1494303941726685,
      "learning_rate": 2.28e-06,
      "loss": 1.3789,
      "step": 114
    },
    {
      "epoch": 0.0342523129618943,
      "grad_norm": 1.1667444705963135,
      "learning_rate": 2.3000000000000004e-06,
      "loss": 1.3625,
      "step": 115
    },
    {
      "epoch": 0.03455015916156295,
      "grad_norm": 1.163591980934143,
      "learning_rate": 2.3200000000000002e-06,
      "loss": 1.3638,
      "step": 116
    },
    {
      "epoch": 0.034848005361231595,
      "grad_norm": 1.1720987558364868,
      "learning_rate": 2.3400000000000005e-06,
      "loss": 1.3481,
      "step": 117
    },
    {
      "epoch": 0.03514585156090024,
      "grad_norm": 1.0097452402114868,
      "learning_rate": 2.3600000000000003e-06,
      "loss": 1.3571,
      "step": 118
    },
    {
      "epoch": 0.03544369776056889,
      "grad_norm": 1.0675311088562012,
      "learning_rate": 2.38e-06,
      "loss": 1.3711,
      "step": 119
    },
    {
      "epoch": 0.03574154396023753,
      "grad_norm": 1.0158478021621704,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 1.3404,
      "step": 120
    },
    {
      "epoch": 0.036039390159906176,
      "grad_norm": 1.000169277191162,
      "learning_rate": 2.42e-06,
      "loss": 1.3661,
      "step": 121
    },
    {
      "epoch": 0.03633723635957482,
      "grad_norm": 1.017183780670166,
      "learning_rate": 2.4400000000000004e-06,
      "loss": 1.3579,
      "step": 122
    },
    {
      "epoch": 0.03663508255924347,
      "grad_norm": 1.0582289695739746,
      "learning_rate": 2.46e-06,
      "loss": 1.3553,
      "step": 123
    },
    {
      "epoch": 0.03693292875891212,
      "grad_norm": 1.0005322694778442,
      "learning_rate": 2.4800000000000004e-06,
      "loss": 1.3568,
      "step": 124
    },
    {
      "epoch": 0.037230774958580765,
      "grad_norm": 1.0457476377487183,
      "learning_rate": 2.5e-06,
      "loss": 1.3551,
      "step": 125
    },
    {
      "epoch": 0.03752862115824941,
      "grad_norm": 0.9791341423988342,
      "learning_rate": 2.52e-06,
      "loss": 1.3547,
      "step": 126
    },
    {
      "epoch": 0.03782646735791805,
      "grad_norm": 0.9721407890319824,
      "learning_rate": 2.5400000000000002e-06,
      "loss": 1.3525,
      "step": 127
    },
    {
      "epoch": 0.0381243135575867,
      "grad_norm": 1.0835869312286377,
      "learning_rate": 2.56e-06,
      "loss": 1.3603,
      "step": 128
    },
    {
      "epoch": 0.038422159757255346,
      "grad_norm": 1.0536324977874756,
      "learning_rate": 2.5800000000000003e-06,
      "loss": 1.3351,
      "step": 129
    },
    {
      "epoch": 0.03872000595692399,
      "grad_norm": 1.0062717199325562,
      "learning_rate": 2.6e-06,
      "loss": 1.3402,
      "step": 130
    },
    {
      "epoch": 0.03901785215659264,
      "grad_norm": 0.9959192276000977,
      "learning_rate": 2.6200000000000003e-06,
      "loss": 1.3468,
      "step": 131
    },
    {
      "epoch": 0.03931569835626129,
      "grad_norm": 1.0131630897521973,
      "learning_rate": 2.64e-06,
      "loss": 1.3361,
      "step": 132
    },
    {
      "epoch": 0.039613544555929935,
      "grad_norm": 1.0662428140640259,
      "learning_rate": 2.6600000000000004e-06,
      "loss": 1.3458,
      "step": 133
    },
    {
      "epoch": 0.039911390755598575,
      "grad_norm": 1.0271987915039062,
      "learning_rate": 2.68e-06,
      "loss": 1.3407,
      "step": 134
    },
    {
      "epoch": 0.04020923695526722,
      "grad_norm": 1.0550216436386108,
      "learning_rate": 2.7000000000000004e-06,
      "loss": 1.3522,
      "step": 135
    },
    {
      "epoch": 0.04050708315493587,
      "grad_norm": 1.1374857425689697,
      "learning_rate": 2.7200000000000002e-06,
      "loss": 1.3424,
      "step": 136
    },
    {
      "epoch": 0.040804929354604516,
      "grad_norm": 1.0215955972671509,
      "learning_rate": 2.7400000000000004e-06,
      "loss": 1.3453,
      "step": 137
    },
    {
      "epoch": 0.04110277555427316,
      "grad_norm": 1.0955356359481812,
      "learning_rate": 2.7600000000000003e-06,
      "loss": 1.3665,
      "step": 138
    },
    {
      "epoch": 0.04140062175394181,
      "grad_norm": 1.0931814908981323,
      "learning_rate": 2.7800000000000005e-06,
      "loss": 1.3111,
      "step": 139
    },
    {
      "epoch": 0.04169846795361046,
      "grad_norm": 1.1277751922607422,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 1.3379,
      "step": 140
    },
    {
      "epoch": 0.0419963141532791,
      "grad_norm": 1.0209782123565674,
      "learning_rate": 2.82e-06,
      "loss": 1.3272,
      "step": 141
    },
    {
      "epoch": 0.042294160352947745,
      "grad_norm": 1.0969637632369995,
      "learning_rate": 2.84e-06,
      "loss": 1.3357,
      "step": 142
    },
    {
      "epoch": 0.04259200655261639,
      "grad_norm": 1.035527229309082,
      "learning_rate": 2.86e-06,
      "loss": 1.3182,
      "step": 143
    },
    {
      "epoch": 0.04288985275228504,
      "grad_norm": 1.0502499341964722,
      "learning_rate": 2.88e-06,
      "loss": 1.3394,
      "step": 144
    },
    {
      "epoch": 0.043187698951953686,
      "grad_norm": 1.084592342376709,
      "learning_rate": 2.9e-06,
      "loss": 1.3296,
      "step": 145
    },
    {
      "epoch": 0.04348554515162233,
      "grad_norm": 1.082802414894104,
      "learning_rate": 2.92e-06,
      "loss": 1.325,
      "step": 146
    },
    {
      "epoch": 0.04378339135129098,
      "grad_norm": 1.0582551956176758,
      "learning_rate": 2.9400000000000002e-06,
      "loss": 1.3229,
      "step": 147
    },
    {
      "epoch": 0.04408123755095962,
      "grad_norm": 1.0280773639678955,
      "learning_rate": 2.96e-06,
      "loss": 1.3273,
      "step": 148
    },
    {
      "epoch": 0.04437908375062827,
      "grad_norm": 1.0459145307540894,
      "learning_rate": 2.9800000000000003e-06,
      "loss": 1.3359,
      "step": 149
    },
    {
      "epoch": 0.044676929950296915,
      "grad_norm": 1.0233818292617798,
      "learning_rate": 3e-06,
      "loss": 1.315,
      "step": 150
    },
    {
      "epoch": 0.04497477614996556,
      "grad_norm": 0.932620644569397,
      "learning_rate": 3.0200000000000003e-06,
      "loss": 1.3192,
      "step": 151
    },
    {
      "epoch": 0.04527262234963421,
      "grad_norm": 1.0429304838180542,
      "learning_rate": 3.04e-06,
      "loss": 1.3427,
      "step": 152
    },
    {
      "epoch": 0.045570468549302856,
      "grad_norm": 0.9749703407287598,
      "learning_rate": 3.0600000000000003e-06,
      "loss": 1.3171,
      "step": 153
    },
    {
      "epoch": 0.0458683147489715,
      "grad_norm": 0.9947569966316223,
      "learning_rate": 3.08e-06,
      "loss": 1.3289,
      "step": 154
    },
    {
      "epoch": 0.04616616094864014,
      "grad_norm": 0.9949209690093994,
      "learning_rate": 3.1000000000000004e-06,
      "loss": 1.318,
      "step": 155
    },
    {
      "epoch": 0.04646400714830879,
      "grad_norm": 1.0718587636947632,
      "learning_rate": 3.12e-06,
      "loss": 1.3089,
      "step": 156
    },
    {
      "epoch": 0.04676185334797744,
      "grad_norm": 0.9929345846176147,
      "learning_rate": 3.1400000000000004e-06,
      "loss": 1.3197,
      "step": 157
    },
    {
      "epoch": 0.047059699547646085,
      "grad_norm": 1.0306274890899658,
      "learning_rate": 3.1600000000000002e-06,
      "loss": 1.2952,
      "step": 158
    },
    {
      "epoch": 0.04735754574731473,
      "grad_norm": 1.0411291122436523,
      "learning_rate": 3.1800000000000005e-06,
      "loss": 1.3093,
      "step": 159
    },
    {
      "epoch": 0.04765539194698338,
      "grad_norm": 0.992257297039032,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 1.3207,
      "step": 160
    },
    {
      "epoch": 0.04795323814665202,
      "grad_norm": 1.0157545804977417,
      "learning_rate": 3.2200000000000005e-06,
      "loss": 1.3164,
      "step": 161
    },
    {
      "epoch": 0.048251084346320666,
      "grad_norm": 0.9970518350601196,
      "learning_rate": 3.2400000000000003e-06,
      "loss": 1.2977,
      "step": 162
    },
    {
      "epoch": 0.04854893054598931,
      "grad_norm": 0.9732163548469543,
      "learning_rate": 3.2600000000000006e-06,
      "loss": 1.3175,
      "step": 163
    },
    {
      "epoch": 0.04884677674565796,
      "grad_norm": 1.0057116746902466,
      "learning_rate": 3.2800000000000004e-06,
      "loss": 1.3238,
      "step": 164
    },
    {
      "epoch": 0.04914462294532661,
      "grad_norm": 1.011751413345337,
      "learning_rate": 3.3000000000000006e-06,
      "loss": 1.3051,
      "step": 165
    },
    {
      "epoch": 0.049442469144995255,
      "grad_norm": 1.004631519317627,
      "learning_rate": 3.3200000000000004e-06,
      "loss": 1.3096,
      "step": 166
    },
    {
      "epoch": 0.0497403153446639,
      "grad_norm": 0.9908256530761719,
      "learning_rate": 3.3400000000000006e-06,
      "loss": 1.3005,
      "step": 167
    },
    {
      "epoch": 0.05003816154433254,
      "grad_norm": 0.9953544735908508,
      "learning_rate": 3.3600000000000004e-06,
      "loss": 1.3195,
      "step": 168
    },
    {
      "epoch": 0.05033600774400119,
      "grad_norm": 1.019623875617981,
      "learning_rate": 3.3800000000000007e-06,
      "loss": 1.3069,
      "step": 169
    },
    {
      "epoch": 0.050633853943669836,
      "grad_norm": 1.0260343551635742,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 1.3155,
      "step": 170
    },
    {
      "epoch": 0.05093170014333848,
      "grad_norm": 0.956617534160614,
      "learning_rate": 3.4200000000000007e-06,
      "loss": 1.3031,
      "step": 171
    },
    {
      "epoch": 0.05122954634300713,
      "grad_norm": 1.0194716453552246,
      "learning_rate": 3.44e-06,
      "loss": 1.3009,
      "step": 172
    },
    {
      "epoch": 0.05152739254267578,
      "grad_norm": 1.022883415222168,
      "learning_rate": 3.46e-06,
      "loss": 1.3235,
      "step": 173
    },
    {
      "epoch": 0.051825238742344425,
      "grad_norm": 0.9606360793113708,
      "learning_rate": 3.48e-06,
      "loss": 1.3068,
      "step": 174
    },
    {
      "epoch": 0.052123084942013065,
      "grad_norm": 0.9691116213798523,
      "learning_rate": 3.5e-06,
      "loss": 1.314,
      "step": 175
    },
    {
      "epoch": 0.05242093114168171,
      "grad_norm": 1.0154898166656494,
      "learning_rate": 3.52e-06,
      "loss": 1.309,
      "step": 176
    },
    {
      "epoch": 0.05271877734135036,
      "grad_norm": 1.0041991472244263,
      "learning_rate": 3.54e-06,
      "loss": 1.2996,
      "step": 177
    },
    {
      "epoch": 0.053016623541019006,
      "grad_norm": 0.9984833002090454,
      "learning_rate": 3.5600000000000002e-06,
      "loss": 1.3088,
      "step": 178
    },
    {
      "epoch": 0.05331446974068765,
      "grad_norm": 1.0310505628585815,
      "learning_rate": 3.58e-06,
      "loss": 1.3057,
      "step": 179
    },
    {
      "epoch": 0.0536123159403563,
      "grad_norm": 1.004197597503662,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 1.3103,
      "step": 180
    },
    {
      "epoch": 0.05391016214002495,
      "grad_norm": 0.9437684416770935,
      "learning_rate": 3.62e-06,
      "loss": 1.2985,
      "step": 181
    },
    {
      "epoch": 0.05420800833969359,
      "grad_norm": 1.027902603149414,
      "learning_rate": 3.6400000000000003e-06,
      "loss": 1.3059,
      "step": 182
    },
    {
      "epoch": 0.054505854539362235,
      "grad_norm": 1.0154935121536255,
      "learning_rate": 3.66e-06,
      "loss": 1.2915,
      "step": 183
    },
    {
      "epoch": 0.05480370073903088,
      "grad_norm": 1.0160235166549683,
      "learning_rate": 3.6800000000000003e-06,
      "loss": 1.3152,
      "step": 184
    },
    {
      "epoch": 0.05510154693869953,
      "grad_norm": 0.9995388984680176,
      "learning_rate": 3.7e-06,
      "loss": 1.2837,
      "step": 185
    },
    {
      "epoch": 0.055399393138368176,
      "grad_norm": 0.9829667806625366,
      "learning_rate": 3.7200000000000004e-06,
      "loss": 1.2949,
      "step": 186
    },
    {
      "epoch": 0.05569723933803682,
      "grad_norm": 0.9536057710647583,
      "learning_rate": 3.74e-06,
      "loss": 1.2988,
      "step": 187
    },
    {
      "epoch": 0.05599508553770547,
      "grad_norm": 1.0392848253250122,
      "learning_rate": 3.7600000000000004e-06,
      "loss": 1.3125,
      "step": 188
    },
    {
      "epoch": 0.05629293173737411,
      "grad_norm": 0.9442264437675476,
      "learning_rate": 3.7800000000000002e-06,
      "loss": 1.2973,
      "step": 189
    },
    {
      "epoch": 0.05659077793704276,
      "grad_norm": 1.0178556442260742,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 1.298,
      "step": 190
    },
    {
      "epoch": 0.056888624136711405,
      "grad_norm": 1.0009549856185913,
      "learning_rate": 3.820000000000001e-06,
      "loss": 1.3203,
      "step": 191
    },
    {
      "epoch": 0.05718647033638005,
      "grad_norm": 0.9697956442832947,
      "learning_rate": 3.8400000000000005e-06,
      "loss": 1.3147,
      "step": 192
    },
    {
      "epoch": 0.0574843165360487,
      "grad_norm": 0.97239089012146,
      "learning_rate": 3.86e-06,
      "loss": 1.3061,
      "step": 193
    },
    {
      "epoch": 0.057782162735717346,
      "grad_norm": 1.0006135702133179,
      "learning_rate": 3.88e-06,
      "loss": 1.3116,
      "step": 194
    },
    {
      "epoch": 0.05808000893538599,
      "grad_norm": 0.9536453485488892,
      "learning_rate": 3.900000000000001e-06,
      "loss": 1.2914,
      "step": 195
    },
    {
      "epoch": 0.058377855135054633,
      "grad_norm": 0.9326615929603577,
      "learning_rate": 3.920000000000001e-06,
      "loss": 1.2791,
      "step": 196
    },
    {
      "epoch": 0.05867570133472328,
      "grad_norm": 0.970112681388855,
      "learning_rate": 3.94e-06,
      "loss": 1.2992,
      "step": 197
    },
    {
      "epoch": 0.05897354753439193,
      "grad_norm": 1.0164474248886108,
      "learning_rate": 3.96e-06,
      "loss": 1.2886,
      "step": 198
    },
    {
      "epoch": 0.059271393734060575,
      "grad_norm": 0.9721841216087341,
      "learning_rate": 3.980000000000001e-06,
      "loss": 1.2922,
      "step": 199
    },
    {
      "epoch": 0.05956923993372922,
      "grad_norm": 0.9503871202468872,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.2803,
      "step": 200
    },
    {
      "epoch": 0.05986708613339787,
      "grad_norm": 0.9525935649871826,
      "learning_rate": 4.0200000000000005e-06,
      "loss": 1.2994,
      "step": 201
    },
    {
      "epoch": 0.060164932333066516,
      "grad_norm": 1.0139000415802002,
      "learning_rate": 4.04e-06,
      "loss": 1.2797,
      "step": 202
    },
    {
      "epoch": 0.060462778532735156,
      "grad_norm": 0.971802830696106,
      "learning_rate": 4.060000000000001e-06,
      "loss": 1.3056,
      "step": 203
    },
    {
      "epoch": 0.0607606247324038,
      "grad_norm": 0.9166972637176514,
      "learning_rate": 4.08e-06,
      "loss": 1.2869,
      "step": 204
    },
    {
      "epoch": 0.06105847093207245,
      "grad_norm": 0.9614176750183105,
      "learning_rate": 4.1e-06,
      "loss": 1.2846,
      "step": 205
    },
    {
      "epoch": 0.0613563171317411,
      "grad_norm": 0.9437718987464905,
      "learning_rate": 4.12e-06,
      "loss": 1.2774,
      "step": 206
    },
    {
      "epoch": 0.061654163331409745,
      "grad_norm": 0.970997154712677,
      "learning_rate": 4.14e-06,
      "loss": 1.2899,
      "step": 207
    },
    {
      "epoch": 0.06195200953107839,
      "grad_norm": 0.9311395883560181,
      "learning_rate": 4.16e-06,
      "loss": 1.2959,
      "step": 208
    },
    {
      "epoch": 0.06224985573074703,
      "grad_norm": 0.9635146260261536,
      "learning_rate": 4.18e-06,
      "loss": 1.3112,
      "step": 209
    },
    {
      "epoch": 0.06254770193041569,
      "grad_norm": 0.9802709817886353,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 1.3015,
      "step": 210
    },
    {
      "epoch": 0.06284554813008433,
      "grad_norm": 1.0112463235855103,
      "learning_rate": 4.22e-06,
      "loss": 1.2901,
      "step": 211
    },
    {
      "epoch": 0.06314339432975298,
      "grad_norm": 0.9580091238021851,
      "learning_rate": 4.24e-06,
      "loss": 1.2939,
      "step": 212
    },
    {
      "epoch": 0.06344124052942161,
      "grad_norm": 0.9963754415512085,
      "learning_rate": 4.26e-06,
      "loss": 1.2736,
      "step": 213
    },
    {
      "epoch": 0.06373908672909026,
      "grad_norm": 0.9625868797302246,
      "learning_rate": 4.2800000000000005e-06,
      "loss": 1.2964,
      "step": 214
    },
    {
      "epoch": 0.06403693292875891,
      "grad_norm": 0.9352360367774963,
      "learning_rate": 4.3e-06,
      "loss": 1.2682,
      "step": 215
    },
    {
      "epoch": 0.06433477912842755,
      "grad_norm": 0.9316005110740662,
      "learning_rate": 4.32e-06,
      "loss": 1.2864,
      "step": 216
    },
    {
      "epoch": 0.0646326253280962,
      "grad_norm": 1.0154376029968262,
      "learning_rate": 4.34e-06,
      "loss": 1.294,
      "step": 217
    },
    {
      "epoch": 0.06493047152776485,
      "grad_norm": 0.9745945930480957,
      "learning_rate": 4.360000000000001e-06,
      "loss": 1.2926,
      "step": 218
    },
    {
      "epoch": 0.0652283177274335,
      "grad_norm": 0.9170344471931458,
      "learning_rate": 4.38e-06,
      "loss": 1.3001,
      "step": 219
    },
    {
      "epoch": 0.06552616392710214,
      "grad_norm": 0.9875105023384094,
      "learning_rate": 4.4e-06,
      "loss": 1.2911,
      "step": 220
    },
    {
      "epoch": 0.06582401012677079,
      "grad_norm": 0.9788703322410583,
      "learning_rate": 4.42e-06,
      "loss": 1.2875,
      "step": 221
    },
    {
      "epoch": 0.06612185632643944,
      "grad_norm": 0.9753866791725159,
      "learning_rate": 4.440000000000001e-06,
      "loss": 1.2836,
      "step": 222
    },
    {
      "epoch": 0.06641970252610808,
      "grad_norm": 0.9529588222503662,
      "learning_rate": 4.4600000000000005e-06,
      "loss": 1.2741,
      "step": 223
    },
    {
      "epoch": 0.06671754872577673,
      "grad_norm": 0.9625334739685059,
      "learning_rate": 4.48e-06,
      "loss": 1.2755,
      "step": 224
    },
    {
      "epoch": 0.06701539492544538,
      "grad_norm": 0.9622026085853577,
      "learning_rate": 4.5e-06,
      "loss": 1.2752,
      "step": 225
    },
    {
      "epoch": 0.06731324112511403,
      "grad_norm": 0.9499610066413879,
      "learning_rate": 4.520000000000001e-06,
      "loss": 1.2695,
      "step": 226
    },
    {
      "epoch": 0.06761108732478266,
      "grad_norm": 0.9504403471946716,
      "learning_rate": 4.540000000000001e-06,
      "loss": 1.2882,
      "step": 227
    },
    {
      "epoch": 0.0679089335244513,
      "grad_norm": 0.9664710164070129,
      "learning_rate": 4.56e-06,
      "loss": 1.2768,
      "step": 228
    },
    {
      "epoch": 0.06820677972411995,
      "grad_norm": 0.9604812860488892,
      "learning_rate": 4.58e-06,
      "loss": 1.295,
      "step": 229
    },
    {
      "epoch": 0.0685046259237886,
      "grad_norm": 1.0047502517700195,
      "learning_rate": 4.600000000000001e-06,
      "loss": 1.279,
      "step": 230
    },
    {
      "epoch": 0.06880247212345725,
      "grad_norm": 0.9735051393508911,
      "learning_rate": 4.620000000000001e-06,
      "loss": 1.2892,
      "step": 231
    },
    {
      "epoch": 0.0691003183231259,
      "grad_norm": 0.971291184425354,
      "learning_rate": 4.6400000000000005e-06,
      "loss": 1.2977,
      "step": 232
    },
    {
      "epoch": 0.06939816452279454,
      "grad_norm": 0.9723486304283142,
      "learning_rate": 4.66e-06,
      "loss": 1.2838,
      "step": 233
    },
    {
      "epoch": 0.06969601072246319,
      "grad_norm": 0.9506466388702393,
      "learning_rate": 4.680000000000001e-06,
      "loss": 1.2795,
      "step": 234
    },
    {
      "epoch": 0.06999385692213184,
      "grad_norm": 0.9426427483558655,
      "learning_rate": 4.7e-06,
      "loss": 1.2781,
      "step": 235
    },
    {
      "epoch": 0.07029170312180048,
      "grad_norm": 0.9696216583251953,
      "learning_rate": 4.7200000000000005e-06,
      "loss": 1.282,
      "step": 236
    },
    {
      "epoch": 0.07058954932146913,
      "grad_norm": 0.9565459489822388,
      "learning_rate": 4.74e-06,
      "loss": 1.2771,
      "step": 237
    },
    {
      "epoch": 0.07088739552113778,
      "grad_norm": 0.9609894752502441,
      "learning_rate": 4.76e-06,
      "loss": 1.2843,
      "step": 238
    },
    {
      "epoch": 0.07118524172080642,
      "grad_norm": 0.9644325375556946,
      "learning_rate": 4.78e-06,
      "loss": 1.2863,
      "step": 239
    },
    {
      "epoch": 0.07148308792047506,
      "grad_norm": 0.9697712063789368,
      "learning_rate": 4.800000000000001e-06,
      "loss": 1.2997,
      "step": 240
    },
    {
      "epoch": 0.0717809341201437,
      "grad_norm": 0.9858734607696533,
      "learning_rate": 4.8200000000000004e-06,
      "loss": 1.2831,
      "step": 241
    },
    {
      "epoch": 0.07207878031981235,
      "grad_norm": 0.9530293345451355,
      "learning_rate": 4.84e-06,
      "loss": 1.2687,
      "step": 242
    },
    {
      "epoch": 0.072376626519481,
      "grad_norm": 0.9498800039291382,
      "learning_rate": 4.86e-06,
      "loss": 1.2617,
      "step": 243
    },
    {
      "epoch": 0.07267447271914965,
      "grad_norm": 0.9868090748786926,
      "learning_rate": 4.880000000000001e-06,
      "loss": 1.2565,
      "step": 244
    },
    {
      "epoch": 0.0729723189188183,
      "grad_norm": 0.9538284540176392,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 1.2558,
      "step": 245
    },
    {
      "epoch": 0.07327016511848694,
      "grad_norm": 0.9679785370826721,
      "learning_rate": 4.92e-06,
      "loss": 1.2739,
      "step": 246
    },
    {
      "epoch": 0.07356801131815559,
      "grad_norm": 0.9972037076950073,
      "learning_rate": 4.94e-06,
      "loss": 1.2802,
      "step": 247
    },
    {
      "epoch": 0.07386585751782423,
      "grad_norm": 1.0301584005355835,
      "learning_rate": 4.960000000000001e-06,
      "loss": 1.2703,
      "step": 248
    },
    {
      "epoch": 0.07416370371749288,
      "grad_norm": 0.939154326915741,
      "learning_rate": 4.980000000000001e-06,
      "loss": 1.2743,
      "step": 249
    },
    {
      "epoch": 0.07446154991716153,
      "grad_norm": 0.986122190952301,
      "learning_rate": 5e-06,
      "loss": 1.2915,
      "step": 250
    },
    {
      "epoch": 0.07475939611683018,
      "grad_norm": 0.9798609614372253,
      "learning_rate": 5.02e-06,
      "loss": 1.2653,
      "step": 251
    },
    {
      "epoch": 0.07505724231649882,
      "grad_norm": 0.8938835859298706,
      "learning_rate": 5.04e-06,
      "loss": 1.2665,
      "step": 252
    },
    {
      "epoch": 0.07535508851616747,
      "grad_norm": 0.9504467248916626,
      "learning_rate": 5.060000000000001e-06,
      "loss": 1.2775,
      "step": 253
    },
    {
      "epoch": 0.0756529347158361,
      "grad_norm": 0.9114778637886047,
      "learning_rate": 5.0800000000000005e-06,
      "loss": 1.2694,
      "step": 254
    },
    {
      "epoch": 0.07595078091550475,
      "grad_norm": 0.9156755805015564,
      "learning_rate": 5.1e-06,
      "loss": 1.2819,
      "step": 255
    },
    {
      "epoch": 0.0762486271151734,
      "grad_norm": 0.9251748323440552,
      "learning_rate": 5.12e-06,
      "loss": 1.26,
      "step": 256
    },
    {
      "epoch": 0.07654647331484205,
      "grad_norm": 1.0092203617095947,
      "learning_rate": 5.140000000000001e-06,
      "loss": 1.2902,
      "step": 257
    },
    {
      "epoch": 0.07684431951451069,
      "grad_norm": 0.9251083731651306,
      "learning_rate": 5.1600000000000006e-06,
      "loss": 1.274,
      "step": 258
    },
    {
      "epoch": 0.07714216571417934,
      "grad_norm": 0.9026808142662048,
      "learning_rate": 5.18e-06,
      "loss": 1.2694,
      "step": 259
    },
    {
      "epoch": 0.07744001191384799,
      "grad_norm": 0.9614836573600769,
      "learning_rate": 5.2e-06,
      "loss": 1.2727,
      "step": 260
    },
    {
      "epoch": 0.07773785811351663,
      "grad_norm": 0.941034197807312,
      "learning_rate": 5.220000000000001e-06,
      "loss": 1.2706,
      "step": 261
    },
    {
      "epoch": 0.07803570431318528,
      "grad_norm": 0.9711002111434937,
      "learning_rate": 5.240000000000001e-06,
      "loss": 1.2571,
      "step": 262
    },
    {
      "epoch": 0.07833355051285393,
      "grad_norm": 0.9186269640922546,
      "learning_rate": 5.2600000000000005e-06,
      "loss": 1.2778,
      "step": 263
    },
    {
      "epoch": 0.07863139671252257,
      "grad_norm": 0.9966695308685303,
      "learning_rate": 5.28e-06,
      "loss": 1.2742,
      "step": 264
    },
    {
      "epoch": 0.07892924291219122,
      "grad_norm": 0.9744868874549866,
      "learning_rate": 5.300000000000001e-06,
      "loss": 1.286,
      "step": 265
    },
    {
      "epoch": 0.07922708911185987,
      "grad_norm": 0.9661166071891785,
      "learning_rate": 5.320000000000001e-06,
      "loss": 1.2743,
      "step": 266
    },
    {
      "epoch": 0.07952493531152852,
      "grad_norm": 0.9896745681762695,
      "learning_rate": 5.3400000000000005e-06,
      "loss": 1.2704,
      "step": 267
    },
    {
      "epoch": 0.07982278151119715,
      "grad_norm": 0.9759904146194458,
      "learning_rate": 5.36e-06,
      "loss": 1.2704,
      "step": 268
    },
    {
      "epoch": 0.0801206277108658,
      "grad_norm": 1.0013540983200073,
      "learning_rate": 5.380000000000001e-06,
      "loss": 1.2842,
      "step": 269
    },
    {
      "epoch": 0.08041847391053444,
      "grad_norm": 0.9929929971694946,
      "learning_rate": 5.400000000000001e-06,
      "loss": 1.2787,
      "step": 270
    },
    {
      "epoch": 0.08071632011020309,
      "grad_norm": 0.9690012335777283,
      "learning_rate": 5.420000000000001e-06,
      "loss": 1.2812,
      "step": 271
    },
    {
      "epoch": 0.08101416630987174,
      "grad_norm": 0.9972385168075562,
      "learning_rate": 5.4400000000000004e-06,
      "loss": 1.2643,
      "step": 272
    },
    {
      "epoch": 0.08131201250954039,
      "grad_norm": 0.9709213376045227,
      "learning_rate": 5.460000000000001e-06,
      "loss": 1.2778,
      "step": 273
    },
    {
      "epoch": 0.08160985870920903,
      "grad_norm": 1.0174850225448608,
      "learning_rate": 5.480000000000001e-06,
      "loss": 1.2595,
      "step": 274
    },
    {
      "epoch": 0.08190770490887768,
      "grad_norm": 0.9746271967887878,
      "learning_rate": 5.500000000000001e-06,
      "loss": 1.2694,
      "step": 275
    },
    {
      "epoch": 0.08220555110854633,
      "grad_norm": 0.9794725775718689,
      "learning_rate": 5.5200000000000005e-06,
      "loss": 1.2869,
      "step": 276
    },
    {
      "epoch": 0.08250339730821497,
      "grad_norm": 0.9307675957679749,
      "learning_rate": 5.540000000000001e-06,
      "loss": 1.2636,
      "step": 277
    },
    {
      "epoch": 0.08280124350788362,
      "grad_norm": 0.9829320907592773,
      "learning_rate": 5.560000000000001e-06,
      "loss": 1.2652,
      "step": 278
    },
    {
      "epoch": 0.08309908970755227,
      "grad_norm": 0.9712953567504883,
      "learning_rate": 5.580000000000001e-06,
      "loss": 1.2523,
      "step": 279
    },
    {
      "epoch": 0.08339693590722091,
      "grad_norm": 0.9294509887695312,
      "learning_rate": 5.600000000000001e-06,
      "loss": 1.2743,
      "step": 280
    },
    {
      "epoch": 0.08369478210688955,
      "grad_norm": 1.0231119394302368,
      "learning_rate": 5.620000000000001e-06,
      "loss": 1.263,
      "step": 281
    },
    {
      "epoch": 0.0839926283065582,
      "grad_norm": 0.9840803146362305,
      "learning_rate": 5.64e-06,
      "loss": 1.2721,
      "step": 282
    },
    {
      "epoch": 0.08429047450622684,
      "grad_norm": 0.9343154430389404,
      "learning_rate": 5.66e-06,
      "loss": 1.2581,
      "step": 283
    },
    {
      "epoch": 0.08458832070589549,
      "grad_norm": 0.9271340370178223,
      "learning_rate": 5.68e-06,
      "loss": 1.2677,
      "step": 284
    },
    {
      "epoch": 0.08488616690556414,
      "grad_norm": 0.9336981177330017,
      "learning_rate": 5.7e-06,
      "loss": 1.2446,
      "step": 285
    },
    {
      "epoch": 0.08518401310523278,
      "grad_norm": 0.9839180707931519,
      "learning_rate": 5.72e-06,
      "loss": 1.276,
      "step": 286
    },
    {
      "epoch": 0.08548185930490143,
      "grad_norm": 0.9934493899345398,
      "learning_rate": 5.74e-06,
      "loss": 1.2714,
      "step": 287
    },
    {
      "epoch": 0.08577970550457008,
      "grad_norm": 0.9257272481918335,
      "learning_rate": 5.76e-06,
      "loss": 1.2663,
      "step": 288
    },
    {
      "epoch": 0.08607755170423873,
      "grad_norm": 0.9175654649734497,
      "learning_rate": 5.78e-06,
      "loss": 1.2801,
      "step": 289
    },
    {
      "epoch": 0.08637539790390737,
      "grad_norm": 0.9797008037567139,
      "learning_rate": 5.8e-06,
      "loss": 1.2625,
      "step": 290
    },
    {
      "epoch": 0.08667324410357602,
      "grad_norm": 0.9508514404296875,
      "learning_rate": 5.82e-06,
      "loss": 1.2561,
      "step": 291
    },
    {
      "epoch": 0.08697109030324467,
      "grad_norm": 0.9426538348197937,
      "learning_rate": 5.84e-06,
      "loss": 1.2726,
      "step": 292
    },
    {
      "epoch": 0.08726893650291331,
      "grad_norm": 0.9558387398719788,
      "learning_rate": 5.86e-06,
      "loss": 1.2747,
      "step": 293
    },
    {
      "epoch": 0.08756678270258196,
      "grad_norm": 0.8807733058929443,
      "learning_rate": 5.8800000000000005e-06,
      "loss": 1.2671,
      "step": 294
    },
    {
      "epoch": 0.0878646289022506,
      "grad_norm": 0.9699344635009766,
      "learning_rate": 5.9e-06,
      "loss": 1.2598,
      "step": 295
    },
    {
      "epoch": 0.08816247510191924,
      "grad_norm": 0.945518434047699,
      "learning_rate": 5.92e-06,
      "loss": 1.2565,
      "step": 296
    },
    {
      "epoch": 0.08846032130158789,
      "grad_norm": 0.9229912757873535,
      "learning_rate": 5.94e-06,
      "loss": 1.2459,
      "step": 297
    },
    {
      "epoch": 0.08875816750125654,
      "grad_norm": 0.9146087169647217,
      "learning_rate": 5.9600000000000005e-06,
      "loss": 1.2481,
      "step": 298
    },
    {
      "epoch": 0.08905601370092518,
      "grad_norm": 0.9101441502571106,
      "learning_rate": 5.98e-06,
      "loss": 1.2327,
      "step": 299
    },
    {
      "epoch": 0.08935385990059383,
      "grad_norm": 0.9883213043212891,
      "learning_rate": 6e-06,
      "loss": 1.2616,
      "step": 300
    },
    {
      "epoch": 0.08965170610026248,
      "grad_norm": 0.926801860332489,
      "learning_rate": 6.02e-06,
      "loss": 1.2451,
      "step": 301
    },
    {
      "epoch": 0.08994955229993112,
      "grad_norm": 0.9054859280586243,
      "learning_rate": 6.040000000000001e-06,
      "loss": 1.2637,
      "step": 302
    },
    {
      "epoch": 0.09024739849959977,
      "grad_norm": 0.988732099533081,
      "learning_rate": 6.0600000000000004e-06,
      "loss": 1.2746,
      "step": 303
    },
    {
      "epoch": 0.09054524469926842,
      "grad_norm": 0.9347218871116638,
      "learning_rate": 6.08e-06,
      "loss": 1.2617,
      "step": 304
    },
    {
      "epoch": 0.09084309089893707,
      "grad_norm": 0.9000351428985596,
      "learning_rate": 6.1e-06,
      "loss": 1.2617,
      "step": 305
    },
    {
      "epoch": 0.09114093709860571,
      "grad_norm": 0.9307082891464233,
      "learning_rate": 6.120000000000001e-06,
      "loss": 1.2659,
      "step": 306
    },
    {
      "epoch": 0.09143878329827436,
      "grad_norm": 0.9266918897628784,
      "learning_rate": 6.1400000000000005e-06,
      "loss": 1.246,
      "step": 307
    },
    {
      "epoch": 0.091736629497943,
      "grad_norm": 0.9378978610038757,
      "learning_rate": 6.16e-06,
      "loss": 1.2666,
      "step": 308
    },
    {
      "epoch": 0.09203447569761164,
      "grad_norm": 0.9447678923606873,
      "learning_rate": 6.18e-06,
      "loss": 1.2692,
      "step": 309
    },
    {
      "epoch": 0.09233232189728029,
      "grad_norm": 0.9284104704856873,
      "learning_rate": 6.200000000000001e-06,
      "loss": 1.2681,
      "step": 310
    },
    {
      "epoch": 0.09263016809694893,
      "grad_norm": 0.9121084809303284,
      "learning_rate": 6.220000000000001e-06,
      "loss": 1.2537,
      "step": 311
    },
    {
      "epoch": 0.09292801429661758,
      "grad_norm": 0.9484390020370483,
      "learning_rate": 6.24e-06,
      "loss": 1.2807,
      "step": 312
    },
    {
      "epoch": 0.09322586049628623,
      "grad_norm": 0.9707821011543274,
      "learning_rate": 6.26e-06,
      "loss": 1.25,
      "step": 313
    },
    {
      "epoch": 0.09352370669595488,
      "grad_norm": 0.9659841656684875,
      "learning_rate": 6.280000000000001e-06,
      "loss": 1.2701,
      "step": 314
    },
    {
      "epoch": 0.09382155289562352,
      "grad_norm": 0.9175627827644348,
      "learning_rate": 6.300000000000001e-06,
      "loss": 1.2422,
      "step": 315
    },
    {
      "epoch": 0.09411939909529217,
      "grad_norm": 0.9576548337936401,
      "learning_rate": 6.3200000000000005e-06,
      "loss": 1.2389,
      "step": 316
    },
    {
      "epoch": 0.09441724529496082,
      "grad_norm": 0.956176221370697,
      "learning_rate": 6.34e-06,
      "loss": 1.2395,
      "step": 317
    },
    {
      "epoch": 0.09471509149462946,
      "grad_norm": 0.9568424820899963,
      "learning_rate": 6.360000000000001e-06,
      "loss": 1.2624,
      "step": 318
    },
    {
      "epoch": 0.09501293769429811,
      "grad_norm": 0.9627560973167419,
      "learning_rate": 6.380000000000001e-06,
      "loss": 1.2332,
      "step": 319
    },
    {
      "epoch": 0.09531078389396676,
      "grad_norm": 0.9709836840629578,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 1.2559,
      "step": 320
    },
    {
      "epoch": 0.0956086300936354,
      "grad_norm": 0.9256501197814941,
      "learning_rate": 6.42e-06,
      "loss": 1.2667,
      "step": 321
    },
    {
      "epoch": 0.09590647629330404,
      "grad_norm": 0.9080420732498169,
      "learning_rate": 6.440000000000001e-06,
      "loss": 1.2475,
      "step": 322
    },
    {
      "epoch": 0.09620432249297269,
      "grad_norm": 0.9155151844024658,
      "learning_rate": 6.460000000000001e-06,
      "loss": 1.2549,
      "step": 323
    },
    {
      "epoch": 0.09650216869264133,
      "grad_norm": 0.9808725714683533,
      "learning_rate": 6.480000000000001e-06,
      "loss": 1.2595,
      "step": 324
    },
    {
      "epoch": 0.09680001489230998,
      "grad_norm": 0.9659640192985535,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 1.2752,
      "step": 325
    },
    {
      "epoch": 0.09709786109197863,
      "grad_norm": 0.8983002305030823,
      "learning_rate": 6.520000000000001e-06,
      "loss": 1.2651,
      "step": 326
    },
    {
      "epoch": 0.09739570729164727,
      "grad_norm": 0.9381516575813293,
      "learning_rate": 6.540000000000001e-06,
      "loss": 1.2489,
      "step": 327
    },
    {
      "epoch": 0.09769355349131592,
      "grad_norm": 0.9322483539581299,
      "learning_rate": 6.560000000000001e-06,
      "loss": 1.2579,
      "step": 328
    },
    {
      "epoch": 0.09799139969098457,
      "grad_norm": 0.929489254951477,
      "learning_rate": 6.5800000000000005e-06,
      "loss": 1.2563,
      "step": 329
    },
    {
      "epoch": 0.09828924589065322,
      "grad_norm": 0.9390615224838257,
      "learning_rate": 6.600000000000001e-06,
      "loss": 1.2511,
      "step": 330
    },
    {
      "epoch": 0.09858709209032186,
      "grad_norm": 0.9023354053497314,
      "learning_rate": 6.620000000000001e-06,
      "loss": 1.2622,
      "step": 331
    },
    {
      "epoch": 0.09888493828999051,
      "grad_norm": 0.9320018887519836,
      "learning_rate": 6.640000000000001e-06,
      "loss": 1.2666,
      "step": 332
    },
    {
      "epoch": 0.09918278448965916,
      "grad_norm": 0.9349921941757202,
      "learning_rate": 6.660000000000001e-06,
      "loss": 1.262,
      "step": 333
    },
    {
      "epoch": 0.0994806306893278,
      "grad_norm": 0.9168747067451477,
      "learning_rate": 6.680000000000001e-06,
      "loss": 1.2601,
      "step": 334
    },
    {
      "epoch": 0.09977847688899645,
      "grad_norm": 0.9716922044754028,
      "learning_rate": 6.700000000000001e-06,
      "loss": 1.2456,
      "step": 335
    },
    {
      "epoch": 0.10007632308866508,
      "grad_norm": 0.9088549017906189,
      "learning_rate": 6.720000000000001e-06,
      "loss": 1.2645,
      "step": 336
    },
    {
      "epoch": 0.10037416928833373,
      "grad_norm": 0.8880696892738342,
      "learning_rate": 6.740000000000001e-06,
      "loss": 1.2407,
      "step": 337
    },
    {
      "epoch": 0.10067201548800238,
      "grad_norm": 1.0113357305526733,
      "learning_rate": 6.760000000000001e-06,
      "loss": 1.2499,
      "step": 338
    },
    {
      "epoch": 0.10096986168767103,
      "grad_norm": 0.9354126453399658,
      "learning_rate": 6.780000000000001e-06,
      "loss": 1.2565,
      "step": 339
    },
    {
      "epoch": 0.10126770788733967,
      "grad_norm": 0.9319290518760681,
      "learning_rate": 6.800000000000001e-06,
      "loss": 1.2594,
      "step": 340
    },
    {
      "epoch": 0.10156555408700832,
      "grad_norm": 0.956741988658905,
      "learning_rate": 6.820000000000001e-06,
      "loss": 1.241,
      "step": 341
    },
    {
      "epoch": 0.10186340028667697,
      "grad_norm": 0.9170883297920227,
      "learning_rate": 6.8400000000000014e-06,
      "loss": 1.252,
      "step": 342
    },
    {
      "epoch": 0.10216124648634561,
      "grad_norm": 0.9752845764160156,
      "learning_rate": 6.860000000000001e-06,
      "loss": 1.2696,
      "step": 343
    },
    {
      "epoch": 0.10245909268601426,
      "grad_norm": 0.9555333256721497,
      "learning_rate": 6.88e-06,
      "loss": 1.2482,
      "step": 344
    },
    {
      "epoch": 0.10275693888568291,
      "grad_norm": 0.9118161201477051,
      "learning_rate": 6.9e-06,
      "loss": 1.2531,
      "step": 345
    },
    {
      "epoch": 0.10305478508535156,
      "grad_norm": 0.92820805311203,
      "learning_rate": 6.92e-06,
      "loss": 1.246,
      "step": 346
    },
    {
      "epoch": 0.1033526312850202,
      "grad_norm": 0.8858067393302917,
      "learning_rate": 6.9400000000000005e-06,
      "loss": 1.256,
      "step": 347
    },
    {
      "epoch": 0.10365047748468885,
      "grad_norm": 0.9146501421928406,
      "learning_rate": 6.96e-06,
      "loss": 1.2507,
      "step": 348
    },
    {
      "epoch": 0.1039483236843575,
      "grad_norm": 0.9195312261581421,
      "learning_rate": 6.98e-06,
      "loss": 1.2433,
      "step": 349
    },
    {
      "epoch": 0.10424616988402613,
      "grad_norm": 0.9402062892913818,
      "learning_rate": 7e-06,
      "loss": 1.2651,
      "step": 350
    },
    {
      "epoch": 0.10454401608369478,
      "grad_norm": 0.9383959174156189,
      "learning_rate": 7.0200000000000006e-06,
      "loss": 1.2402,
      "step": 351
    },
    {
      "epoch": 0.10484186228336342,
      "grad_norm": 0.9008394479751587,
      "learning_rate": 7.04e-06,
      "loss": 1.2575,
      "step": 352
    },
    {
      "epoch": 0.10513970848303207,
      "grad_norm": 0.944844663143158,
      "learning_rate": 7.06e-06,
      "loss": 1.2612,
      "step": 353
    },
    {
      "epoch": 0.10543755468270072,
      "grad_norm": 0.9473034143447876,
      "learning_rate": 7.08e-06,
      "loss": 1.2561,
      "step": 354
    },
    {
      "epoch": 0.10573540088236937,
      "grad_norm": 0.9180167317390442,
      "learning_rate": 7.100000000000001e-06,
      "loss": 1.242,
      "step": 355
    },
    {
      "epoch": 0.10603324708203801,
      "grad_norm": 0.9363252520561218,
      "learning_rate": 7.1200000000000004e-06,
      "loss": 1.2331,
      "step": 356
    },
    {
      "epoch": 0.10633109328170666,
      "grad_norm": 0.968022346496582,
      "learning_rate": 7.14e-06,
      "loss": 1.2444,
      "step": 357
    },
    {
      "epoch": 0.1066289394813753,
      "grad_norm": 0.9318762421607971,
      "learning_rate": 7.16e-06,
      "loss": 1.2425,
      "step": 358
    },
    {
      "epoch": 0.10692678568104395,
      "grad_norm": 0.966212809085846,
      "learning_rate": 7.180000000000001e-06,
      "loss": 1.2535,
      "step": 359
    },
    {
      "epoch": 0.1072246318807126,
      "grad_norm": 0.9267631769180298,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 1.2589,
      "step": 360
    },
    {
      "epoch": 0.10752247808038125,
      "grad_norm": 0.9328450560569763,
      "learning_rate": 7.22e-06,
      "loss": 1.2675,
      "step": 361
    },
    {
      "epoch": 0.1078203242800499,
      "grad_norm": 0.9472862482070923,
      "learning_rate": 7.24e-06,
      "loss": 1.2527,
      "step": 362
    },
    {
      "epoch": 0.10811817047971854,
      "grad_norm": 0.9217740893363953,
      "learning_rate": 7.260000000000001e-06,
      "loss": 1.2511,
      "step": 363
    },
    {
      "epoch": 0.10841601667938718,
      "grad_norm": 0.9169412851333618,
      "learning_rate": 7.280000000000001e-06,
      "loss": 1.2468,
      "step": 364
    },
    {
      "epoch": 0.10871386287905582,
      "grad_norm": 0.9022817611694336,
      "learning_rate": 7.3e-06,
      "loss": 1.2321,
      "step": 365
    },
    {
      "epoch": 0.10901170907872447,
      "grad_norm": 0.9114884734153748,
      "learning_rate": 7.32e-06,
      "loss": 1.2546,
      "step": 366
    },
    {
      "epoch": 0.10930955527839312,
      "grad_norm": 0.9416398406028748,
      "learning_rate": 7.340000000000001e-06,
      "loss": 1.2461,
      "step": 367
    },
    {
      "epoch": 0.10960740147806176,
      "grad_norm": 0.8965981006622314,
      "learning_rate": 7.360000000000001e-06,
      "loss": 1.2541,
      "step": 368
    },
    {
      "epoch": 0.10990524767773041,
      "grad_norm": 0.881142795085907,
      "learning_rate": 7.3800000000000005e-06,
      "loss": 1.2195,
      "step": 369
    },
    {
      "epoch": 0.11020309387739906,
      "grad_norm": 0.963137686252594,
      "learning_rate": 7.4e-06,
      "loss": 1.248,
      "step": 370
    },
    {
      "epoch": 0.1105009400770677,
      "grad_norm": 0.8961619138717651,
      "learning_rate": 7.420000000000001e-06,
      "loss": 1.2432,
      "step": 371
    },
    {
      "epoch": 0.11079878627673635,
      "grad_norm": 0.8863562941551208,
      "learning_rate": 7.440000000000001e-06,
      "loss": 1.2464,
      "step": 372
    },
    {
      "epoch": 0.111096632476405,
      "grad_norm": 0.91034334897995,
      "learning_rate": 7.4600000000000006e-06,
      "loss": 1.2449,
      "step": 373
    },
    {
      "epoch": 0.11139447867607365,
      "grad_norm": 0.9478059411048889,
      "learning_rate": 7.48e-06,
      "loss": 1.248,
      "step": 374
    },
    {
      "epoch": 0.1116923248757423,
      "grad_norm": 0.918700098991394,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.2533,
      "step": 375
    },
    {
      "epoch": 0.11199017107541094,
      "grad_norm": 0.9177455306053162,
      "learning_rate": 7.520000000000001e-06,
      "loss": 1.2308,
      "step": 376
    },
    {
      "epoch": 0.11228801727507957,
      "grad_norm": 0.9344789385795593,
      "learning_rate": 7.540000000000001e-06,
      "loss": 1.2285,
      "step": 377
    },
    {
      "epoch": 0.11258586347474822,
      "grad_norm": 0.8945629596710205,
      "learning_rate": 7.5600000000000005e-06,
      "loss": 1.2308,
      "step": 378
    },
    {
      "epoch": 0.11288370967441687,
      "grad_norm": 0.9492330551147461,
      "learning_rate": 7.58e-06,
      "loss": 1.2559,
      "step": 379
    },
    {
      "epoch": 0.11318155587408552,
      "grad_norm": 0.9249116778373718,
      "learning_rate": 7.600000000000001e-06,
      "loss": 1.2291,
      "step": 380
    },
    {
      "epoch": 0.11347940207375416,
      "grad_norm": 0.9113664627075195,
      "learning_rate": 7.620000000000001e-06,
      "loss": 1.2447,
      "step": 381
    },
    {
      "epoch": 0.11377724827342281,
      "grad_norm": 0.9052017331123352,
      "learning_rate": 7.640000000000001e-06,
      "loss": 1.2457,
      "step": 382
    },
    {
      "epoch": 0.11407509447309146,
      "grad_norm": 0.9360484480857849,
      "learning_rate": 7.660000000000001e-06,
      "loss": 1.2443,
      "step": 383
    },
    {
      "epoch": 0.1143729406727601,
      "grad_norm": 0.946617841720581,
      "learning_rate": 7.680000000000001e-06,
      "loss": 1.2328,
      "step": 384
    },
    {
      "epoch": 0.11467078687242875,
      "grad_norm": 0.952668309211731,
      "learning_rate": 7.7e-06,
      "loss": 1.2442,
      "step": 385
    },
    {
      "epoch": 0.1149686330720974,
      "grad_norm": 0.9053682088851929,
      "learning_rate": 7.72e-06,
      "loss": 1.2241,
      "step": 386
    },
    {
      "epoch": 0.11526647927176605,
      "grad_norm": 0.9697679877281189,
      "learning_rate": 7.74e-06,
      "loss": 1.2441,
      "step": 387
    },
    {
      "epoch": 0.11556432547143469,
      "grad_norm": 0.9620223045349121,
      "learning_rate": 7.76e-06,
      "loss": 1.2286,
      "step": 388
    },
    {
      "epoch": 0.11586217167110334,
      "grad_norm": 0.9781268835067749,
      "learning_rate": 7.78e-06,
      "loss": 1.2341,
      "step": 389
    },
    {
      "epoch": 0.11616001787077199,
      "grad_norm": 0.9378364682197571,
      "learning_rate": 7.800000000000002e-06,
      "loss": 1.2472,
      "step": 390
    },
    {
      "epoch": 0.11645786407044062,
      "grad_norm": 1.0310090780258179,
      "learning_rate": 7.820000000000001e-06,
      "loss": 1.2639,
      "step": 391
    },
    {
      "epoch": 0.11675571027010927,
      "grad_norm": 0.9500730037689209,
      "learning_rate": 7.840000000000001e-06,
      "loss": 1.2305,
      "step": 392
    },
    {
      "epoch": 0.11705355646977791,
      "grad_norm": 0.9176587462425232,
      "learning_rate": 7.860000000000001e-06,
      "loss": 1.2445,
      "step": 393
    },
    {
      "epoch": 0.11735140266944656,
      "grad_norm": 0.9089423418045044,
      "learning_rate": 7.88e-06,
      "loss": 1.224,
      "step": 394
    },
    {
      "epoch": 0.11764924886911521,
      "grad_norm": 0.98666912317276,
      "learning_rate": 7.9e-06,
      "loss": 1.2416,
      "step": 395
    },
    {
      "epoch": 0.11794709506878386,
      "grad_norm": 0.8849412202835083,
      "learning_rate": 7.92e-06,
      "loss": 1.233,
      "step": 396
    },
    {
      "epoch": 0.1182449412684525,
      "grad_norm": 0.9893229603767395,
      "learning_rate": 7.94e-06,
      "loss": 1.2516,
      "step": 397
    },
    {
      "epoch": 0.11854278746812115,
      "grad_norm": 1.0308648347854614,
      "learning_rate": 7.960000000000002e-06,
      "loss": 1.223,
      "step": 398
    },
    {
      "epoch": 0.1188406336677898,
      "grad_norm": 0.9404271841049194,
      "learning_rate": 7.980000000000002e-06,
      "loss": 1.2491,
      "step": 399
    },
    {
      "epoch": 0.11913847986745844,
      "grad_norm": 0.9053499102592468,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.2326,
      "step": 400
    },
    {
      "epoch": 0.11943632606712709,
      "grad_norm": 0.9721001386642456,
      "learning_rate": 8.020000000000001e-06,
      "loss": 1.2279,
      "step": 401
    },
    {
      "epoch": 0.11973417226679574,
      "grad_norm": 0.9912161827087402,
      "learning_rate": 8.040000000000001e-06,
      "loss": 1.2346,
      "step": 402
    },
    {
      "epoch": 0.12003201846646439,
      "grad_norm": 0.8813495635986328,
      "learning_rate": 8.06e-06,
      "loss": 1.2436,
      "step": 403
    },
    {
      "epoch": 0.12032986466613303,
      "grad_norm": 0.9342209696769714,
      "learning_rate": 8.08e-06,
      "loss": 1.2542,
      "step": 404
    },
    {
      "epoch": 0.12062771086580167,
      "grad_norm": 0.9780386090278625,
      "learning_rate": 8.1e-06,
      "loss": 1.2651,
      "step": 405
    },
    {
      "epoch": 0.12092555706547031,
      "grad_norm": 0.8965239524841309,
      "learning_rate": 8.120000000000002e-06,
      "loss": 1.2524,
      "step": 406
    },
    {
      "epoch": 0.12122340326513896,
      "grad_norm": 0.9501399993896484,
      "learning_rate": 8.14e-06,
      "loss": 1.2457,
      "step": 407
    },
    {
      "epoch": 0.1215212494648076,
      "grad_norm": 0.9022490382194519,
      "learning_rate": 8.16e-06,
      "loss": 1.2567,
      "step": 408
    },
    {
      "epoch": 0.12181909566447625,
      "grad_norm": 0.9594461917877197,
      "learning_rate": 8.18e-06,
      "loss": 1.2521,
      "step": 409
    },
    {
      "epoch": 0.1221169418641449,
      "grad_norm": 0.901566207408905,
      "learning_rate": 8.2e-06,
      "loss": 1.2402,
      "step": 410
    },
    {
      "epoch": 0.12241478806381355,
      "grad_norm": 0.8931708335876465,
      "learning_rate": 8.220000000000001e-06,
      "loss": 1.241,
      "step": 411
    },
    {
      "epoch": 0.1227126342634822,
      "grad_norm": 0.9423035979270935,
      "learning_rate": 8.24e-06,
      "loss": 1.2335,
      "step": 412
    },
    {
      "epoch": 0.12301048046315084,
      "grad_norm": 0.8927134871482849,
      "learning_rate": 8.26e-06,
      "loss": 1.2393,
      "step": 413
    },
    {
      "epoch": 0.12330832666281949,
      "grad_norm": 0.9339447021484375,
      "learning_rate": 8.28e-06,
      "loss": 1.2618,
      "step": 414
    },
    {
      "epoch": 0.12360617286248814,
      "grad_norm": 0.943602979183197,
      "learning_rate": 8.3e-06,
      "loss": 1.2584,
      "step": 415
    },
    {
      "epoch": 0.12390401906215678,
      "grad_norm": 0.9646636843681335,
      "learning_rate": 8.32e-06,
      "loss": 1.2341,
      "step": 416
    },
    {
      "epoch": 0.12420186526182543,
      "grad_norm": 0.9566707611083984,
      "learning_rate": 8.34e-06,
      "loss": 1.2495,
      "step": 417
    },
    {
      "epoch": 0.12449971146149406,
      "grad_norm": 0.9131056070327759,
      "learning_rate": 8.36e-06,
      "loss": 1.2576,
      "step": 418
    },
    {
      "epoch": 0.12479755766116271,
      "grad_norm": 0.9560261368751526,
      "learning_rate": 8.380000000000001e-06,
      "loss": 1.2542,
      "step": 419
    },
    {
      "epoch": 0.12509540386083137,
      "grad_norm": 0.9510844349861145,
      "learning_rate": 8.400000000000001e-06,
      "loss": 1.2334,
      "step": 420
    },
    {
      "epoch": 0.1253932500605,
      "grad_norm": 0.9644795060157776,
      "learning_rate": 8.42e-06,
      "loss": 1.2516,
      "step": 421
    },
    {
      "epoch": 0.12569109626016867,
      "grad_norm": 0.9237042665481567,
      "learning_rate": 8.44e-06,
      "loss": 1.2151,
      "step": 422
    },
    {
      "epoch": 0.1259889424598373,
      "grad_norm": 0.931945264339447,
      "learning_rate": 8.46e-06,
      "loss": 1.2485,
      "step": 423
    },
    {
      "epoch": 0.12628678865950596,
      "grad_norm": 0.9151108860969543,
      "learning_rate": 8.48e-06,
      "loss": 1.2413,
      "step": 424
    },
    {
      "epoch": 0.1265846348591746,
      "grad_norm": 0.9182891845703125,
      "learning_rate": 8.5e-06,
      "loss": 1.2543,
      "step": 425
    },
    {
      "epoch": 0.12688248105884323,
      "grad_norm": 0.90055912733078,
      "learning_rate": 8.52e-06,
      "loss": 1.2382,
      "step": 426
    },
    {
      "epoch": 0.1271803272585119,
      "grad_norm": 0.9547976851463318,
      "learning_rate": 8.540000000000001e-06,
      "loss": 1.2522,
      "step": 427
    },
    {
      "epoch": 0.12747817345818052,
      "grad_norm": 0.8878069519996643,
      "learning_rate": 8.560000000000001e-06,
      "loss": 1.2256,
      "step": 428
    },
    {
      "epoch": 0.12777601965784918,
      "grad_norm": 0.9147696495056152,
      "learning_rate": 8.580000000000001e-06,
      "loss": 1.2513,
      "step": 429
    },
    {
      "epoch": 0.12807386585751782,
      "grad_norm": 0.9284038543701172,
      "learning_rate": 8.6e-06,
      "loss": 1.2393,
      "step": 430
    },
    {
      "epoch": 0.12837171205718648,
      "grad_norm": 0.9363827109336853,
      "learning_rate": 8.62e-06,
      "loss": 1.2234,
      "step": 431
    },
    {
      "epoch": 0.1286695582568551,
      "grad_norm": 0.88382488489151,
      "learning_rate": 8.64e-06,
      "loss": 1.2184,
      "step": 432
    },
    {
      "epoch": 0.12896740445652377,
      "grad_norm": 0.8739042282104492,
      "learning_rate": 8.66e-06,
      "loss": 1.2336,
      "step": 433
    },
    {
      "epoch": 0.1292652506561924,
      "grad_norm": 0.9750383496284485,
      "learning_rate": 8.68e-06,
      "loss": 1.2357,
      "step": 434
    },
    {
      "epoch": 0.12956309685586107,
      "grad_norm": 0.9510998725891113,
      "learning_rate": 8.700000000000001e-06,
      "loss": 1.2304,
      "step": 435
    },
    {
      "epoch": 0.1298609430555297,
      "grad_norm": 0.933433473110199,
      "learning_rate": 8.720000000000001e-06,
      "loss": 1.2492,
      "step": 436
    },
    {
      "epoch": 0.13015878925519836,
      "grad_norm": 0.9198402762413025,
      "learning_rate": 8.740000000000001e-06,
      "loss": 1.2276,
      "step": 437
    },
    {
      "epoch": 0.130456635454867,
      "grad_norm": 0.9637134671211243,
      "learning_rate": 8.76e-06,
      "loss": 1.2343,
      "step": 438
    },
    {
      "epoch": 0.13075448165453563,
      "grad_norm": 0.9523952603340149,
      "learning_rate": 8.78e-06,
      "loss": 1.2199,
      "step": 439
    },
    {
      "epoch": 0.1310523278542043,
      "grad_norm": 0.9594414234161377,
      "learning_rate": 8.8e-06,
      "loss": 1.2301,
      "step": 440
    },
    {
      "epoch": 0.13135017405387292,
      "grad_norm": 0.9743456840515137,
      "learning_rate": 8.82e-06,
      "loss": 1.244,
      "step": 441
    },
    {
      "epoch": 0.13164802025354158,
      "grad_norm": 0.9492051005363464,
      "learning_rate": 8.84e-06,
      "loss": 1.2271,
      "step": 442
    },
    {
      "epoch": 0.13194586645321021,
      "grad_norm": 0.9626035094261169,
      "learning_rate": 8.860000000000002e-06,
      "loss": 1.2257,
      "step": 443
    },
    {
      "epoch": 0.13224371265287888,
      "grad_norm": 0.9121590852737427,
      "learning_rate": 8.880000000000001e-06,
      "loss": 1.2394,
      "step": 444
    },
    {
      "epoch": 0.1325415588525475,
      "grad_norm": 1.0002115964889526,
      "learning_rate": 8.900000000000001e-06,
      "loss": 1.2358,
      "step": 445
    },
    {
      "epoch": 0.13283940505221617,
      "grad_norm": 0.878296971321106,
      "learning_rate": 8.920000000000001e-06,
      "loss": 1.2216,
      "step": 446
    },
    {
      "epoch": 0.1331372512518848,
      "grad_norm": 0.9509201049804688,
      "learning_rate": 8.94e-06,
      "loss": 1.2313,
      "step": 447
    },
    {
      "epoch": 0.13343509745155346,
      "grad_norm": 0.9199770092964172,
      "learning_rate": 8.96e-06,
      "loss": 1.2399,
      "step": 448
    },
    {
      "epoch": 0.1337329436512221,
      "grad_norm": 0.8914684057235718,
      "learning_rate": 8.98e-06,
      "loss": 1.2495,
      "step": 449
    },
    {
      "epoch": 0.13403078985089076,
      "grad_norm": 0.9377488493919373,
      "learning_rate": 9e-06,
      "loss": 1.2335,
      "step": 450
    },
    {
      "epoch": 0.1343286360505594,
      "grad_norm": 0.9135779738426208,
      "learning_rate": 9.020000000000002e-06,
      "loss": 1.2136,
      "step": 451
    },
    {
      "epoch": 0.13462648225022805,
      "grad_norm": 0.9123891592025757,
      "learning_rate": 9.040000000000002e-06,
      "loss": 1.2425,
      "step": 452
    },
    {
      "epoch": 0.13492432844989669,
      "grad_norm": 0.8983786106109619,
      "learning_rate": 9.060000000000001e-06,
      "loss": 1.2306,
      "step": 453
    },
    {
      "epoch": 0.13522217464956532,
      "grad_norm": 0.9097176790237427,
      "learning_rate": 9.080000000000001e-06,
      "loss": 1.2294,
      "step": 454
    },
    {
      "epoch": 0.13552002084923398,
      "grad_norm": 0.9312183856964111,
      "learning_rate": 9.100000000000001e-06,
      "loss": 1.2459,
      "step": 455
    },
    {
      "epoch": 0.1358178670489026,
      "grad_norm": 0.8798116445541382,
      "learning_rate": 9.12e-06,
      "loss": 1.2357,
      "step": 456
    },
    {
      "epoch": 0.13611571324857127,
      "grad_norm": 0.910600483417511,
      "learning_rate": 9.14e-06,
      "loss": 1.2499,
      "step": 457
    },
    {
      "epoch": 0.1364135594482399,
      "grad_norm": 0.9578474760055542,
      "learning_rate": 9.16e-06,
      "loss": 1.2568,
      "step": 458
    },
    {
      "epoch": 0.13671140564790857,
      "grad_norm": 0.8519591093063354,
      "learning_rate": 9.180000000000002e-06,
      "loss": 1.2455,
      "step": 459
    },
    {
      "epoch": 0.1370092518475772,
      "grad_norm": 0.8745837807655334,
      "learning_rate": 9.200000000000002e-06,
      "loss": 1.2394,
      "step": 460
    },
    {
      "epoch": 0.13730709804724586,
      "grad_norm": 0.9189319014549255,
      "learning_rate": 9.220000000000002e-06,
      "loss": 1.2142,
      "step": 461
    },
    {
      "epoch": 0.1376049442469145,
      "grad_norm": 0.897632360458374,
      "learning_rate": 9.240000000000001e-06,
      "loss": 1.2125,
      "step": 462
    },
    {
      "epoch": 0.13790279044658316,
      "grad_norm": 0.8797314167022705,
      "learning_rate": 9.260000000000001e-06,
      "loss": 1.25,
      "step": 463
    },
    {
      "epoch": 0.1382006366462518,
      "grad_norm": 0.8879082798957825,
      "learning_rate": 9.280000000000001e-06,
      "loss": 1.2202,
      "step": 464
    },
    {
      "epoch": 0.13849848284592045,
      "grad_norm": 0.9235281944274902,
      "learning_rate": 9.3e-06,
      "loss": 1.2402,
      "step": 465
    },
    {
      "epoch": 0.13879632904558908,
      "grad_norm": 0.868775486946106,
      "learning_rate": 9.32e-06,
      "loss": 1.2432,
      "step": 466
    },
    {
      "epoch": 0.13909417524525772,
      "grad_norm": 0.8809604048728943,
      "learning_rate": 9.340000000000002e-06,
      "loss": 1.2279,
      "step": 467
    },
    {
      "epoch": 0.13939202144492638,
      "grad_norm": 0.8650441765785217,
      "learning_rate": 9.360000000000002e-06,
      "loss": 1.2218,
      "step": 468
    },
    {
      "epoch": 0.139689867644595,
      "grad_norm": 0.8861599564552307,
      "learning_rate": 9.38e-06,
      "loss": 1.2386,
      "step": 469
    },
    {
      "epoch": 0.13998771384426367,
      "grad_norm": 0.9290902018547058,
      "learning_rate": 9.4e-06,
      "loss": 1.2456,
      "step": 470
    },
    {
      "epoch": 0.1402855600439323,
      "grad_norm": 0.8369218707084656,
      "learning_rate": 9.42e-06,
      "loss": 1.2075,
      "step": 471
    },
    {
      "epoch": 0.14058340624360097,
      "grad_norm": 0.8635169863700867,
      "learning_rate": 9.440000000000001e-06,
      "loss": 1.2399,
      "step": 472
    },
    {
      "epoch": 0.1408812524432696,
      "grad_norm": 0.8932274580001831,
      "learning_rate": 9.460000000000001e-06,
      "loss": 1.2307,
      "step": 473
    },
    {
      "epoch": 0.14117909864293826,
      "grad_norm": 0.8765782713890076,
      "learning_rate": 9.48e-06,
      "loss": 1.2312,
      "step": 474
    },
    {
      "epoch": 0.1414769448426069,
      "grad_norm": 0.845764696598053,
      "learning_rate": 9.5e-06,
      "loss": 1.2262,
      "step": 475
    },
    {
      "epoch": 0.14177479104227556,
      "grad_norm": 0.8652346730232239,
      "learning_rate": 9.52e-06,
      "loss": 1.2381,
      "step": 476
    },
    {
      "epoch": 0.1420726372419442,
      "grad_norm": 0.9064956903457642,
      "learning_rate": 9.54e-06,
      "loss": 1.235,
      "step": 477
    },
    {
      "epoch": 0.14237048344161285,
      "grad_norm": 0.8930155634880066,
      "learning_rate": 9.56e-06,
      "loss": 1.2369,
      "step": 478
    },
    {
      "epoch": 0.14266832964128148,
      "grad_norm": 0.9362769722938538,
      "learning_rate": 9.58e-06,
      "loss": 1.225,
      "step": 479
    },
    {
      "epoch": 0.14296617584095012,
      "grad_norm": 0.9055190682411194,
      "learning_rate": 9.600000000000001e-06,
      "loss": 1.2132,
      "step": 480
    },
    {
      "epoch": 0.14326402204061878,
      "grad_norm": 0.8966508507728577,
      "learning_rate": 9.620000000000001e-06,
      "loss": 1.2278,
      "step": 481
    },
    {
      "epoch": 0.1435618682402874,
      "grad_norm": 0.9678097367286682,
      "learning_rate": 9.640000000000001e-06,
      "loss": 1.241,
      "step": 482
    },
    {
      "epoch": 0.14385971443995607,
      "grad_norm": 0.8886237740516663,
      "learning_rate": 9.66e-06,
      "loss": 1.229,
      "step": 483
    },
    {
      "epoch": 0.1441575606396247,
      "grad_norm": 1.013090968132019,
      "learning_rate": 9.68e-06,
      "loss": 1.2374,
      "step": 484
    },
    {
      "epoch": 0.14445540683929337,
      "grad_norm": 0.9113328456878662,
      "learning_rate": 9.7e-06,
      "loss": 1.2393,
      "step": 485
    },
    {
      "epoch": 0.144753253038962,
      "grad_norm": 0.957122802734375,
      "learning_rate": 9.72e-06,
      "loss": 1.2397,
      "step": 486
    },
    {
      "epoch": 0.14505109923863066,
      "grad_norm": 0.9432862401008606,
      "learning_rate": 9.74e-06,
      "loss": 1.2341,
      "step": 487
    },
    {
      "epoch": 0.1453489454382993,
      "grad_norm": 0.9090948700904846,
      "learning_rate": 9.760000000000001e-06,
      "loss": 1.2429,
      "step": 488
    },
    {
      "epoch": 0.14564679163796795,
      "grad_norm": 0.9642869830131531,
      "learning_rate": 9.780000000000001e-06,
      "loss": 1.2328,
      "step": 489
    },
    {
      "epoch": 0.1459446378376366,
      "grad_norm": 0.9003296494483948,
      "learning_rate": 9.800000000000001e-06,
      "loss": 1.2502,
      "step": 490
    },
    {
      "epoch": 0.14624248403730525,
      "grad_norm": 0.8932400345802307,
      "learning_rate": 9.820000000000001e-06,
      "loss": 1.2092,
      "step": 491
    },
    {
      "epoch": 0.14654033023697388,
      "grad_norm": 0.8575664758682251,
      "learning_rate": 9.84e-06,
      "loss": 1.2112,
      "step": 492
    },
    {
      "epoch": 0.14683817643664254,
      "grad_norm": 0.9423137307167053,
      "learning_rate": 9.86e-06,
      "loss": 1.2242,
      "step": 493
    },
    {
      "epoch": 0.14713602263631118,
      "grad_norm": 0.855263352394104,
      "learning_rate": 9.88e-06,
      "loss": 1.248,
      "step": 494
    },
    {
      "epoch": 0.1474338688359798,
      "grad_norm": 0.8522682189941406,
      "learning_rate": 9.9e-06,
      "loss": 1.2509,
      "step": 495
    },
    {
      "epoch": 0.14773171503564847,
      "grad_norm": 0.8839662671089172,
      "learning_rate": 9.920000000000002e-06,
      "loss": 1.2207,
      "step": 496
    },
    {
      "epoch": 0.1480295612353171,
      "grad_norm": 0.907123327255249,
      "learning_rate": 9.940000000000001e-06,
      "loss": 1.2259,
      "step": 497
    },
    {
      "epoch": 0.14832740743498576,
      "grad_norm": 0.9126421809196472,
      "learning_rate": 9.960000000000001e-06,
      "loss": 1.221,
      "step": 498
    },
    {
      "epoch": 0.1486252536346544,
      "grad_norm": 0.8731040358543396,
      "learning_rate": 9.980000000000001e-06,
      "loss": 1.2503,
      "step": 499
    },
    {
      "epoch": 0.14892309983432306,
      "grad_norm": 0.9176982641220093,
      "learning_rate": 1e-05,
      "loss": 1.2307,
      "step": 500
    },
    {
      "epoch": 0.1492209460339917,
      "grad_norm": 0.9213762283325195,
      "learning_rate": 1.002e-05,
      "loss": 1.233,
      "step": 501
    },
    {
      "epoch": 0.14951879223366035,
      "grad_norm": 0.8724860548973083,
      "learning_rate": 1.004e-05,
      "loss": 1.2208,
      "step": 502
    },
    {
      "epoch": 0.14981663843332899,
      "grad_norm": 0.8980942964553833,
      "learning_rate": 1.006e-05,
      "loss": 1.2273,
      "step": 503
    },
    {
      "epoch": 0.15011448463299765,
      "grad_norm": 0.9424652457237244,
      "learning_rate": 1.008e-05,
      "loss": 1.2458,
      "step": 504
    },
    {
      "epoch": 0.15041233083266628,
      "grad_norm": 0.8460282683372498,
      "learning_rate": 1.0100000000000002e-05,
      "loss": 1.2276,
      "step": 505
    },
    {
      "epoch": 0.15071017703233494,
      "grad_norm": 0.9348447322845459,
      "learning_rate": 1.0120000000000001e-05,
      "loss": 1.2256,
      "step": 506
    },
    {
      "epoch": 0.15100802323200357,
      "grad_norm": 0.8438363075256348,
      "learning_rate": 1.0140000000000001e-05,
      "loss": 1.2341,
      "step": 507
    },
    {
      "epoch": 0.1513058694316722,
      "grad_norm": 0.8746684789657593,
      "learning_rate": 1.0160000000000001e-05,
      "loss": 1.2145,
      "step": 508
    },
    {
      "epoch": 0.15160371563134087,
      "grad_norm": 0.8724362850189209,
      "learning_rate": 1.018e-05,
      "loss": 1.2431,
      "step": 509
    },
    {
      "epoch": 0.1519015618310095,
      "grad_norm": 0.8627363443374634,
      "learning_rate": 1.02e-05,
      "loss": 1.221,
      "step": 510
    },
    {
      "epoch": 0.15219940803067816,
      "grad_norm": 0.8830870985984802,
      "learning_rate": 1.022e-05,
      "loss": 1.2346,
      "step": 511
    },
    {
      "epoch": 0.1524972542303468,
      "grad_norm": 0.8937995433807373,
      "learning_rate": 1.024e-05,
      "loss": 1.2372,
      "step": 512
    },
    {
      "epoch": 0.15279510043001546,
      "grad_norm": 0.8519806861877441,
      "learning_rate": 1.0260000000000002e-05,
      "loss": 1.234,
      "step": 513
    },
    {
      "epoch": 0.1530929466296841,
      "grad_norm": 0.886282742023468,
      "learning_rate": 1.0280000000000002e-05,
      "loss": 1.2286,
      "step": 514
    },
    {
      "epoch": 0.15339079282935275,
      "grad_norm": 0.8713241815567017,
      "learning_rate": 1.0300000000000001e-05,
      "loss": 1.2316,
      "step": 515
    },
    {
      "epoch": 0.15368863902902138,
      "grad_norm": 0.8357083797454834,
      "learning_rate": 1.0320000000000001e-05,
      "loss": 1.2223,
      "step": 516
    },
    {
      "epoch": 0.15398648522869005,
      "grad_norm": 0.8487093448638916,
      "learning_rate": 1.0340000000000001e-05,
      "loss": 1.2301,
      "step": 517
    },
    {
      "epoch": 0.15428433142835868,
      "grad_norm": 0.8660985827445984,
      "learning_rate": 1.036e-05,
      "loss": 1.2286,
      "step": 518
    },
    {
      "epoch": 0.15458217762802734,
      "grad_norm": 0.8656616806983948,
      "learning_rate": 1.038e-05,
      "loss": 1.2296,
      "step": 519
    },
    {
      "epoch": 0.15488002382769597,
      "grad_norm": 0.8939107656478882,
      "learning_rate": 1.04e-05,
      "loss": 1.2267,
      "step": 520
    },
    {
      "epoch": 0.1551778700273646,
      "grad_norm": 0.9108803868293762,
      "learning_rate": 1.0420000000000002e-05,
      "loss": 1.2475,
      "step": 521
    },
    {
      "epoch": 0.15547571622703327,
      "grad_norm": 0.8460344076156616,
      "learning_rate": 1.0440000000000002e-05,
      "loss": 1.2212,
      "step": 522
    },
    {
      "epoch": 0.1557735624267019,
      "grad_norm": 0.8857704997062683,
      "learning_rate": 1.0460000000000001e-05,
      "loss": 1.2052,
      "step": 523
    },
    {
      "epoch": 0.15607140862637056,
      "grad_norm": 0.911916196346283,
      "learning_rate": 1.0480000000000001e-05,
      "loss": 1.2603,
      "step": 524
    },
    {
      "epoch": 0.1563692548260392,
      "grad_norm": 0.8645758032798767,
      "learning_rate": 1.0500000000000001e-05,
      "loss": 1.2407,
      "step": 525
    },
    {
      "epoch": 0.15666710102570786,
      "grad_norm": 0.8820276260375977,
      "learning_rate": 1.0520000000000001e-05,
      "loss": 1.222,
      "step": 526
    },
    {
      "epoch": 0.1569649472253765,
      "grad_norm": 0.8692166805267334,
      "learning_rate": 1.054e-05,
      "loss": 1.2249,
      "step": 527
    },
    {
      "epoch": 0.15726279342504515,
      "grad_norm": 0.8875836133956909,
      "learning_rate": 1.056e-05,
      "loss": 1.2432,
      "step": 528
    },
    {
      "epoch": 0.15756063962471378,
      "grad_norm": 0.8823421597480774,
      "learning_rate": 1.0580000000000002e-05,
      "loss": 1.2162,
      "step": 529
    },
    {
      "epoch": 0.15785848582438244,
      "grad_norm": 0.8723965883255005,
      "learning_rate": 1.0600000000000002e-05,
      "loss": 1.2218,
      "step": 530
    },
    {
      "epoch": 0.15815633202405108,
      "grad_norm": 0.8678070306777954,
      "learning_rate": 1.0620000000000002e-05,
      "loss": 1.215,
      "step": 531
    },
    {
      "epoch": 0.15845417822371974,
      "grad_norm": 0.8555147051811218,
      "learning_rate": 1.0640000000000001e-05,
      "loss": 1.2224,
      "step": 532
    },
    {
      "epoch": 0.15875202442338837,
      "grad_norm": 0.9082272052764893,
      "learning_rate": 1.0660000000000001e-05,
      "loss": 1.2372,
      "step": 533
    },
    {
      "epoch": 0.15904987062305703,
      "grad_norm": 0.8980268239974976,
      "learning_rate": 1.0680000000000001e-05,
      "loss": 1.2253,
      "step": 534
    },
    {
      "epoch": 0.15934771682272567,
      "grad_norm": 0.8838905692100525,
      "learning_rate": 1.0700000000000001e-05,
      "loss": 1.2296,
      "step": 535
    },
    {
      "epoch": 0.1596455630223943,
      "grad_norm": 0.9228916168212891,
      "learning_rate": 1.072e-05,
      "loss": 1.222,
      "step": 536
    },
    {
      "epoch": 0.15994340922206296,
      "grad_norm": 0.8654788732528687,
      "learning_rate": 1.0740000000000002e-05,
      "loss": 1.2161,
      "step": 537
    },
    {
      "epoch": 0.1602412554217316,
      "grad_norm": 0.8972218036651611,
      "learning_rate": 1.0760000000000002e-05,
      "loss": 1.2077,
      "step": 538
    },
    {
      "epoch": 0.16053910162140025,
      "grad_norm": 0.8432794213294983,
      "learning_rate": 1.0780000000000002e-05,
      "loss": 1.2206,
      "step": 539
    },
    {
      "epoch": 0.1608369478210689,
      "grad_norm": 0.9376282095909119,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 1.2252,
      "step": 540
    },
    {
      "epoch": 0.16113479402073755,
      "grad_norm": 0.8277711272239685,
      "learning_rate": 1.0820000000000001e-05,
      "loss": 1.2301,
      "step": 541
    },
    {
      "epoch": 0.16143264022040618,
      "grad_norm": 0.8760358691215515,
      "learning_rate": 1.0840000000000001e-05,
      "loss": 1.225,
      "step": 542
    },
    {
      "epoch": 0.16173048642007484,
      "grad_norm": 0.8795649409294128,
      "learning_rate": 1.0860000000000001e-05,
      "loss": 1.2334,
      "step": 543
    },
    {
      "epoch": 0.16202833261974348,
      "grad_norm": 0.8629165887832642,
      "learning_rate": 1.0880000000000001e-05,
      "loss": 1.2147,
      "step": 544
    },
    {
      "epoch": 0.16232617881941214,
      "grad_norm": 0.8989906311035156,
      "learning_rate": 1.0900000000000002e-05,
      "loss": 1.2441,
      "step": 545
    },
    {
      "epoch": 0.16262402501908077,
      "grad_norm": 0.8646058440208435,
      "learning_rate": 1.0920000000000002e-05,
      "loss": 1.2303,
      "step": 546
    },
    {
      "epoch": 0.16292187121874943,
      "grad_norm": 0.874629557132721,
      "learning_rate": 1.0940000000000002e-05,
      "loss": 1.2428,
      "step": 547
    },
    {
      "epoch": 0.16321971741841806,
      "grad_norm": 1.095576524734497,
      "learning_rate": 1.0960000000000002e-05,
      "loss": 1.2268,
      "step": 548
    },
    {
      "epoch": 0.1635175636180867,
      "grad_norm": 0.8792427778244019,
      "learning_rate": 1.0980000000000002e-05,
      "loss": 1.2131,
      "step": 549
    },
    {
      "epoch": 0.16381540981775536,
      "grad_norm": 0.8681890368461609,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 1.2151,
      "step": 550
    },
    {
      "epoch": 0.164113256017424,
      "grad_norm": 0.8687170147895813,
      "learning_rate": 1.1020000000000001e-05,
      "loss": 1.2287,
      "step": 551
    },
    {
      "epoch": 0.16441110221709265,
      "grad_norm": 0.8476585745811462,
      "learning_rate": 1.1040000000000001e-05,
      "loss": 1.2084,
      "step": 552
    },
    {
      "epoch": 0.16470894841676129,
      "grad_norm": 0.8646326065063477,
      "learning_rate": 1.1060000000000003e-05,
      "loss": 1.2155,
      "step": 553
    },
    {
      "epoch": 0.16500679461642995,
      "grad_norm": 0.8767011761665344,
      "learning_rate": 1.1080000000000002e-05,
      "loss": 1.225,
      "step": 554
    },
    {
      "epoch": 0.16530464081609858,
      "grad_norm": 0.8663629293441772,
      "learning_rate": 1.1100000000000002e-05,
      "loss": 1.2245,
      "step": 555
    },
    {
      "epoch": 0.16560248701576724,
      "grad_norm": 0.9188178777694702,
      "learning_rate": 1.1120000000000002e-05,
      "loss": 1.2148,
      "step": 556
    },
    {
      "epoch": 0.16590033321543587,
      "grad_norm": 0.8939908146858215,
      "learning_rate": 1.1140000000000002e-05,
      "loss": 1.2347,
      "step": 557
    },
    {
      "epoch": 0.16619817941510454,
      "grad_norm": 0.8432437181472778,
      "learning_rate": 1.1160000000000002e-05,
      "loss": 1.2188,
      "step": 558
    },
    {
      "epoch": 0.16649602561477317,
      "grad_norm": 0.8336137533187866,
      "learning_rate": 1.1180000000000001e-05,
      "loss": 1.2255,
      "step": 559
    },
    {
      "epoch": 0.16679387181444183,
      "grad_norm": 0.8656436204910278,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 1.2221,
      "step": 560
    },
    {
      "epoch": 0.16709171801411046,
      "grad_norm": 0.8557179570198059,
      "learning_rate": 1.1220000000000003e-05,
      "loss": 1.2121,
      "step": 561
    },
    {
      "epoch": 0.1673895642137791,
      "grad_norm": 0.8736665844917297,
      "learning_rate": 1.1240000000000002e-05,
      "loss": 1.2199,
      "step": 562
    },
    {
      "epoch": 0.16768741041344776,
      "grad_norm": 0.8865178227424622,
      "learning_rate": 1.126e-05,
      "loss": 1.2323,
      "step": 563
    },
    {
      "epoch": 0.1679852566131164,
      "grad_norm": 0.9162757992744446,
      "learning_rate": 1.128e-05,
      "loss": 1.2184,
      "step": 564
    },
    {
      "epoch": 0.16828310281278505,
      "grad_norm": 0.8775267004966736,
      "learning_rate": 1.13e-05,
      "loss": 1.2153,
      "step": 565
    },
    {
      "epoch": 0.16858094901245368,
      "grad_norm": 0.8618582487106323,
      "learning_rate": 1.132e-05,
      "loss": 1.2211,
      "step": 566
    },
    {
      "epoch": 0.16887879521212235,
      "grad_norm": 0.8913806080818176,
      "learning_rate": 1.134e-05,
      "loss": 1.2435,
      "step": 567
    },
    {
      "epoch": 0.16917664141179098,
      "grad_norm": 0.8814314007759094,
      "learning_rate": 1.136e-05,
      "loss": 1.226,
      "step": 568
    },
    {
      "epoch": 0.16947448761145964,
      "grad_norm": 0.8640012741088867,
      "learning_rate": 1.138e-05,
      "loss": 1.2199,
      "step": 569
    },
    {
      "epoch": 0.16977233381112827,
      "grad_norm": 0.8627512454986572,
      "learning_rate": 1.14e-05,
      "loss": 1.2139,
      "step": 570
    },
    {
      "epoch": 0.17007018001079693,
      "grad_norm": 0.8850782513618469,
      "learning_rate": 1.142e-05,
      "loss": 1.2412,
      "step": 571
    },
    {
      "epoch": 0.17036802621046557,
      "grad_norm": 0.8755152821540833,
      "learning_rate": 1.144e-05,
      "loss": 1.2254,
      "step": 572
    },
    {
      "epoch": 0.17066587241013423,
      "grad_norm": 0.8645856380462646,
      "learning_rate": 1.146e-05,
      "loss": 1.2088,
      "step": 573
    },
    {
      "epoch": 0.17096371860980286,
      "grad_norm": 0.8511538505554199,
      "learning_rate": 1.148e-05,
      "loss": 1.2135,
      "step": 574
    },
    {
      "epoch": 0.17126156480947152,
      "grad_norm": 0.920762300491333,
      "learning_rate": 1.15e-05,
      "loss": 1.2203,
      "step": 575
    },
    {
      "epoch": 0.17155941100914016,
      "grad_norm": 0.9491738677024841,
      "learning_rate": 1.152e-05,
      "loss": 1.2252,
      "step": 576
    },
    {
      "epoch": 0.1718572572088088,
      "grad_norm": 0.9299079179763794,
      "learning_rate": 1.154e-05,
      "loss": 1.2157,
      "step": 577
    },
    {
      "epoch": 0.17215510340847745,
      "grad_norm": 0.8946446776390076,
      "learning_rate": 1.156e-05,
      "loss": 1.2125,
      "step": 578
    },
    {
      "epoch": 0.17245294960814608,
      "grad_norm": 0.898821234703064,
      "learning_rate": 1.1580000000000001e-05,
      "loss": 1.2318,
      "step": 579
    },
    {
      "epoch": 0.17275079580781474,
      "grad_norm": 0.895999014377594,
      "learning_rate": 1.16e-05,
      "loss": 1.1936,
      "step": 580
    },
    {
      "epoch": 0.17304864200748338,
      "grad_norm": 0.9102035760879517,
      "learning_rate": 1.162e-05,
      "loss": 1.236,
      "step": 581
    },
    {
      "epoch": 0.17334648820715204,
      "grad_norm": 0.8564568758010864,
      "learning_rate": 1.164e-05,
      "loss": 1.2186,
      "step": 582
    },
    {
      "epoch": 0.17364433440682067,
      "grad_norm": 0.9261096715927124,
      "learning_rate": 1.166e-05,
      "loss": 1.2521,
      "step": 583
    },
    {
      "epoch": 0.17394218060648933,
      "grad_norm": 0.8637592792510986,
      "learning_rate": 1.168e-05,
      "loss": 1.2331,
      "step": 584
    },
    {
      "epoch": 0.17424002680615797,
      "grad_norm": 0.9201040267944336,
      "learning_rate": 1.17e-05,
      "loss": 1.2354,
      "step": 585
    },
    {
      "epoch": 0.17453787300582663,
      "grad_norm": 0.8958178162574768,
      "learning_rate": 1.172e-05,
      "loss": 1.2108,
      "step": 586
    },
    {
      "epoch": 0.17483571920549526,
      "grad_norm": 0.8750001192092896,
      "learning_rate": 1.1740000000000001e-05,
      "loss": 1.2161,
      "step": 587
    },
    {
      "epoch": 0.17513356540516392,
      "grad_norm": 0.8706455826759338,
      "learning_rate": 1.1760000000000001e-05,
      "loss": 1.206,
      "step": 588
    },
    {
      "epoch": 0.17543141160483255,
      "grad_norm": 0.9035860896110535,
      "learning_rate": 1.178e-05,
      "loss": 1.2261,
      "step": 589
    },
    {
      "epoch": 0.1757292578045012,
      "grad_norm": 0.8816605806350708,
      "learning_rate": 1.18e-05,
      "loss": 1.2134,
      "step": 590
    },
    {
      "epoch": 0.17602710400416985,
      "grad_norm": 0.8740524649620056,
      "learning_rate": 1.182e-05,
      "loss": 1.2253,
      "step": 591
    },
    {
      "epoch": 0.17632495020383848,
      "grad_norm": 0.8848228454589844,
      "learning_rate": 1.184e-05,
      "loss": 1.2258,
      "step": 592
    },
    {
      "epoch": 0.17662279640350714,
      "grad_norm": 0.9045940041542053,
      "learning_rate": 1.186e-05,
      "loss": 1.2095,
      "step": 593
    },
    {
      "epoch": 0.17692064260317578,
      "grad_norm": 0.8732846975326538,
      "learning_rate": 1.188e-05,
      "loss": 1.2151,
      "step": 594
    },
    {
      "epoch": 0.17721848880284444,
      "grad_norm": 0.9114317297935486,
      "learning_rate": 1.1900000000000001e-05,
      "loss": 1.2191,
      "step": 595
    },
    {
      "epoch": 0.17751633500251307,
      "grad_norm": 0.9063019156455994,
      "learning_rate": 1.1920000000000001e-05,
      "loss": 1.2227,
      "step": 596
    },
    {
      "epoch": 0.17781418120218173,
      "grad_norm": 0.8572113513946533,
      "learning_rate": 1.1940000000000001e-05,
      "loss": 1.2306,
      "step": 597
    },
    {
      "epoch": 0.17811202740185036,
      "grad_norm": 0.904134213924408,
      "learning_rate": 1.196e-05,
      "loss": 1.2273,
      "step": 598
    },
    {
      "epoch": 0.17840987360151903,
      "grad_norm": 0.8823001980781555,
      "learning_rate": 1.198e-05,
      "loss": 1.2155,
      "step": 599
    },
    {
      "epoch": 0.17870771980118766,
      "grad_norm": 0.9021396636962891,
      "learning_rate": 1.2e-05,
      "loss": 1.2359,
      "step": 600
    },
    {
      "epoch": 0.17900556600085632,
      "grad_norm": 0.8999623656272888,
      "learning_rate": 1.202e-05,
      "loss": 1.2357,
      "step": 601
    },
    {
      "epoch": 0.17930341220052495,
      "grad_norm": 0.9321269392967224,
      "learning_rate": 1.204e-05,
      "loss": 1.2134,
      "step": 602
    },
    {
      "epoch": 0.1796012584001936,
      "grad_norm": 0.8833494782447815,
      "learning_rate": 1.2060000000000001e-05,
      "loss": 1.2195,
      "step": 603
    },
    {
      "epoch": 0.17989910459986225,
      "grad_norm": 0.9311667084693909,
      "learning_rate": 1.2080000000000001e-05,
      "loss": 1.2327,
      "step": 604
    },
    {
      "epoch": 0.18019695079953088,
      "grad_norm": 0.8968342542648315,
      "learning_rate": 1.2100000000000001e-05,
      "loss": 1.209,
      "step": 605
    },
    {
      "epoch": 0.18049479699919954,
      "grad_norm": 0.9245573282241821,
      "learning_rate": 1.2120000000000001e-05,
      "loss": 1.2115,
      "step": 606
    },
    {
      "epoch": 0.18079264319886817,
      "grad_norm": 0.8936393857002258,
      "learning_rate": 1.214e-05,
      "loss": 1.2316,
      "step": 607
    },
    {
      "epoch": 0.18109048939853684,
      "grad_norm": 0.8967474699020386,
      "learning_rate": 1.216e-05,
      "loss": 1.2054,
      "step": 608
    },
    {
      "epoch": 0.18138833559820547,
      "grad_norm": 0.8783530592918396,
      "learning_rate": 1.218e-05,
      "loss": 1.2143,
      "step": 609
    },
    {
      "epoch": 0.18168618179787413,
      "grad_norm": 0.8443657755851746,
      "learning_rate": 1.22e-05,
      "loss": 1.2271,
      "step": 610
    },
    {
      "epoch": 0.18198402799754276,
      "grad_norm": 0.8986846208572388,
      "learning_rate": 1.2220000000000002e-05,
      "loss": 1.2056,
      "step": 611
    },
    {
      "epoch": 0.18228187419721142,
      "grad_norm": 0.8509023785591125,
      "learning_rate": 1.2240000000000001e-05,
      "loss": 1.2333,
      "step": 612
    },
    {
      "epoch": 0.18257972039688006,
      "grad_norm": 0.8772693276405334,
      "learning_rate": 1.2260000000000001e-05,
      "loss": 1.2263,
      "step": 613
    },
    {
      "epoch": 0.18287756659654872,
      "grad_norm": 0.8497651219367981,
      "learning_rate": 1.2280000000000001e-05,
      "loss": 1.2046,
      "step": 614
    },
    {
      "epoch": 0.18317541279621735,
      "grad_norm": 0.8755048513412476,
      "learning_rate": 1.23e-05,
      "loss": 1.2175,
      "step": 615
    },
    {
      "epoch": 0.183473258995886,
      "grad_norm": 0.8201553821563721,
      "learning_rate": 1.232e-05,
      "loss": 1.2298,
      "step": 616
    },
    {
      "epoch": 0.18377110519555465,
      "grad_norm": 0.8523305654525757,
      "learning_rate": 1.234e-05,
      "loss": 1.2368,
      "step": 617
    },
    {
      "epoch": 0.18406895139522328,
      "grad_norm": 0.8759392499923706,
      "learning_rate": 1.236e-05,
      "loss": 1.2106,
      "step": 618
    },
    {
      "epoch": 0.18436679759489194,
      "grad_norm": 0.8349262475967407,
      "learning_rate": 1.2380000000000002e-05,
      "loss": 1.2202,
      "step": 619
    },
    {
      "epoch": 0.18466464379456057,
      "grad_norm": 0.9339987635612488,
      "learning_rate": 1.2400000000000002e-05,
      "loss": 1.2203,
      "step": 620
    },
    {
      "epoch": 0.18496248999422923,
      "grad_norm": 0.9326422810554504,
      "learning_rate": 1.2420000000000001e-05,
      "loss": 1.2417,
      "step": 621
    },
    {
      "epoch": 0.18526033619389787,
      "grad_norm": 0.8437680006027222,
      "learning_rate": 1.2440000000000001e-05,
      "loss": 1.211,
      "step": 622
    },
    {
      "epoch": 0.18555818239356653,
      "grad_norm": 0.8542959690093994,
      "learning_rate": 1.2460000000000001e-05,
      "loss": 1.2284,
      "step": 623
    },
    {
      "epoch": 0.18585602859323516,
      "grad_norm": 0.9304774403572083,
      "learning_rate": 1.248e-05,
      "loss": 1.2279,
      "step": 624
    },
    {
      "epoch": 0.18615387479290382,
      "grad_norm": 0.8755531907081604,
      "learning_rate": 1.25e-05,
      "loss": 1.2159,
      "step": 625
    },
    {
      "epoch": 0.18645172099257246,
      "grad_norm": 0.8923066854476929,
      "learning_rate": 1.252e-05,
      "loss": 1.2077,
      "step": 626
    },
    {
      "epoch": 0.18674956719224112,
      "grad_norm": 0.8681371212005615,
      "learning_rate": 1.254e-05,
      "loss": 1.2391,
      "step": 627
    },
    {
      "epoch": 0.18704741339190975,
      "grad_norm": 0.8848538398742676,
      "learning_rate": 1.2560000000000002e-05,
      "loss": 1.2271,
      "step": 628
    },
    {
      "epoch": 0.1873452595915784,
      "grad_norm": 0.8799881339073181,
      "learning_rate": 1.2580000000000002e-05,
      "loss": 1.215,
      "step": 629
    },
    {
      "epoch": 0.18764310579124704,
      "grad_norm": 0.8865759372711182,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 1.2293,
      "step": 630
    },
    {
      "epoch": 0.18794095199091568,
      "grad_norm": 0.8680864572525024,
      "learning_rate": 1.2620000000000001e-05,
      "loss": 1.2225,
      "step": 631
    },
    {
      "epoch": 0.18823879819058434,
      "grad_norm": 0.8316558599472046,
      "learning_rate": 1.2640000000000001e-05,
      "loss": 1.2178,
      "step": 632
    },
    {
      "epoch": 0.18853664439025297,
      "grad_norm": 0.8623107075691223,
      "learning_rate": 1.266e-05,
      "loss": 1.2064,
      "step": 633
    },
    {
      "epoch": 0.18883449058992163,
      "grad_norm": 0.9063116312026978,
      "learning_rate": 1.268e-05,
      "loss": 1.2118,
      "step": 634
    },
    {
      "epoch": 0.18913233678959027,
      "grad_norm": 0.8657467365264893,
      "learning_rate": 1.27e-05,
      "loss": 1.2098,
      "step": 635
    },
    {
      "epoch": 0.18943018298925893,
      "grad_norm": 0.9021874666213989,
      "learning_rate": 1.2720000000000002e-05,
      "loss": 1.2238,
      "step": 636
    },
    {
      "epoch": 0.18972802918892756,
      "grad_norm": 0.8496937155723572,
      "learning_rate": 1.2740000000000002e-05,
      "loss": 1.2027,
      "step": 637
    },
    {
      "epoch": 0.19002587538859622,
      "grad_norm": 0.9396972060203552,
      "learning_rate": 1.2760000000000001e-05,
      "loss": 1.2232,
      "step": 638
    },
    {
      "epoch": 0.19032372158826485,
      "grad_norm": 0.8533748984336853,
      "learning_rate": 1.2780000000000001e-05,
      "loss": 1.1946,
      "step": 639
    },
    {
      "epoch": 0.19062156778793352,
      "grad_norm": 0.8975105881690979,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 1.2118,
      "step": 640
    },
    {
      "epoch": 0.19091941398760215,
      "grad_norm": 0.9015722274780273,
      "learning_rate": 1.2820000000000001e-05,
      "loss": 1.2291,
      "step": 641
    },
    {
      "epoch": 0.1912172601872708,
      "grad_norm": 0.8721552491188049,
      "learning_rate": 1.284e-05,
      "loss": 1.2147,
      "step": 642
    },
    {
      "epoch": 0.19151510638693944,
      "grad_norm": 0.8680423498153687,
      "learning_rate": 1.286e-05,
      "loss": 1.2087,
      "step": 643
    },
    {
      "epoch": 0.19181295258660808,
      "grad_norm": 0.9278917908668518,
      "learning_rate": 1.2880000000000002e-05,
      "loss": 1.2333,
      "step": 644
    },
    {
      "epoch": 0.19211079878627674,
      "grad_norm": 0.8421360850334167,
      "learning_rate": 1.2900000000000002e-05,
      "loss": 1.2135,
      "step": 645
    },
    {
      "epoch": 0.19240864498594537,
      "grad_norm": 0.8617555499076843,
      "learning_rate": 1.2920000000000002e-05,
      "loss": 1.2224,
      "step": 646
    },
    {
      "epoch": 0.19270649118561403,
      "grad_norm": 0.9047665596008301,
      "learning_rate": 1.2940000000000001e-05,
      "loss": 1.2215,
      "step": 647
    },
    {
      "epoch": 0.19300433738528266,
      "grad_norm": 0.8864332437515259,
      "learning_rate": 1.2960000000000001e-05,
      "loss": 1.2132,
      "step": 648
    },
    {
      "epoch": 0.19330218358495133,
      "grad_norm": 0.8074783682823181,
      "learning_rate": 1.2980000000000001e-05,
      "loss": 1.2177,
      "step": 649
    },
    {
      "epoch": 0.19360002978461996,
      "grad_norm": 0.8635845184326172,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 1.212,
      "step": 650
    },
    {
      "epoch": 0.19389787598428862,
      "grad_norm": 0.8365917205810547,
      "learning_rate": 1.302e-05,
      "loss": 1.2192,
      "step": 651
    },
    {
      "epoch": 0.19419572218395725,
      "grad_norm": 0.8393798470497131,
      "learning_rate": 1.3040000000000002e-05,
      "loss": 1.2045,
      "step": 652
    },
    {
      "epoch": 0.19449356838362591,
      "grad_norm": 0.8882096409797668,
      "learning_rate": 1.3060000000000002e-05,
      "loss": 1.2135,
      "step": 653
    },
    {
      "epoch": 0.19479141458329455,
      "grad_norm": 0.931267499923706,
      "learning_rate": 1.3080000000000002e-05,
      "loss": 1.2149,
      "step": 654
    },
    {
      "epoch": 0.1950892607829632,
      "grad_norm": 0.8753933906555176,
      "learning_rate": 1.3100000000000002e-05,
      "loss": 1.1922,
      "step": 655
    },
    {
      "epoch": 0.19538710698263184,
      "grad_norm": 0.8708509206771851,
      "learning_rate": 1.3120000000000001e-05,
      "loss": 1.2332,
      "step": 656
    },
    {
      "epoch": 0.1956849531823005,
      "grad_norm": 0.902732789516449,
      "learning_rate": 1.3140000000000001e-05,
      "loss": 1.2277,
      "step": 657
    },
    {
      "epoch": 0.19598279938196914,
      "grad_norm": 0.8589592576026917,
      "learning_rate": 1.3160000000000001e-05,
      "loss": 1.2239,
      "step": 658
    },
    {
      "epoch": 0.19628064558163777,
      "grad_norm": 0.8285161852836609,
      "learning_rate": 1.3180000000000001e-05,
      "loss": 1.2121,
      "step": 659
    },
    {
      "epoch": 0.19657849178130643,
      "grad_norm": 0.8366507291793823,
      "learning_rate": 1.3200000000000002e-05,
      "loss": 1.2141,
      "step": 660
    },
    {
      "epoch": 0.19687633798097506,
      "grad_norm": 0.826694905757904,
      "learning_rate": 1.3220000000000002e-05,
      "loss": 1.2033,
      "step": 661
    },
    {
      "epoch": 0.19717418418064372,
      "grad_norm": 0.9147154688835144,
      "learning_rate": 1.3240000000000002e-05,
      "loss": 1.2242,
      "step": 662
    },
    {
      "epoch": 0.19747203038031236,
      "grad_norm": 0.869482159614563,
      "learning_rate": 1.3260000000000002e-05,
      "loss": 1.2162,
      "step": 663
    },
    {
      "epoch": 0.19776987657998102,
      "grad_norm": 0.8541890978813171,
      "learning_rate": 1.3280000000000002e-05,
      "loss": 1.2342,
      "step": 664
    },
    {
      "epoch": 0.19806772277964965,
      "grad_norm": 0.882982611656189,
      "learning_rate": 1.3300000000000001e-05,
      "loss": 1.2091,
      "step": 665
    },
    {
      "epoch": 0.1983655689793183,
      "grad_norm": 0.9164595007896423,
      "learning_rate": 1.3320000000000001e-05,
      "loss": 1.2147,
      "step": 666
    },
    {
      "epoch": 0.19866341517898695,
      "grad_norm": 0.878693163394928,
      "learning_rate": 1.3340000000000001e-05,
      "loss": 1.2013,
      "step": 667
    },
    {
      "epoch": 0.1989612613786556,
      "grad_norm": 0.92546147108078,
      "learning_rate": 1.3360000000000003e-05,
      "loss": 1.2109,
      "step": 668
    },
    {
      "epoch": 0.19925910757832424,
      "grad_norm": 0.9355611801147461,
      "learning_rate": 1.3380000000000002e-05,
      "loss": 1.2136,
      "step": 669
    },
    {
      "epoch": 0.1995569537779929,
      "grad_norm": 0.9295053482055664,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 1.2186,
      "step": 670
    },
    {
      "epoch": 0.19985479997766153,
      "grad_norm": 0.9733754992485046,
      "learning_rate": 1.3420000000000002e-05,
      "loss": 1.233,
      "step": 671
    },
    {
      "epoch": 0.20015264617733017,
      "grad_norm": 0.8522588610649109,
      "learning_rate": 1.3440000000000002e-05,
      "loss": 1.2064,
      "step": 672
    },
    {
      "epoch": 0.20045049237699883,
      "grad_norm": 0.8661471009254456,
      "learning_rate": 1.3460000000000002e-05,
      "loss": 1.2129,
      "step": 673
    },
    {
      "epoch": 0.20074833857666746,
      "grad_norm": 0.8988415598869324,
      "learning_rate": 1.3480000000000001e-05,
      "loss": 1.2088,
      "step": 674
    },
    {
      "epoch": 0.20104618477633612,
      "grad_norm": 0.861110270023346,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 1.2088,
      "step": 675
    },
    {
      "epoch": 0.20134403097600476,
      "grad_norm": 0.8818802237510681,
      "learning_rate": 1.3520000000000003e-05,
      "loss": 1.1982,
      "step": 676
    },
    {
      "epoch": 0.20164187717567342,
      "grad_norm": 0.962843120098114,
      "learning_rate": 1.3540000000000003e-05,
      "loss": 1.2205,
      "step": 677
    },
    {
      "epoch": 0.20193972337534205,
      "grad_norm": 0.8584957122802734,
      "learning_rate": 1.3560000000000002e-05,
      "loss": 1.219,
      "step": 678
    },
    {
      "epoch": 0.2022375695750107,
      "grad_norm": 0.8659563064575195,
      "learning_rate": 1.3580000000000002e-05,
      "loss": 1.2384,
      "step": 679
    },
    {
      "epoch": 0.20253541577467934,
      "grad_norm": 0.8857707381248474,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 1.2162,
      "step": 680
    },
    {
      "epoch": 0.202833261974348,
      "grad_norm": 0.8685858845710754,
      "learning_rate": 1.3620000000000002e-05,
      "loss": 1.2327,
      "step": 681
    },
    {
      "epoch": 0.20313110817401664,
      "grad_norm": 0.8808422088623047,
      "learning_rate": 1.3640000000000002e-05,
      "loss": 1.2081,
      "step": 682
    },
    {
      "epoch": 0.2034289543736853,
      "grad_norm": 0.8810634613037109,
      "learning_rate": 1.3660000000000001e-05,
      "loss": 1.2035,
      "step": 683
    },
    {
      "epoch": 0.20372680057335393,
      "grad_norm": 0.7999762892723083,
      "learning_rate": 1.3680000000000003e-05,
      "loss": 1.2288,
      "step": 684
    },
    {
      "epoch": 0.2040246467730226,
      "grad_norm": 0.8779205083847046,
      "learning_rate": 1.3700000000000003e-05,
      "loss": 1.2054,
      "step": 685
    },
    {
      "epoch": 0.20432249297269123,
      "grad_norm": 0.8029731512069702,
      "learning_rate": 1.3720000000000002e-05,
      "loss": 1.2196,
      "step": 686
    },
    {
      "epoch": 0.20462033917235986,
      "grad_norm": 0.8677182793617249,
      "learning_rate": 1.3740000000000002e-05,
      "loss": 1.2152,
      "step": 687
    },
    {
      "epoch": 0.20491818537202852,
      "grad_norm": 0.8408791422843933,
      "learning_rate": 1.376e-05,
      "loss": 1.214,
      "step": 688
    },
    {
      "epoch": 0.20521603157169716,
      "grad_norm": 0.828201174736023,
      "learning_rate": 1.378e-05,
      "loss": 1.2108,
      "step": 689
    },
    {
      "epoch": 0.20551387777136582,
      "grad_norm": 0.8862516283988953,
      "learning_rate": 1.38e-05,
      "loss": 1.2056,
      "step": 690
    },
    {
      "epoch": 0.20581172397103445,
      "grad_norm": 0.8860002160072327,
      "learning_rate": 1.382e-05,
      "loss": 1.2099,
      "step": 691
    },
    {
      "epoch": 0.2061095701707031,
      "grad_norm": 0.8548729419708252,
      "learning_rate": 1.384e-05,
      "loss": 1.2139,
      "step": 692
    },
    {
      "epoch": 0.20640741637037174,
      "grad_norm": 0.9217012524604797,
      "learning_rate": 1.386e-05,
      "loss": 1.2093,
      "step": 693
    },
    {
      "epoch": 0.2067052625700404,
      "grad_norm": 0.8493072390556335,
      "learning_rate": 1.3880000000000001e-05,
      "loss": 1.2197,
      "step": 694
    },
    {
      "epoch": 0.20700310876970904,
      "grad_norm": 0.855212926864624,
      "learning_rate": 1.39e-05,
      "loss": 1.216,
      "step": 695
    },
    {
      "epoch": 0.2073009549693777,
      "grad_norm": 0.8438214659690857,
      "learning_rate": 1.392e-05,
      "loss": 1.2262,
      "step": 696
    },
    {
      "epoch": 0.20759880116904633,
      "grad_norm": 0.8461406826972961,
      "learning_rate": 1.394e-05,
      "loss": 1.2073,
      "step": 697
    },
    {
      "epoch": 0.207896647368715,
      "grad_norm": 0.7991374731063843,
      "learning_rate": 1.396e-05,
      "loss": 1.1951,
      "step": 698
    },
    {
      "epoch": 0.20819449356838363,
      "grad_norm": 0.8273499608039856,
      "learning_rate": 1.398e-05,
      "loss": 1.1955,
      "step": 699
    },
    {
      "epoch": 0.20849233976805226,
      "grad_norm": 0.811980128288269,
      "learning_rate": 1.4e-05,
      "loss": 1.2092,
      "step": 700
    },
    {
      "epoch": 0.20879018596772092,
      "grad_norm": 0.8373486995697021,
      "learning_rate": 1.402e-05,
      "loss": 1.211,
      "step": 701
    },
    {
      "epoch": 0.20908803216738955,
      "grad_norm": 0.8518791794776917,
      "learning_rate": 1.4040000000000001e-05,
      "loss": 1.2213,
      "step": 702
    },
    {
      "epoch": 0.20938587836705821,
      "grad_norm": 0.8460239768028259,
      "learning_rate": 1.4060000000000001e-05,
      "loss": 1.1904,
      "step": 703
    },
    {
      "epoch": 0.20968372456672685,
      "grad_norm": 0.837502121925354,
      "learning_rate": 1.408e-05,
      "loss": 1.2075,
      "step": 704
    },
    {
      "epoch": 0.2099815707663955,
      "grad_norm": 0.8522453904151917,
      "learning_rate": 1.41e-05,
      "loss": 1.214,
      "step": 705
    },
    {
      "epoch": 0.21027941696606414,
      "grad_norm": 0.8058056831359863,
      "learning_rate": 1.412e-05,
      "loss": 1.2154,
      "step": 706
    },
    {
      "epoch": 0.2105772631657328,
      "grad_norm": 0.8278490304946899,
      "learning_rate": 1.414e-05,
      "loss": 1.2017,
      "step": 707
    },
    {
      "epoch": 0.21087510936540144,
      "grad_norm": 0.855830192565918,
      "learning_rate": 1.416e-05,
      "loss": 1.2151,
      "step": 708
    },
    {
      "epoch": 0.2111729555650701,
      "grad_norm": 0.8688766360282898,
      "learning_rate": 1.418e-05,
      "loss": 1.2247,
      "step": 709
    },
    {
      "epoch": 0.21147080176473873,
      "grad_norm": 0.8472169637680054,
      "learning_rate": 1.4200000000000001e-05,
      "loss": 1.2057,
      "step": 710
    },
    {
      "epoch": 0.2117686479644074,
      "grad_norm": 0.8659970760345459,
      "learning_rate": 1.4220000000000001e-05,
      "loss": 1.2086,
      "step": 711
    },
    {
      "epoch": 0.21206649416407602,
      "grad_norm": 0.7935543656349182,
      "learning_rate": 1.4240000000000001e-05,
      "loss": 1.2294,
      "step": 712
    },
    {
      "epoch": 0.21236434036374466,
      "grad_norm": 0.825660765171051,
      "learning_rate": 1.426e-05,
      "loss": 1.2297,
      "step": 713
    },
    {
      "epoch": 0.21266218656341332,
      "grad_norm": 0.8227145075798035,
      "learning_rate": 1.428e-05,
      "loss": 1.2059,
      "step": 714
    },
    {
      "epoch": 0.21296003276308195,
      "grad_norm": 0.7915933728218079,
      "learning_rate": 1.43e-05,
      "loss": 1.2002,
      "step": 715
    },
    {
      "epoch": 0.2132578789627506,
      "grad_norm": 0.8464688062667847,
      "learning_rate": 1.432e-05,
      "loss": 1.1971,
      "step": 716
    },
    {
      "epoch": 0.21355572516241925,
      "grad_norm": 0.8985934257507324,
      "learning_rate": 1.434e-05,
      "loss": 1.2039,
      "step": 717
    },
    {
      "epoch": 0.2138535713620879,
      "grad_norm": 0.8358312249183655,
      "learning_rate": 1.4360000000000001e-05,
      "loss": 1.1948,
      "step": 718
    },
    {
      "epoch": 0.21415141756175654,
      "grad_norm": 0.9404199123382568,
      "learning_rate": 1.4380000000000001e-05,
      "loss": 1.2225,
      "step": 719
    },
    {
      "epoch": 0.2144492637614252,
      "grad_norm": 0.8191746473312378,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 1.2128,
      "step": 720
    },
    {
      "epoch": 0.21474710996109383,
      "grad_norm": 0.8840652108192444,
      "learning_rate": 1.4420000000000001e-05,
      "loss": 1.2052,
      "step": 721
    },
    {
      "epoch": 0.2150449561607625,
      "grad_norm": 0.8269261121749878,
      "learning_rate": 1.444e-05,
      "loss": 1.2168,
      "step": 722
    },
    {
      "epoch": 0.21534280236043113,
      "grad_norm": 0.8701027035713196,
      "learning_rate": 1.446e-05,
      "loss": 1.2131,
      "step": 723
    },
    {
      "epoch": 0.2156406485600998,
      "grad_norm": 0.8917607069015503,
      "learning_rate": 1.448e-05,
      "loss": 1.2271,
      "step": 724
    },
    {
      "epoch": 0.21593849475976842,
      "grad_norm": 0.8657019734382629,
      "learning_rate": 1.45e-05,
      "loss": 1.2084,
      "step": 725
    },
    {
      "epoch": 0.21623634095943708,
      "grad_norm": 0.8441585898399353,
      "learning_rate": 1.4520000000000002e-05,
      "loss": 1.2148,
      "step": 726
    },
    {
      "epoch": 0.21653418715910572,
      "grad_norm": 0.8501395583152771,
      "learning_rate": 1.4540000000000001e-05,
      "loss": 1.2108,
      "step": 727
    },
    {
      "epoch": 0.21683203335877435,
      "grad_norm": 0.813061535358429,
      "learning_rate": 1.4560000000000001e-05,
      "loss": 1.2231,
      "step": 728
    },
    {
      "epoch": 0.217129879558443,
      "grad_norm": 0.8619956970214844,
      "learning_rate": 1.4580000000000001e-05,
      "loss": 1.2127,
      "step": 729
    },
    {
      "epoch": 0.21742772575811165,
      "grad_norm": 0.8398110866546631,
      "learning_rate": 1.46e-05,
      "loss": 1.1859,
      "step": 730
    },
    {
      "epoch": 0.2177255719577803,
      "grad_norm": 0.859654426574707,
      "learning_rate": 1.462e-05,
      "loss": 1.2029,
      "step": 731
    },
    {
      "epoch": 0.21802341815744894,
      "grad_norm": 0.8375681638717651,
      "learning_rate": 1.464e-05,
      "loss": 1.1856,
      "step": 732
    },
    {
      "epoch": 0.2183212643571176,
      "grad_norm": 0.8332937359809875,
      "learning_rate": 1.466e-05,
      "loss": 1.2124,
      "step": 733
    },
    {
      "epoch": 0.21861911055678623,
      "grad_norm": 0.8546724915504456,
      "learning_rate": 1.4680000000000002e-05,
      "loss": 1.217,
      "step": 734
    },
    {
      "epoch": 0.2189169567564549,
      "grad_norm": 0.8700973987579346,
      "learning_rate": 1.4700000000000002e-05,
      "loss": 1.2218,
      "step": 735
    },
    {
      "epoch": 0.21921480295612353,
      "grad_norm": 0.8805535435676575,
      "learning_rate": 1.4720000000000001e-05,
      "loss": 1.216,
      "step": 736
    },
    {
      "epoch": 0.2195126491557922,
      "grad_norm": 0.8366180062294006,
      "learning_rate": 1.4740000000000001e-05,
      "loss": 1.2053,
      "step": 737
    },
    {
      "epoch": 0.21981049535546082,
      "grad_norm": 0.8360991477966309,
      "learning_rate": 1.4760000000000001e-05,
      "loss": 1.2282,
      "step": 738
    },
    {
      "epoch": 0.22010834155512948,
      "grad_norm": 0.8774970769882202,
      "learning_rate": 1.478e-05,
      "loss": 1.2064,
      "step": 739
    },
    {
      "epoch": 0.22040618775479812,
      "grad_norm": 0.8220171332359314,
      "learning_rate": 1.48e-05,
      "loss": 1.2124,
      "step": 740
    },
    {
      "epoch": 0.22070403395446675,
      "grad_norm": 0.8188339471817017,
      "learning_rate": 1.482e-05,
      "loss": 1.1978,
      "step": 741
    },
    {
      "epoch": 0.2210018801541354,
      "grad_norm": 0.8306236267089844,
      "learning_rate": 1.4840000000000002e-05,
      "loss": 1.2078,
      "step": 742
    },
    {
      "epoch": 0.22129972635380404,
      "grad_norm": 0.8979356288909912,
      "learning_rate": 1.4860000000000002e-05,
      "loss": 1.2141,
      "step": 743
    },
    {
      "epoch": 0.2215975725534727,
      "grad_norm": 0.8510664701461792,
      "learning_rate": 1.4880000000000002e-05,
      "loss": 1.2196,
      "step": 744
    },
    {
      "epoch": 0.22189541875314134,
      "grad_norm": 0.9481486678123474,
      "learning_rate": 1.4900000000000001e-05,
      "loss": 1.2263,
      "step": 745
    },
    {
      "epoch": 0.22219326495281,
      "grad_norm": 0.8476606607437134,
      "learning_rate": 1.4920000000000001e-05,
      "loss": 1.1982,
      "step": 746
    },
    {
      "epoch": 0.22249111115247863,
      "grad_norm": 0.8507776856422424,
      "learning_rate": 1.4940000000000001e-05,
      "loss": 1.1948,
      "step": 747
    },
    {
      "epoch": 0.2227889573521473,
      "grad_norm": 0.8643898963928223,
      "learning_rate": 1.496e-05,
      "loss": 1.2062,
      "step": 748
    },
    {
      "epoch": 0.22308680355181593,
      "grad_norm": 0.8651342391967773,
      "learning_rate": 1.498e-05,
      "loss": 1.2158,
      "step": 749
    },
    {
      "epoch": 0.2233846497514846,
      "grad_norm": 0.8332622647285461,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.2211,
      "step": 750
    },
    {
      "epoch": 0.22368249595115322,
      "grad_norm": 0.8560720086097717,
      "learning_rate": 1.5020000000000002e-05,
      "loss": 1.2014,
      "step": 751
    },
    {
      "epoch": 0.22398034215082188,
      "grad_norm": 0.837689220905304,
      "learning_rate": 1.5040000000000002e-05,
      "loss": 1.2029,
      "step": 752
    },
    {
      "epoch": 0.22427818835049051,
      "grad_norm": 0.8622657060623169,
      "learning_rate": 1.5060000000000001e-05,
      "loss": 1.1922,
      "step": 753
    },
    {
      "epoch": 0.22457603455015915,
      "grad_norm": 0.8364995121955872,
      "learning_rate": 1.5080000000000001e-05,
      "loss": 1.209,
      "step": 754
    },
    {
      "epoch": 0.2248738807498278,
      "grad_norm": 0.8895785808563232,
      "learning_rate": 1.5100000000000001e-05,
      "loss": 1.2205,
      "step": 755
    },
    {
      "epoch": 0.22517172694949644,
      "grad_norm": 0.868441641330719,
      "learning_rate": 1.5120000000000001e-05,
      "loss": 1.213,
      "step": 756
    },
    {
      "epoch": 0.2254695731491651,
      "grad_norm": 0.8537010550498962,
      "learning_rate": 1.514e-05,
      "loss": 1.1928,
      "step": 757
    },
    {
      "epoch": 0.22576741934883374,
      "grad_norm": 0.8216389417648315,
      "learning_rate": 1.516e-05,
      "loss": 1.2002,
      "step": 758
    },
    {
      "epoch": 0.2260652655485024,
      "grad_norm": 0.8112812638282776,
      "learning_rate": 1.5180000000000002e-05,
      "loss": 1.1963,
      "step": 759
    },
    {
      "epoch": 0.22636311174817103,
      "grad_norm": 0.8622816205024719,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 1.2067,
      "step": 760
    },
    {
      "epoch": 0.2266609579478397,
      "grad_norm": 0.7909964323043823,
      "learning_rate": 1.5220000000000002e-05,
      "loss": 1.1944,
      "step": 761
    },
    {
      "epoch": 0.22695880414750833,
      "grad_norm": 0.9343780875205994,
      "learning_rate": 1.5240000000000001e-05,
      "loss": 1.2176,
      "step": 762
    },
    {
      "epoch": 0.22725665034717699,
      "grad_norm": 0.8086036443710327,
      "learning_rate": 1.5260000000000003e-05,
      "loss": 1.2145,
      "step": 763
    },
    {
      "epoch": 0.22755449654684562,
      "grad_norm": 0.878097414970398,
      "learning_rate": 1.5280000000000003e-05,
      "loss": 1.2223,
      "step": 764
    },
    {
      "epoch": 0.22785234274651428,
      "grad_norm": 0.8477746248245239,
      "learning_rate": 1.5300000000000003e-05,
      "loss": 1.2025,
      "step": 765
    },
    {
      "epoch": 0.2281501889461829,
      "grad_norm": 0.9618206024169922,
      "learning_rate": 1.5320000000000002e-05,
      "loss": 1.2194,
      "step": 766
    },
    {
      "epoch": 0.22844803514585157,
      "grad_norm": 0.8408910632133484,
      "learning_rate": 1.5340000000000002e-05,
      "loss": 1.2019,
      "step": 767
    },
    {
      "epoch": 0.2287458813455202,
      "grad_norm": 0.9743706583976746,
      "learning_rate": 1.5360000000000002e-05,
      "loss": 1.2085,
      "step": 768
    },
    {
      "epoch": 0.22904372754518884,
      "grad_norm": 0.8604052066802979,
      "learning_rate": 1.5380000000000002e-05,
      "loss": 1.2027,
      "step": 769
    },
    {
      "epoch": 0.2293415737448575,
      "grad_norm": 1.0062904357910156,
      "learning_rate": 1.54e-05,
      "loss": 1.2083,
      "step": 770
    },
    {
      "epoch": 0.22963941994452614,
      "grad_norm": 0.8215556144714355,
      "learning_rate": 1.542e-05,
      "loss": 1.1928,
      "step": 771
    },
    {
      "epoch": 0.2299372661441948,
      "grad_norm": 0.9532316327095032,
      "learning_rate": 1.544e-05,
      "loss": 1.2084,
      "step": 772
    },
    {
      "epoch": 0.23023511234386343,
      "grad_norm": 0.8274873495101929,
      "learning_rate": 1.546e-05,
      "loss": 1.2066,
      "step": 773
    },
    {
      "epoch": 0.2305329585435321,
      "grad_norm": 0.9274572730064392,
      "learning_rate": 1.548e-05,
      "loss": 1.2232,
      "step": 774
    },
    {
      "epoch": 0.23083080474320072,
      "grad_norm": 0.8092347979545593,
      "learning_rate": 1.55e-05,
      "loss": 1.2044,
      "step": 775
    },
    {
      "epoch": 0.23112865094286938,
      "grad_norm": 0.9040752053260803,
      "learning_rate": 1.552e-05,
      "loss": 1.2122,
      "step": 776
    },
    {
      "epoch": 0.23142649714253802,
      "grad_norm": 0.8546992540359497,
      "learning_rate": 1.554e-05,
      "loss": 1.2151,
      "step": 777
    },
    {
      "epoch": 0.23172434334220668,
      "grad_norm": 0.9295095801353455,
      "learning_rate": 1.556e-05,
      "loss": 1.2102,
      "step": 778
    },
    {
      "epoch": 0.2320221895418753,
      "grad_norm": 0.8288260102272034,
      "learning_rate": 1.5580000000000003e-05,
      "loss": 1.2095,
      "step": 779
    },
    {
      "epoch": 0.23232003574154397,
      "grad_norm": 0.9736093878746033,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 1.2259,
      "step": 780
    },
    {
      "epoch": 0.2326178819412126,
      "grad_norm": 0.8532735109329224,
      "learning_rate": 1.5620000000000003e-05,
      "loss": 1.1978,
      "step": 781
    },
    {
      "epoch": 0.23291572814088124,
      "grad_norm": 0.8977347016334534,
      "learning_rate": 1.5640000000000003e-05,
      "loss": 1.2152,
      "step": 782
    },
    {
      "epoch": 0.2332135743405499,
      "grad_norm": 0.9258580803871155,
      "learning_rate": 1.5660000000000003e-05,
      "loss": 1.2065,
      "step": 783
    },
    {
      "epoch": 0.23351142054021853,
      "grad_norm": 0.8673878908157349,
      "learning_rate": 1.5680000000000002e-05,
      "loss": 1.2187,
      "step": 784
    },
    {
      "epoch": 0.2338092667398872,
      "grad_norm": 1.0159471035003662,
      "learning_rate": 1.5700000000000002e-05,
      "loss": 1.2041,
      "step": 785
    },
    {
      "epoch": 0.23410711293955583,
      "grad_norm": 0.8775511384010315,
      "learning_rate": 1.5720000000000002e-05,
      "loss": 1.2027,
      "step": 786
    },
    {
      "epoch": 0.2344049591392245,
      "grad_norm": 0.919565737247467,
      "learning_rate": 1.5740000000000002e-05,
      "loss": 1.2045,
      "step": 787
    },
    {
      "epoch": 0.23470280533889312,
      "grad_norm": 0.8786361217498779,
      "learning_rate": 1.576e-05,
      "loss": 1.2231,
      "step": 788
    },
    {
      "epoch": 0.23500065153856178,
      "grad_norm": 0.8838209509849548,
      "learning_rate": 1.578e-05,
      "loss": 1.2054,
      "step": 789
    },
    {
      "epoch": 0.23529849773823042,
      "grad_norm": 0.8410146236419678,
      "learning_rate": 1.58e-05,
      "loss": 1.2005,
      "step": 790
    },
    {
      "epoch": 0.23559634393789908,
      "grad_norm": 0.8445044755935669,
      "learning_rate": 1.582e-05,
      "loss": 1.2225,
      "step": 791
    },
    {
      "epoch": 0.2358941901375677,
      "grad_norm": 0.8243874311447144,
      "learning_rate": 1.584e-05,
      "loss": 1.2095,
      "step": 792
    },
    {
      "epoch": 0.23619203633723637,
      "grad_norm": 0.8203811049461365,
      "learning_rate": 1.586e-05,
      "loss": 1.1973,
      "step": 793
    },
    {
      "epoch": 0.236489882536905,
      "grad_norm": 0.8111893534660339,
      "learning_rate": 1.588e-05,
      "loss": 1.1952,
      "step": 794
    },
    {
      "epoch": 0.23678772873657364,
      "grad_norm": 0.8844137787818909,
      "learning_rate": 1.5900000000000004e-05,
      "loss": 1.2191,
      "step": 795
    },
    {
      "epoch": 0.2370855749362423,
      "grad_norm": 0.8572410941123962,
      "learning_rate": 1.5920000000000003e-05,
      "loss": 1.2184,
      "step": 796
    },
    {
      "epoch": 0.23738342113591093,
      "grad_norm": 0.8745017051696777,
      "learning_rate": 1.5940000000000003e-05,
      "loss": 1.209,
      "step": 797
    },
    {
      "epoch": 0.2376812673355796,
      "grad_norm": 0.8740428686141968,
      "learning_rate": 1.5960000000000003e-05,
      "loss": 1.2217,
      "step": 798
    },
    {
      "epoch": 0.23797911353524823,
      "grad_norm": 0.8808128237724304,
      "learning_rate": 1.5980000000000003e-05,
      "loss": 1.2117,
      "step": 799
    },
    {
      "epoch": 0.2382769597349169,
      "grad_norm": 0.8756093978881836,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.224,
      "step": 800
    },
    {
      "epoch": 0.23857480593458552,
      "grad_norm": 0.8535891771316528,
      "learning_rate": 1.6020000000000002e-05,
      "loss": 1.21,
      "step": 801
    },
    {
      "epoch": 0.23887265213425418,
      "grad_norm": 0.8931506872177124,
      "learning_rate": 1.6040000000000002e-05,
      "loss": 1.2082,
      "step": 802
    },
    {
      "epoch": 0.23917049833392282,
      "grad_norm": 0.8423414826393127,
      "learning_rate": 1.6060000000000002e-05,
      "loss": 1.2042,
      "step": 803
    },
    {
      "epoch": 0.23946834453359148,
      "grad_norm": 0.8824076652526855,
      "learning_rate": 1.6080000000000002e-05,
      "loss": 1.2129,
      "step": 804
    },
    {
      "epoch": 0.2397661907332601,
      "grad_norm": 0.8630048036575317,
      "learning_rate": 1.6100000000000002e-05,
      "loss": 1.1961,
      "step": 805
    },
    {
      "epoch": 0.24006403693292877,
      "grad_norm": 0.907599151134491,
      "learning_rate": 1.612e-05,
      "loss": 1.2236,
      "step": 806
    },
    {
      "epoch": 0.2403618831325974,
      "grad_norm": 0.9025611281394958,
      "learning_rate": 1.614e-05,
      "loss": 1.2154,
      "step": 807
    },
    {
      "epoch": 0.24065972933226606,
      "grad_norm": 0.8655932545661926,
      "learning_rate": 1.616e-05,
      "loss": 1.2205,
      "step": 808
    },
    {
      "epoch": 0.2409575755319347,
      "grad_norm": 0.8551158905029297,
      "learning_rate": 1.618e-05,
      "loss": 1.2035,
      "step": 809
    },
    {
      "epoch": 0.24125542173160333,
      "grad_norm": 0.8754732608795166,
      "learning_rate": 1.62e-05,
      "loss": 1.2022,
      "step": 810
    },
    {
      "epoch": 0.241553267931272,
      "grad_norm": 0.8469142913818359,
      "learning_rate": 1.6220000000000004e-05,
      "loss": 1.2156,
      "step": 811
    },
    {
      "epoch": 0.24185111413094063,
      "grad_norm": 0.9227187037467957,
      "learning_rate": 1.6240000000000004e-05,
      "loss": 1.2145,
      "step": 812
    },
    {
      "epoch": 0.2421489603306093,
      "grad_norm": 0.8639239072799683,
      "learning_rate": 1.626e-05,
      "loss": 1.2177,
      "step": 813
    },
    {
      "epoch": 0.24244680653027792,
      "grad_norm": 0.9511845707893372,
      "learning_rate": 1.628e-05,
      "loss": 1.2158,
      "step": 814
    },
    {
      "epoch": 0.24274465272994658,
      "grad_norm": 0.8417127728462219,
      "learning_rate": 1.63e-05,
      "loss": 1.2025,
      "step": 815
    },
    {
      "epoch": 0.2430424989296152,
      "grad_norm": 0.8147373795509338,
      "learning_rate": 1.632e-05,
      "loss": 1.2089,
      "step": 816
    },
    {
      "epoch": 0.24334034512928387,
      "grad_norm": 0.8666256070137024,
      "learning_rate": 1.634e-05,
      "loss": 1.2095,
      "step": 817
    },
    {
      "epoch": 0.2436381913289525,
      "grad_norm": 0.8636316657066345,
      "learning_rate": 1.636e-05,
      "loss": 1.1976,
      "step": 818
    },
    {
      "epoch": 0.24393603752862117,
      "grad_norm": 0.8087793588638306,
      "learning_rate": 1.638e-05,
      "loss": 1.2085,
      "step": 819
    },
    {
      "epoch": 0.2442338837282898,
      "grad_norm": 0.8696207404136658,
      "learning_rate": 1.64e-05,
      "loss": 1.2263,
      "step": 820
    },
    {
      "epoch": 0.24453172992795846,
      "grad_norm": 0.8158828616142273,
      "learning_rate": 1.6420000000000002e-05,
      "loss": 1.2123,
      "step": 821
    },
    {
      "epoch": 0.2448295761276271,
      "grad_norm": 0.8344365954399109,
      "learning_rate": 1.6440000000000002e-05,
      "loss": 1.1895,
      "step": 822
    },
    {
      "epoch": 0.24512742232729573,
      "grad_norm": 0.871940553188324,
      "learning_rate": 1.646e-05,
      "loss": 1.2061,
      "step": 823
    },
    {
      "epoch": 0.2454252685269644,
      "grad_norm": 0.8790808320045471,
      "learning_rate": 1.648e-05,
      "loss": 1.2166,
      "step": 824
    },
    {
      "epoch": 0.24572311472663302,
      "grad_norm": 0.8303203582763672,
      "learning_rate": 1.65e-05,
      "loss": 1.199,
      "step": 825
    },
    {
      "epoch": 0.24602096092630169,
      "grad_norm": 0.8460865616798401,
      "learning_rate": 1.652e-05,
      "loss": 1.2135,
      "step": 826
    },
    {
      "epoch": 0.24631880712597032,
      "grad_norm": 0.8239215612411499,
      "learning_rate": 1.654e-05,
      "loss": 1.2071,
      "step": 827
    },
    {
      "epoch": 0.24661665332563898,
      "grad_norm": 0.8556895852088928,
      "learning_rate": 1.656e-05,
      "loss": 1.1923,
      "step": 828
    },
    {
      "epoch": 0.2469144995253076,
      "grad_norm": 0.8784323930740356,
      "learning_rate": 1.658e-05,
      "loss": 1.2267,
      "step": 829
    },
    {
      "epoch": 0.24721234572497627,
      "grad_norm": 0.838881254196167,
      "learning_rate": 1.66e-05,
      "loss": 1.2251,
      "step": 830
    },
    {
      "epoch": 0.2475101919246449,
      "grad_norm": 0.8431991934776306,
      "learning_rate": 1.662e-05,
      "loss": 1.1932,
      "step": 831
    },
    {
      "epoch": 0.24780803812431357,
      "grad_norm": 0.7851248383522034,
      "learning_rate": 1.664e-05,
      "loss": 1.2043,
      "step": 832
    },
    {
      "epoch": 0.2481058843239822,
      "grad_norm": 0.8126862049102783,
      "learning_rate": 1.666e-05,
      "loss": 1.1849,
      "step": 833
    },
    {
      "epoch": 0.24840373052365086,
      "grad_norm": 0.8328498601913452,
      "learning_rate": 1.668e-05,
      "loss": 1.2141,
      "step": 834
    },
    {
      "epoch": 0.2487015767233195,
      "grad_norm": 0.8406838178634644,
      "learning_rate": 1.67e-05,
      "loss": 1.2043,
      "step": 835
    },
    {
      "epoch": 0.24899942292298813,
      "grad_norm": 0.8363956212997437,
      "learning_rate": 1.672e-05,
      "loss": 1.193,
      "step": 836
    },
    {
      "epoch": 0.2492972691226568,
      "grad_norm": 0.8512202501296997,
      "learning_rate": 1.6740000000000002e-05,
      "loss": 1.2038,
      "step": 837
    },
    {
      "epoch": 0.24959511532232542,
      "grad_norm": 0.8159688115119934,
      "learning_rate": 1.6760000000000002e-05,
      "loss": 1.2182,
      "step": 838
    },
    {
      "epoch": 0.24989296152199408,
      "grad_norm": 0.833097517490387,
      "learning_rate": 1.6780000000000002e-05,
      "loss": 1.1983,
      "step": 839
    },
    {
      "epoch": 0.25019080772166274,
      "grad_norm": 0.8369312882423401,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 1.1988,
      "step": 840
    },
    {
      "epoch": 0.25048865392133135,
      "grad_norm": 0.884112536907196,
      "learning_rate": 1.682e-05,
      "loss": 1.2213,
      "step": 841
    },
    {
      "epoch": 0.250786500121,
      "grad_norm": 0.8264943361282349,
      "learning_rate": 1.684e-05,
      "loss": 1.2097,
      "step": 842
    },
    {
      "epoch": 0.25108434632066867,
      "grad_norm": 0.8816318511962891,
      "learning_rate": 1.686e-05,
      "loss": 1.1963,
      "step": 843
    },
    {
      "epoch": 0.25138219252033733,
      "grad_norm": 0.7869309186935425,
      "learning_rate": 1.688e-05,
      "loss": 1.1938,
      "step": 844
    },
    {
      "epoch": 0.25168003872000594,
      "grad_norm": 0.8592996001243591,
      "learning_rate": 1.69e-05,
      "loss": 1.2221,
      "step": 845
    },
    {
      "epoch": 0.2519778849196746,
      "grad_norm": 0.7969801425933838,
      "learning_rate": 1.692e-05,
      "loss": 1.2029,
      "step": 846
    },
    {
      "epoch": 0.25227573111934326,
      "grad_norm": 0.862031102180481,
      "learning_rate": 1.694e-05,
      "loss": 1.201,
      "step": 847
    },
    {
      "epoch": 0.2525735773190119,
      "grad_norm": 0.7971358895301819,
      "learning_rate": 1.696e-05,
      "loss": 1.2025,
      "step": 848
    },
    {
      "epoch": 0.2528714235186805,
      "grad_norm": 0.8102288842201233,
      "learning_rate": 1.698e-05,
      "loss": 1.2034,
      "step": 849
    },
    {
      "epoch": 0.2531692697183492,
      "grad_norm": 0.7993308901786804,
      "learning_rate": 1.7e-05,
      "loss": 1.1965,
      "step": 850
    },
    {
      "epoch": 0.25346711591801785,
      "grad_norm": 0.7983496189117432,
      "learning_rate": 1.702e-05,
      "loss": 1.2118,
      "step": 851
    },
    {
      "epoch": 0.25376496211768645,
      "grad_norm": 0.8213385343551636,
      "learning_rate": 1.704e-05,
      "loss": 1.2025,
      "step": 852
    },
    {
      "epoch": 0.2540628083173551,
      "grad_norm": 0.783064067363739,
      "learning_rate": 1.7060000000000003e-05,
      "loss": 1.1981,
      "step": 853
    },
    {
      "epoch": 0.2543606545170238,
      "grad_norm": 0.7969495058059692,
      "learning_rate": 1.7080000000000002e-05,
      "loss": 1.2109,
      "step": 854
    },
    {
      "epoch": 0.25465850071669244,
      "grad_norm": 0.7907369136810303,
      "learning_rate": 1.7100000000000002e-05,
      "loss": 1.2058,
      "step": 855
    },
    {
      "epoch": 0.25495634691636104,
      "grad_norm": 0.7799652218818665,
      "learning_rate": 1.7120000000000002e-05,
      "loss": 1.1995,
      "step": 856
    },
    {
      "epoch": 0.2552541931160297,
      "grad_norm": 0.8215786814689636,
      "learning_rate": 1.7140000000000002e-05,
      "loss": 1.2067,
      "step": 857
    },
    {
      "epoch": 0.25555203931569836,
      "grad_norm": 0.7765409350395203,
      "learning_rate": 1.7160000000000002e-05,
      "loss": 1.2088,
      "step": 858
    },
    {
      "epoch": 0.255849885515367,
      "grad_norm": 0.8462494611740112,
      "learning_rate": 1.718e-05,
      "loss": 1.2104,
      "step": 859
    },
    {
      "epoch": 0.25614773171503563,
      "grad_norm": 0.83392733335495,
      "learning_rate": 1.72e-05,
      "loss": 1.2106,
      "step": 860
    },
    {
      "epoch": 0.2564455779147043,
      "grad_norm": 0.8329858183860779,
      "learning_rate": 1.722e-05,
      "loss": 1.1969,
      "step": 861
    },
    {
      "epoch": 0.25674342411437295,
      "grad_norm": 0.8081089854240417,
      "learning_rate": 1.724e-05,
      "loss": 1.1964,
      "step": 862
    },
    {
      "epoch": 0.2570412703140416,
      "grad_norm": 0.7642909288406372,
      "learning_rate": 1.726e-05,
      "loss": 1.2135,
      "step": 863
    },
    {
      "epoch": 0.2573391165137102,
      "grad_norm": 0.8370653390884399,
      "learning_rate": 1.728e-05,
      "loss": 1.2201,
      "step": 864
    },
    {
      "epoch": 0.2576369627133789,
      "grad_norm": 0.8088498711585999,
      "learning_rate": 1.73e-05,
      "loss": 1.1889,
      "step": 865
    },
    {
      "epoch": 0.25793480891304754,
      "grad_norm": 0.7948449850082397,
      "learning_rate": 1.732e-05,
      "loss": 1.2056,
      "step": 866
    },
    {
      "epoch": 0.25823265511271615,
      "grad_norm": 0.7823464274406433,
      "learning_rate": 1.734e-05,
      "loss": 1.1988,
      "step": 867
    },
    {
      "epoch": 0.2585305013123848,
      "grad_norm": 0.773443877696991,
      "learning_rate": 1.736e-05,
      "loss": 1.1902,
      "step": 868
    },
    {
      "epoch": 0.25882834751205347,
      "grad_norm": 0.8593543767929077,
      "learning_rate": 1.7380000000000003e-05,
      "loss": 1.2041,
      "step": 869
    },
    {
      "epoch": 0.25912619371172213,
      "grad_norm": 0.8284367322921753,
      "learning_rate": 1.7400000000000003e-05,
      "loss": 1.1982,
      "step": 870
    },
    {
      "epoch": 0.25942403991139074,
      "grad_norm": 0.802595853805542,
      "learning_rate": 1.7420000000000003e-05,
      "loss": 1.197,
      "step": 871
    },
    {
      "epoch": 0.2597218861110594,
      "grad_norm": 0.8641994595527649,
      "learning_rate": 1.7440000000000002e-05,
      "loss": 1.202,
      "step": 872
    },
    {
      "epoch": 0.26001973231072806,
      "grad_norm": 0.8463013768196106,
      "learning_rate": 1.7460000000000002e-05,
      "loss": 1.201,
      "step": 873
    },
    {
      "epoch": 0.2603175785103967,
      "grad_norm": 0.8413798213005066,
      "learning_rate": 1.7480000000000002e-05,
      "loss": 1.2076,
      "step": 874
    },
    {
      "epoch": 0.2606154247100653,
      "grad_norm": 0.8585672378540039,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 1.2019,
      "step": 875
    },
    {
      "epoch": 0.260913270909734,
      "grad_norm": 0.8081045746803284,
      "learning_rate": 1.752e-05,
      "loss": 1.2025,
      "step": 876
    },
    {
      "epoch": 0.26121111710940265,
      "grad_norm": 0.8568826913833618,
      "learning_rate": 1.754e-05,
      "loss": 1.1943,
      "step": 877
    },
    {
      "epoch": 0.26150896330907125,
      "grad_norm": 0.8199113607406616,
      "learning_rate": 1.756e-05,
      "loss": 1.197,
      "step": 878
    },
    {
      "epoch": 0.2618068095087399,
      "grad_norm": 0.8933385610580444,
      "learning_rate": 1.758e-05,
      "loss": 1.213,
      "step": 879
    },
    {
      "epoch": 0.2621046557084086,
      "grad_norm": 0.8476577401161194,
      "learning_rate": 1.76e-05,
      "loss": 1.2001,
      "step": 880
    },
    {
      "epoch": 0.26240250190807723,
      "grad_norm": 0.9339004755020142,
      "learning_rate": 1.762e-05,
      "loss": 1.2284,
      "step": 881
    },
    {
      "epoch": 0.26270034810774584,
      "grad_norm": 0.845933198928833,
      "learning_rate": 1.764e-05,
      "loss": 1.2005,
      "step": 882
    },
    {
      "epoch": 0.2629981943074145,
      "grad_norm": 0.8759322762489319,
      "learning_rate": 1.766e-05,
      "loss": 1.2091,
      "step": 883
    },
    {
      "epoch": 0.26329604050708316,
      "grad_norm": 0.8563612699508667,
      "learning_rate": 1.768e-05,
      "loss": 1.2183,
      "step": 884
    },
    {
      "epoch": 0.2635938867067518,
      "grad_norm": 0.8960975408554077,
      "learning_rate": 1.77e-05,
      "loss": 1.2022,
      "step": 885
    },
    {
      "epoch": 0.26389173290642043,
      "grad_norm": 0.8683968186378479,
      "learning_rate": 1.7720000000000003e-05,
      "loss": 1.2089,
      "step": 886
    },
    {
      "epoch": 0.2641895791060891,
      "grad_norm": 0.8820282220840454,
      "learning_rate": 1.7740000000000003e-05,
      "loss": 1.2222,
      "step": 887
    },
    {
      "epoch": 0.26448742530575775,
      "grad_norm": 0.834923267364502,
      "learning_rate": 1.7760000000000003e-05,
      "loss": 1.1928,
      "step": 888
    },
    {
      "epoch": 0.2647852715054264,
      "grad_norm": 0.850400447845459,
      "learning_rate": 1.7780000000000003e-05,
      "loss": 1.2092,
      "step": 889
    },
    {
      "epoch": 0.265083117705095,
      "grad_norm": 0.8835453391075134,
      "learning_rate": 1.7800000000000002e-05,
      "loss": 1.1973,
      "step": 890
    },
    {
      "epoch": 0.2653809639047637,
      "grad_norm": 0.7784873843193054,
      "learning_rate": 1.7820000000000002e-05,
      "loss": 1.2208,
      "step": 891
    },
    {
      "epoch": 0.26567881010443234,
      "grad_norm": 0.8298808336257935,
      "learning_rate": 1.7840000000000002e-05,
      "loss": 1.1982,
      "step": 892
    },
    {
      "epoch": 0.26597665630410094,
      "grad_norm": 0.8173578977584839,
      "learning_rate": 1.7860000000000002e-05,
      "loss": 1.1895,
      "step": 893
    },
    {
      "epoch": 0.2662745025037696,
      "grad_norm": 0.8170942068099976,
      "learning_rate": 1.788e-05,
      "loss": 1.2065,
      "step": 894
    },
    {
      "epoch": 0.26657234870343827,
      "grad_norm": 0.8495282530784607,
      "learning_rate": 1.79e-05,
      "loss": 1.1947,
      "step": 895
    },
    {
      "epoch": 0.2668701949031069,
      "grad_norm": 0.8125613331794739,
      "learning_rate": 1.792e-05,
      "loss": 1.2014,
      "step": 896
    },
    {
      "epoch": 0.26716804110277553,
      "grad_norm": 0.8314772248268127,
      "learning_rate": 1.794e-05,
      "loss": 1.1985,
      "step": 897
    },
    {
      "epoch": 0.2674658873024442,
      "grad_norm": 0.7989786267280579,
      "learning_rate": 1.796e-05,
      "loss": 1.2093,
      "step": 898
    },
    {
      "epoch": 0.26776373350211286,
      "grad_norm": 0.789775550365448,
      "learning_rate": 1.798e-05,
      "loss": 1.1977,
      "step": 899
    },
    {
      "epoch": 0.2680615797017815,
      "grad_norm": 0.7902028560638428,
      "learning_rate": 1.8e-05,
      "loss": 1.1916,
      "step": 900
    },
    {
      "epoch": 0.2683594259014501,
      "grad_norm": 0.7982574701309204,
      "learning_rate": 1.802e-05,
      "loss": 1.1923,
      "step": 901
    },
    {
      "epoch": 0.2686572721011188,
      "grad_norm": 0.8205264210700989,
      "learning_rate": 1.8040000000000003e-05,
      "loss": 1.2144,
      "step": 902
    },
    {
      "epoch": 0.26895511830078744,
      "grad_norm": 0.7552690505981445,
      "learning_rate": 1.8060000000000003e-05,
      "loss": 1.1815,
      "step": 903
    },
    {
      "epoch": 0.2692529645004561,
      "grad_norm": 0.780487596988678,
      "learning_rate": 1.8080000000000003e-05,
      "loss": 1.1939,
      "step": 904
    },
    {
      "epoch": 0.2695508107001247,
      "grad_norm": 0.7809114456176758,
      "learning_rate": 1.8100000000000003e-05,
      "loss": 1.1773,
      "step": 905
    },
    {
      "epoch": 0.26984865689979337,
      "grad_norm": 0.825038492679596,
      "learning_rate": 1.8120000000000003e-05,
      "loss": 1.1976,
      "step": 906
    },
    {
      "epoch": 0.27014650309946203,
      "grad_norm": 0.8322103023529053,
      "learning_rate": 1.8140000000000003e-05,
      "loss": 1.2149,
      "step": 907
    },
    {
      "epoch": 0.27044434929913064,
      "grad_norm": 0.8210577964782715,
      "learning_rate": 1.8160000000000002e-05,
      "loss": 1.2056,
      "step": 908
    },
    {
      "epoch": 0.2707421954987993,
      "grad_norm": 0.9032166004180908,
      "learning_rate": 1.8180000000000002e-05,
      "loss": 1.2119,
      "step": 909
    },
    {
      "epoch": 0.27104004169846796,
      "grad_norm": 0.7939310669898987,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 1.1978,
      "step": 910
    },
    {
      "epoch": 0.2713378878981366,
      "grad_norm": 0.8468285202980042,
      "learning_rate": 1.8220000000000002e-05,
      "loss": 1.1897,
      "step": 911
    },
    {
      "epoch": 0.2716357340978052,
      "grad_norm": 0.8237414956092834,
      "learning_rate": 1.824e-05,
      "loss": 1.216,
      "step": 912
    },
    {
      "epoch": 0.2719335802974739,
      "grad_norm": 0.8750325441360474,
      "learning_rate": 1.826e-05,
      "loss": 1.1882,
      "step": 913
    },
    {
      "epoch": 0.27223142649714255,
      "grad_norm": 0.8091866374015808,
      "learning_rate": 1.828e-05,
      "loss": 1.1978,
      "step": 914
    },
    {
      "epoch": 0.2725292726968112,
      "grad_norm": 0.8455181121826172,
      "learning_rate": 1.83e-05,
      "loss": 1.1945,
      "step": 915
    },
    {
      "epoch": 0.2728271188964798,
      "grad_norm": 0.8655949831008911,
      "learning_rate": 1.832e-05,
      "loss": 1.1883,
      "step": 916
    },
    {
      "epoch": 0.2731249650961485,
      "grad_norm": 0.8071688413619995,
      "learning_rate": 1.834e-05,
      "loss": 1.1947,
      "step": 917
    },
    {
      "epoch": 0.27342281129581714,
      "grad_norm": 0.8856038451194763,
      "learning_rate": 1.8360000000000004e-05,
      "loss": 1.2027,
      "step": 918
    },
    {
      "epoch": 0.27372065749548574,
      "grad_norm": 0.8375715017318726,
      "learning_rate": 1.8380000000000004e-05,
      "loss": 1.2041,
      "step": 919
    },
    {
      "epoch": 0.2740185036951544,
      "grad_norm": 0.9118662476539612,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 1.1968,
      "step": 920
    },
    {
      "epoch": 0.27431634989482306,
      "grad_norm": 0.8204622864723206,
      "learning_rate": 1.8420000000000003e-05,
      "loss": 1.1924,
      "step": 921
    },
    {
      "epoch": 0.2746141960944917,
      "grad_norm": 0.9562674164772034,
      "learning_rate": 1.8440000000000003e-05,
      "loss": 1.1805,
      "step": 922
    },
    {
      "epoch": 0.27491204229416033,
      "grad_norm": 0.8254942297935486,
      "learning_rate": 1.8460000000000003e-05,
      "loss": 1.2017,
      "step": 923
    },
    {
      "epoch": 0.275209888493829,
      "grad_norm": 0.9312728643417358,
      "learning_rate": 1.8480000000000003e-05,
      "loss": 1.196,
      "step": 924
    },
    {
      "epoch": 0.27550773469349765,
      "grad_norm": 0.8033220767974854,
      "learning_rate": 1.8500000000000002e-05,
      "loss": 1.2056,
      "step": 925
    },
    {
      "epoch": 0.2758055808931663,
      "grad_norm": 0.8654564023017883,
      "learning_rate": 1.8520000000000002e-05,
      "loss": 1.2112,
      "step": 926
    },
    {
      "epoch": 0.2761034270928349,
      "grad_norm": 0.825669527053833,
      "learning_rate": 1.8540000000000002e-05,
      "loss": 1.2239,
      "step": 927
    },
    {
      "epoch": 0.2764012732925036,
      "grad_norm": 0.8819819688796997,
      "learning_rate": 1.8560000000000002e-05,
      "loss": 1.2055,
      "step": 928
    },
    {
      "epoch": 0.27669911949217224,
      "grad_norm": 0.8378221988677979,
      "learning_rate": 1.858e-05,
      "loss": 1.1908,
      "step": 929
    },
    {
      "epoch": 0.2769969656918409,
      "grad_norm": 0.8161121010780334,
      "learning_rate": 1.86e-05,
      "loss": 1.1953,
      "step": 930
    },
    {
      "epoch": 0.2772948118915095,
      "grad_norm": 0.8052731156349182,
      "learning_rate": 1.862e-05,
      "loss": 1.1954,
      "step": 931
    },
    {
      "epoch": 0.27759265809117817,
      "grad_norm": 0.854401707649231,
      "learning_rate": 1.864e-05,
      "loss": 1.1882,
      "step": 932
    },
    {
      "epoch": 0.27789050429084683,
      "grad_norm": 0.8080146312713623,
      "learning_rate": 1.866e-05,
      "loss": 1.2061,
      "step": 933
    },
    {
      "epoch": 0.27818835049051543,
      "grad_norm": 0.8854711651802063,
      "learning_rate": 1.8680000000000004e-05,
      "loss": 1.1944,
      "step": 934
    },
    {
      "epoch": 0.2784861966901841,
      "grad_norm": 0.7640926241874695,
      "learning_rate": 1.8700000000000004e-05,
      "loss": 1.2088,
      "step": 935
    },
    {
      "epoch": 0.27878404288985276,
      "grad_norm": 0.8555243015289307,
      "learning_rate": 1.8720000000000004e-05,
      "loss": 1.1837,
      "step": 936
    },
    {
      "epoch": 0.2790818890895214,
      "grad_norm": 0.7751480340957642,
      "learning_rate": 1.8740000000000004e-05,
      "loss": 1.2042,
      "step": 937
    },
    {
      "epoch": 0.27937973528919,
      "grad_norm": 0.8217097520828247,
      "learning_rate": 1.876e-05,
      "loss": 1.1981,
      "step": 938
    },
    {
      "epoch": 0.2796775814888587,
      "grad_norm": 0.7984431385993958,
      "learning_rate": 1.878e-05,
      "loss": 1.1837,
      "step": 939
    },
    {
      "epoch": 0.27997542768852735,
      "grad_norm": 0.8163536787033081,
      "learning_rate": 1.88e-05,
      "loss": 1.1918,
      "step": 940
    },
    {
      "epoch": 0.280273273888196,
      "grad_norm": 0.8139321208000183,
      "learning_rate": 1.882e-05,
      "loss": 1.202,
      "step": 941
    },
    {
      "epoch": 0.2805711200878646,
      "grad_norm": 0.8381873369216919,
      "learning_rate": 1.884e-05,
      "loss": 1.1942,
      "step": 942
    },
    {
      "epoch": 0.2808689662875333,
      "grad_norm": 0.851736307144165,
      "learning_rate": 1.886e-05,
      "loss": 1.1954,
      "step": 943
    },
    {
      "epoch": 0.28116681248720193,
      "grad_norm": 0.8058578968048096,
      "learning_rate": 1.8880000000000002e-05,
      "loss": 1.2009,
      "step": 944
    },
    {
      "epoch": 0.2814646586868706,
      "grad_norm": 0.925950825214386,
      "learning_rate": 1.8900000000000002e-05,
      "loss": 1.199,
      "step": 945
    },
    {
      "epoch": 0.2817625048865392,
      "grad_norm": 0.8057994246482849,
      "learning_rate": 1.8920000000000002e-05,
      "loss": 1.1854,
      "step": 946
    },
    {
      "epoch": 0.28206035108620786,
      "grad_norm": 0.9160003066062927,
      "learning_rate": 1.894e-05,
      "loss": 1.1944,
      "step": 947
    },
    {
      "epoch": 0.2823581972858765,
      "grad_norm": 0.8122429251670837,
      "learning_rate": 1.896e-05,
      "loss": 1.2012,
      "step": 948
    },
    {
      "epoch": 0.2826560434855451,
      "grad_norm": 0.809448778629303,
      "learning_rate": 1.898e-05,
      "loss": 1.1988,
      "step": 949
    },
    {
      "epoch": 0.2829538896852138,
      "grad_norm": 0.8422517776489258,
      "learning_rate": 1.9e-05,
      "loss": 1.2024,
      "step": 950
    },
    {
      "epoch": 0.28325173588488245,
      "grad_norm": 0.7887592315673828,
      "learning_rate": 1.902e-05,
      "loss": 1.1879,
      "step": 951
    },
    {
      "epoch": 0.2835495820845511,
      "grad_norm": 0.8457075357437134,
      "learning_rate": 1.904e-05,
      "loss": 1.1964,
      "step": 952
    },
    {
      "epoch": 0.2838474282842197,
      "grad_norm": 0.809822678565979,
      "learning_rate": 1.906e-05,
      "loss": 1.2064,
      "step": 953
    },
    {
      "epoch": 0.2841452744838884,
      "grad_norm": 0.8593079447746277,
      "learning_rate": 1.908e-05,
      "loss": 1.2058,
      "step": 954
    },
    {
      "epoch": 0.28444312068355704,
      "grad_norm": 0.8395040035247803,
      "learning_rate": 1.91e-05,
      "loss": 1.2005,
      "step": 955
    },
    {
      "epoch": 0.2847409668832257,
      "grad_norm": 0.8251676559448242,
      "learning_rate": 1.912e-05,
      "loss": 1.2074,
      "step": 956
    },
    {
      "epoch": 0.2850388130828943,
      "grad_norm": 0.8459216356277466,
      "learning_rate": 1.914e-05,
      "loss": 1.2043,
      "step": 957
    },
    {
      "epoch": 0.28533665928256297,
      "grad_norm": 0.7983732223510742,
      "learning_rate": 1.916e-05,
      "loss": 1.1994,
      "step": 958
    },
    {
      "epoch": 0.2856345054822316,
      "grad_norm": 0.8408694267272949,
      "learning_rate": 1.918e-05,
      "loss": 1.2008,
      "step": 959
    },
    {
      "epoch": 0.28593235168190023,
      "grad_norm": 0.8495489954948425,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 1.2016,
      "step": 960
    },
    {
      "epoch": 0.2862301978815689,
      "grad_norm": 0.8151966333389282,
      "learning_rate": 1.9220000000000002e-05,
      "loss": 1.1851,
      "step": 961
    },
    {
      "epoch": 0.28652804408123755,
      "grad_norm": 0.8616309762001038,
      "learning_rate": 1.9240000000000002e-05,
      "loss": 1.2075,
      "step": 962
    },
    {
      "epoch": 0.2868258902809062,
      "grad_norm": 0.8756716251373291,
      "learning_rate": 1.9260000000000002e-05,
      "loss": 1.2062,
      "step": 963
    },
    {
      "epoch": 0.2871237364805748,
      "grad_norm": 0.7794026136398315,
      "learning_rate": 1.9280000000000002e-05,
      "loss": 1.2102,
      "step": 964
    },
    {
      "epoch": 0.2874215826802435,
      "grad_norm": 0.9570958614349365,
      "learning_rate": 1.93e-05,
      "loss": 1.2025,
      "step": 965
    },
    {
      "epoch": 0.28771942887991214,
      "grad_norm": 0.8152825236320496,
      "learning_rate": 1.932e-05,
      "loss": 1.1975,
      "step": 966
    },
    {
      "epoch": 0.2880172750795808,
      "grad_norm": 0.8646731376647949,
      "learning_rate": 1.934e-05,
      "loss": 1.1899,
      "step": 967
    },
    {
      "epoch": 0.2883151212792494,
      "grad_norm": 0.8867997527122498,
      "learning_rate": 1.936e-05,
      "loss": 1.2114,
      "step": 968
    },
    {
      "epoch": 0.28861296747891807,
      "grad_norm": 0.8051161766052246,
      "learning_rate": 1.938e-05,
      "loss": 1.2102,
      "step": 969
    },
    {
      "epoch": 0.28891081367858673,
      "grad_norm": 0.8601623177528381,
      "learning_rate": 1.94e-05,
      "loss": 1.2079,
      "step": 970
    },
    {
      "epoch": 0.2892086598782554,
      "grad_norm": 0.7886808514595032,
      "learning_rate": 1.942e-05,
      "loss": 1.1818,
      "step": 971
    },
    {
      "epoch": 0.289506506077924,
      "grad_norm": 0.8170090317726135,
      "learning_rate": 1.944e-05,
      "loss": 1.211,
      "step": 972
    },
    {
      "epoch": 0.28980435227759266,
      "grad_norm": 0.8385158777236938,
      "learning_rate": 1.946e-05,
      "loss": 1.1909,
      "step": 973
    },
    {
      "epoch": 0.2901021984772613,
      "grad_norm": 0.7577624320983887,
      "learning_rate": 1.948e-05,
      "loss": 1.2007,
      "step": 974
    },
    {
      "epoch": 0.2904000446769299,
      "grad_norm": 0.874447762966156,
      "learning_rate": 1.95e-05,
      "loss": 1.1983,
      "step": 975
    },
    {
      "epoch": 0.2906978908765986,
      "grad_norm": 0.7828870415687561,
      "learning_rate": 1.9520000000000003e-05,
      "loss": 1.1903,
      "step": 976
    },
    {
      "epoch": 0.29099573707626725,
      "grad_norm": 0.803607702255249,
      "learning_rate": 1.9540000000000003e-05,
      "loss": 1.2196,
      "step": 977
    },
    {
      "epoch": 0.2912935832759359,
      "grad_norm": 0.7986970543861389,
      "learning_rate": 1.9560000000000002e-05,
      "loss": 1.1859,
      "step": 978
    },
    {
      "epoch": 0.2915914294756045,
      "grad_norm": 0.7709547281265259,
      "learning_rate": 1.9580000000000002e-05,
      "loss": 1.1925,
      "step": 979
    },
    {
      "epoch": 0.2918892756752732,
      "grad_norm": 0.8195174336433411,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 1.1901,
      "step": 980
    },
    {
      "epoch": 0.29218712187494184,
      "grad_norm": 0.7611882090568542,
      "learning_rate": 1.9620000000000002e-05,
      "loss": 1.1891,
      "step": 981
    },
    {
      "epoch": 0.2924849680746105,
      "grad_norm": 0.7673640251159668,
      "learning_rate": 1.9640000000000002e-05,
      "loss": 1.1834,
      "step": 982
    },
    {
      "epoch": 0.2927828142742791,
      "grad_norm": 0.8106610774993896,
      "learning_rate": 1.966e-05,
      "loss": 1.1911,
      "step": 983
    },
    {
      "epoch": 0.29308066047394776,
      "grad_norm": 0.7877560257911682,
      "learning_rate": 1.968e-05,
      "loss": 1.1974,
      "step": 984
    },
    {
      "epoch": 0.2933785066736164,
      "grad_norm": 0.8275457620620728,
      "learning_rate": 1.97e-05,
      "loss": 1.2069,
      "step": 985
    },
    {
      "epoch": 0.2936763528732851,
      "grad_norm": 0.7886846661567688,
      "learning_rate": 1.972e-05,
      "loss": 1.2154,
      "step": 986
    },
    {
      "epoch": 0.2939741990729537,
      "grad_norm": 0.8117048144340515,
      "learning_rate": 1.974e-05,
      "loss": 1.2013,
      "step": 987
    },
    {
      "epoch": 0.29427204527262235,
      "grad_norm": 0.7913259863853455,
      "learning_rate": 1.976e-05,
      "loss": 1.1976,
      "step": 988
    },
    {
      "epoch": 0.294569891472291,
      "grad_norm": 0.8291166424751282,
      "learning_rate": 1.978e-05,
      "loss": 1.1867,
      "step": 989
    },
    {
      "epoch": 0.2948677376719596,
      "grad_norm": 0.7826667428016663,
      "learning_rate": 1.98e-05,
      "loss": 1.2058,
      "step": 990
    },
    {
      "epoch": 0.2951655838716283,
      "grad_norm": 0.8778476715087891,
      "learning_rate": 1.982e-05,
      "loss": 1.2031,
      "step": 991
    },
    {
      "epoch": 0.29546343007129694,
      "grad_norm": 0.8096939325332642,
      "learning_rate": 1.9840000000000003e-05,
      "loss": 1.2012,
      "step": 992
    },
    {
      "epoch": 0.2957612762709656,
      "grad_norm": 0.8248263597488403,
      "learning_rate": 1.9860000000000003e-05,
      "loss": 1.1945,
      "step": 993
    },
    {
      "epoch": 0.2960591224706342,
      "grad_norm": 0.8173134326934814,
      "learning_rate": 1.9880000000000003e-05,
      "loss": 1.2022,
      "step": 994
    },
    {
      "epoch": 0.29635696867030287,
      "grad_norm": 0.8006213307380676,
      "learning_rate": 1.9900000000000003e-05,
      "loss": 1.196,
      "step": 995
    },
    {
      "epoch": 0.29665481486997153,
      "grad_norm": 0.7770216464996338,
      "learning_rate": 1.9920000000000002e-05,
      "loss": 1.1956,
      "step": 996
    },
    {
      "epoch": 0.2969526610696402,
      "grad_norm": 0.7982072234153748,
      "learning_rate": 1.9940000000000002e-05,
      "loss": 1.2059,
      "step": 997
    },
    {
      "epoch": 0.2972505072693088,
      "grad_norm": 0.7886027693748474,
      "learning_rate": 1.9960000000000002e-05,
      "loss": 1.1974,
      "step": 998
    },
    {
      "epoch": 0.29754835346897746,
      "grad_norm": 0.8030641674995422,
      "learning_rate": 1.9980000000000002e-05,
      "loss": 1.2043,
      "step": 999
    },
    {
      "epoch": 0.2978461996686461,
      "grad_norm": 0.7905114889144897,
      "learning_rate": 2e-05,
      "loss": 1.2075,
      "step": 1000
    },
    {
      "epoch": 0.2978461996686461,
      "eval_loss": 1.2831172943115234,
      "eval_runtime": 14.8416,
      "eval_samples_per_second": 116.834,
      "eval_steps_per_second": 14.621,
      "step": 1000
    },
    {
      "epoch": 0.2981440458683147,
      "grad_norm": 0.7607181072235107,
      "learning_rate": 1.999999111718698e-05,
      "loss": 1.1921,
      "step": 1001
    },
    {
      "epoch": 0.2984418920679834,
      "grad_norm": 0.8085062503814697,
      "learning_rate": 1.999996446876371e-05,
      "loss": 1.2081,
      "step": 1002
    },
    {
      "epoch": 0.29873973826765204,
      "grad_norm": 0.7902792692184448,
      "learning_rate": 1.9999920054777522e-05,
      "loss": 1.1971,
      "step": 1003
    },
    {
      "epoch": 0.2990375844673207,
      "grad_norm": 0.7862740755081177,
      "learning_rate": 1.9999857875307324e-05,
      "loss": 1.1939,
      "step": 1004
    },
    {
      "epoch": 0.2993354306669893,
      "grad_norm": 0.7781235575675964,
      "learning_rate": 1.999977793046359e-05,
      "loss": 1.1843,
      "step": 1005
    },
    {
      "epoch": 0.29963327686665797,
      "grad_norm": 0.8072807192802429,
      "learning_rate": 1.999968022038833e-05,
      "loss": 1.1998,
      "step": 1006
    },
    {
      "epoch": 0.29993112306632663,
      "grad_norm": 0.8127691149711609,
      "learning_rate": 1.9999564745255148e-05,
      "loss": 1.2054,
      "step": 1007
    },
    {
      "epoch": 0.3002289692659953,
      "grad_norm": 0.9035406112670898,
      "learning_rate": 1.9999431505269185e-05,
      "loss": 1.2104,
      "step": 1008
    },
    {
      "epoch": 0.3005268154656639,
      "grad_norm": 0.7784050703048706,
      "learning_rate": 1.9999280500667154e-05,
      "loss": 1.1969,
      "step": 1009
    },
    {
      "epoch": 0.30082466166533256,
      "grad_norm": 0.8624287247657776,
      "learning_rate": 1.999911173171732e-05,
      "loss": 1.2059,
      "step": 1010
    },
    {
      "epoch": 0.3011225078650012,
      "grad_norm": 0.8037803769111633,
      "learning_rate": 1.9998925198719514e-05,
      "loss": 1.1913,
      "step": 1011
    },
    {
      "epoch": 0.3014203540646699,
      "grad_norm": 0.8232534527778625,
      "learning_rate": 1.9998720902005125e-05,
      "loss": 1.2038,
      "step": 1012
    },
    {
      "epoch": 0.3017182002643385,
      "grad_norm": 0.8869085907936096,
      "learning_rate": 1.999849884193709e-05,
      "loss": 1.2114,
      "step": 1013
    },
    {
      "epoch": 0.30201604646400715,
      "grad_norm": 0.8729265332221985,
      "learning_rate": 1.999825901890993e-05,
      "loss": 1.211,
      "step": 1014
    },
    {
      "epoch": 0.3023138926636758,
      "grad_norm": 0.9199550151824951,
      "learning_rate": 1.999800143334969e-05,
      "loss": 1.1907,
      "step": 1015
    },
    {
      "epoch": 0.3026117388633444,
      "grad_norm": 0.8800138831138611,
      "learning_rate": 1.9997726085713993e-05,
      "loss": 1.2057,
      "step": 1016
    },
    {
      "epoch": 0.3029095850630131,
      "grad_norm": 0.859698474407196,
      "learning_rate": 1.9997432976492006e-05,
      "loss": 1.198,
      "step": 1017
    },
    {
      "epoch": 0.30320743126268174,
      "grad_norm": 0.7960864305496216,
      "learning_rate": 1.9997122106204466e-05,
      "loss": 1.2039,
      "step": 1018
    },
    {
      "epoch": 0.3035052774623504,
      "grad_norm": 0.84726482629776,
      "learning_rate": 1.9996793475403647e-05,
      "loss": 1.1975,
      "step": 1019
    },
    {
      "epoch": 0.303803123662019,
      "grad_norm": 0.8081278800964355,
      "learning_rate": 1.9996447084673383e-05,
      "loss": 1.2077,
      "step": 1020
    },
    {
      "epoch": 0.30410096986168766,
      "grad_norm": 0.8523291945457458,
      "learning_rate": 1.9996082934629057e-05,
      "loss": 1.1975,
      "step": 1021
    },
    {
      "epoch": 0.3043988160613563,
      "grad_norm": 0.8435166478157043,
      "learning_rate": 1.9995701025917607e-05,
      "loss": 1.2017,
      "step": 1022
    },
    {
      "epoch": 0.304696662261025,
      "grad_norm": 0.7795689105987549,
      "learning_rate": 1.999530135921752e-05,
      "loss": 1.1988,
      "step": 1023
    },
    {
      "epoch": 0.3049945084606936,
      "grad_norm": 0.8221864700317383,
      "learning_rate": 1.9994883935238822e-05,
      "loss": 1.2065,
      "step": 1024
    },
    {
      "epoch": 0.30529235466036225,
      "grad_norm": 0.7459162473678589,
      "learning_rate": 1.9994448754723098e-05,
      "loss": 1.1898,
      "step": 1025
    },
    {
      "epoch": 0.3055902008600309,
      "grad_norm": 0.8057251572608948,
      "learning_rate": 1.9993995818443473e-05,
      "loss": 1.208,
      "step": 1026
    },
    {
      "epoch": 0.3058880470596996,
      "grad_norm": 0.8260543942451477,
      "learning_rate": 1.9993525127204615e-05,
      "loss": 1.1876,
      "step": 1027
    },
    {
      "epoch": 0.3061858932593682,
      "grad_norm": 0.8154152631759644,
      "learning_rate": 1.9993036681842737e-05,
      "loss": 1.2015,
      "step": 1028
    },
    {
      "epoch": 0.30648373945903684,
      "grad_norm": 0.8257317543029785,
      "learning_rate": 1.9992530483225596e-05,
      "loss": 1.1842,
      "step": 1029
    },
    {
      "epoch": 0.3067815856587055,
      "grad_norm": 0.8563246130943298,
      "learning_rate": 1.999200653225248e-05,
      "loss": 1.1843,
      "step": 1030
    },
    {
      "epoch": 0.3070794318583741,
      "grad_norm": 0.7984018921852112,
      "learning_rate": 1.9991464829854223e-05,
      "loss": 1.1956,
      "step": 1031
    },
    {
      "epoch": 0.30737727805804277,
      "grad_norm": 0.8103369474411011,
      "learning_rate": 1.999090537699319e-05,
      "loss": 1.1949,
      "step": 1032
    },
    {
      "epoch": 0.30767512425771143,
      "grad_norm": 0.8192564249038696,
      "learning_rate": 1.9990328174663297e-05,
      "loss": 1.2114,
      "step": 1033
    },
    {
      "epoch": 0.3079729704573801,
      "grad_norm": 0.8223491311073303,
      "learning_rate": 1.9989733223889964e-05,
      "loss": 1.1987,
      "step": 1034
    },
    {
      "epoch": 0.3082708166570487,
      "grad_norm": 0.8012005686759949,
      "learning_rate": 1.9989120525730165e-05,
      "loss": 1.2032,
      "step": 1035
    },
    {
      "epoch": 0.30856866285671736,
      "grad_norm": 0.8108108639717102,
      "learning_rate": 1.9988490081272397e-05,
      "loss": 1.193,
      "step": 1036
    },
    {
      "epoch": 0.308866509056386,
      "grad_norm": 0.7977895736694336,
      "learning_rate": 1.9987841891636687e-05,
      "loss": 1.1915,
      "step": 1037
    },
    {
      "epoch": 0.3091643552560547,
      "grad_norm": 0.8073302507400513,
      "learning_rate": 1.9987175957974577e-05,
      "loss": 1.1968,
      "step": 1038
    },
    {
      "epoch": 0.3094622014557233,
      "grad_norm": 0.8165181279182434,
      "learning_rate": 1.9986492281469147e-05,
      "loss": 1.1937,
      "step": 1039
    },
    {
      "epoch": 0.30976004765539195,
      "grad_norm": 0.8280035257339478,
      "learning_rate": 1.9985790863334988e-05,
      "loss": 1.1976,
      "step": 1040
    },
    {
      "epoch": 0.3100578938550606,
      "grad_norm": 0.879883348941803,
      "learning_rate": 1.998507170481821e-05,
      "loss": 1.1887,
      "step": 1041
    },
    {
      "epoch": 0.3103557400547292,
      "grad_norm": 0.8251213431358337,
      "learning_rate": 1.998433480719645e-05,
      "loss": 1.206,
      "step": 1042
    },
    {
      "epoch": 0.3106535862543979,
      "grad_norm": 0.7930659055709839,
      "learning_rate": 1.9983580171778846e-05,
      "loss": 1.1767,
      "step": 1043
    },
    {
      "epoch": 0.31095143245406653,
      "grad_norm": 0.782149612903595,
      "learning_rate": 1.9982807799906057e-05,
      "loss": 1.1902,
      "step": 1044
    },
    {
      "epoch": 0.3112492786537352,
      "grad_norm": 0.7847026586532593,
      "learning_rate": 1.9982017692950254e-05,
      "loss": 1.1881,
      "step": 1045
    },
    {
      "epoch": 0.3115471248534038,
      "grad_norm": 0.800574541091919,
      "learning_rate": 1.9981209852315108e-05,
      "loss": 1.1972,
      "step": 1046
    },
    {
      "epoch": 0.31184497105307246,
      "grad_norm": 0.8125137090682983,
      "learning_rate": 1.99803842794358e-05,
      "loss": 1.2182,
      "step": 1047
    },
    {
      "epoch": 0.3121428172527411,
      "grad_norm": 0.7842181324958801,
      "learning_rate": 1.997954097577901e-05,
      "loss": 1.1908,
      "step": 1048
    },
    {
      "epoch": 0.3124406634524098,
      "grad_norm": 0.7809070944786072,
      "learning_rate": 1.9978679942842922e-05,
      "loss": 1.1776,
      "step": 1049
    },
    {
      "epoch": 0.3127385096520784,
      "grad_norm": 0.775614321231842,
      "learning_rate": 1.9977801182157213e-05,
      "loss": 1.1829,
      "step": 1050
    },
    {
      "epoch": 0.31303635585174705,
      "grad_norm": 0.792310893535614,
      "learning_rate": 1.9976904695283057e-05,
      "loss": 1.1912,
      "step": 1051
    },
    {
      "epoch": 0.3133342020514157,
      "grad_norm": 0.8021408319473267,
      "learning_rate": 1.997599048381312e-05,
      "loss": 1.1917,
      "step": 1052
    },
    {
      "epoch": 0.31363204825108437,
      "grad_norm": 0.7507716417312622,
      "learning_rate": 1.9975058549371553e-05,
      "loss": 1.1864,
      "step": 1053
    },
    {
      "epoch": 0.313929894450753,
      "grad_norm": 0.7538429498672485,
      "learning_rate": 1.9974108893614e-05,
      "loss": 1.193,
      "step": 1054
    },
    {
      "epoch": 0.31422774065042164,
      "grad_norm": 0.7878537774085999,
      "learning_rate": 1.997314151822758e-05,
      "loss": 1.2004,
      "step": 1055
    },
    {
      "epoch": 0.3145255868500903,
      "grad_norm": 0.7723720073699951,
      "learning_rate": 1.9972156424930898e-05,
      "loss": 1.213,
      "step": 1056
    },
    {
      "epoch": 0.3148234330497589,
      "grad_norm": 0.7710390686988831,
      "learning_rate": 1.9971153615474036e-05,
      "loss": 1.2023,
      "step": 1057
    },
    {
      "epoch": 0.31512127924942757,
      "grad_norm": 0.7667935490608215,
      "learning_rate": 1.997013309163854e-05,
      "loss": 1.1669,
      "step": 1058
    },
    {
      "epoch": 0.3154191254490962,
      "grad_norm": 0.815008819103241,
      "learning_rate": 1.9969094855237446e-05,
      "loss": 1.1887,
      "step": 1059
    },
    {
      "epoch": 0.3157169716487649,
      "grad_norm": 0.7756442427635193,
      "learning_rate": 1.9968038908115237e-05,
      "loss": 1.1901,
      "step": 1060
    },
    {
      "epoch": 0.3160148178484335,
      "grad_norm": 0.7865918278694153,
      "learning_rate": 1.9966965252147873e-05,
      "loss": 1.181,
      "step": 1061
    },
    {
      "epoch": 0.31631266404810215,
      "grad_norm": 0.7754029035568237,
      "learning_rate": 1.9965873889242768e-05,
      "loss": 1.2099,
      "step": 1062
    },
    {
      "epoch": 0.3166105102477708,
      "grad_norm": 0.810185432434082,
      "learning_rate": 1.99647648213388e-05,
      "loss": 1.2029,
      "step": 1063
    },
    {
      "epoch": 0.3169083564474395,
      "grad_norm": 0.8116163611412048,
      "learning_rate": 1.9963638050406297e-05,
      "loss": 1.1925,
      "step": 1064
    },
    {
      "epoch": 0.3172062026471081,
      "grad_norm": 0.7758317589759827,
      "learning_rate": 1.9962493578447038e-05,
      "loss": 1.1895,
      "step": 1065
    },
    {
      "epoch": 0.31750404884677674,
      "grad_norm": 0.8024826049804688,
      "learning_rate": 1.9961331407494245e-05,
      "loss": 1.1818,
      "step": 1066
    },
    {
      "epoch": 0.3178018950464454,
      "grad_norm": 0.7944859862327576,
      "learning_rate": 1.9960151539612593e-05,
      "loss": 1.1851,
      "step": 1067
    },
    {
      "epoch": 0.31809974124611406,
      "grad_norm": 0.7910348176956177,
      "learning_rate": 1.995895397689819e-05,
      "loss": 1.1946,
      "step": 1068
    },
    {
      "epoch": 0.31839758744578267,
      "grad_norm": 0.8745784163475037,
      "learning_rate": 1.995773872147858e-05,
      "loss": 1.1903,
      "step": 1069
    },
    {
      "epoch": 0.31869543364545133,
      "grad_norm": 0.8198260068893433,
      "learning_rate": 1.995650577551274e-05,
      "loss": 1.1835,
      "step": 1070
    },
    {
      "epoch": 0.31899327984512,
      "grad_norm": 0.7996460199356079,
      "learning_rate": 1.9955255141191074e-05,
      "loss": 1.1876,
      "step": 1071
    },
    {
      "epoch": 0.3192911260447886,
      "grad_norm": 0.8283758163452148,
      "learning_rate": 1.9953986820735414e-05,
      "loss": 1.193,
      "step": 1072
    },
    {
      "epoch": 0.31958897224445726,
      "grad_norm": 0.8139001727104187,
      "learning_rate": 1.9952700816399014e-05,
      "loss": 1.1772,
      "step": 1073
    },
    {
      "epoch": 0.3198868184441259,
      "grad_norm": 0.8688188195228577,
      "learning_rate": 1.9951397130466535e-05,
      "loss": 1.1848,
      "step": 1074
    },
    {
      "epoch": 0.3201846646437946,
      "grad_norm": 0.7709712982177734,
      "learning_rate": 1.9950075765254063e-05,
      "loss": 1.2024,
      "step": 1075
    },
    {
      "epoch": 0.3204825108434632,
      "grad_norm": 0.7958956360816956,
      "learning_rate": 1.9948736723109082e-05,
      "loss": 1.1998,
      "step": 1076
    },
    {
      "epoch": 0.32078035704313185,
      "grad_norm": 0.8214638829231262,
      "learning_rate": 1.9947380006410484e-05,
      "loss": 1.2095,
      "step": 1077
    },
    {
      "epoch": 0.3210782032428005,
      "grad_norm": 0.8300702571868896,
      "learning_rate": 1.9946005617568563e-05,
      "loss": 1.1846,
      "step": 1078
    },
    {
      "epoch": 0.32137604944246917,
      "grad_norm": 0.8205470442771912,
      "learning_rate": 1.9944613559025005e-05,
      "loss": 1.1882,
      "step": 1079
    },
    {
      "epoch": 0.3216738956421378,
      "grad_norm": 0.900877833366394,
      "learning_rate": 1.994320383325289e-05,
      "loss": 1.2056,
      "step": 1080
    },
    {
      "epoch": 0.32197174184180644,
      "grad_norm": 0.8237030506134033,
      "learning_rate": 1.9941776442756685e-05,
      "loss": 1.2097,
      "step": 1081
    },
    {
      "epoch": 0.3222695880414751,
      "grad_norm": 0.8971453309059143,
      "learning_rate": 1.994033139007224e-05,
      "loss": 1.192,
      "step": 1082
    },
    {
      "epoch": 0.3225674342411437,
      "grad_norm": 0.7471580505371094,
      "learning_rate": 1.9938868677766778e-05,
      "loss": 1.1782,
      "step": 1083
    },
    {
      "epoch": 0.32286528044081236,
      "grad_norm": 0.901508629322052,
      "learning_rate": 1.99373883084389e-05,
      "loss": 1.1898,
      "step": 1084
    },
    {
      "epoch": 0.323163126640481,
      "grad_norm": 0.7786260843276978,
      "learning_rate": 1.9935890284718574e-05,
      "loss": 1.1984,
      "step": 1085
    },
    {
      "epoch": 0.3234609728401497,
      "grad_norm": 0.8482844233512878,
      "learning_rate": 1.993437460926714e-05,
      "loss": 1.2019,
      "step": 1086
    },
    {
      "epoch": 0.3237588190398183,
      "grad_norm": 0.7786232233047485,
      "learning_rate": 1.9932841284777277e-05,
      "loss": 1.1949,
      "step": 1087
    },
    {
      "epoch": 0.32405666523948695,
      "grad_norm": 0.8808726072311401,
      "learning_rate": 1.9931290313973043e-05,
      "loss": 1.1954,
      "step": 1088
    },
    {
      "epoch": 0.3243545114391556,
      "grad_norm": 0.8082985877990723,
      "learning_rate": 1.9929721699609828e-05,
      "loss": 1.1911,
      "step": 1089
    },
    {
      "epoch": 0.3246523576388243,
      "grad_norm": 0.7995214462280273,
      "learning_rate": 1.9928135444474382e-05,
      "loss": 1.1821,
      "step": 1090
    },
    {
      "epoch": 0.3249502038384929,
      "grad_norm": 0.7639725208282471,
      "learning_rate": 1.992653155138478e-05,
      "loss": 1.1684,
      "step": 1091
    },
    {
      "epoch": 0.32524805003816154,
      "grad_norm": 0.8273347020149231,
      "learning_rate": 1.9924910023190434e-05,
      "loss": 1.1789,
      "step": 1092
    },
    {
      "epoch": 0.3255458962378302,
      "grad_norm": 0.7688165307044983,
      "learning_rate": 1.9923270862772104e-05,
      "loss": 1.2132,
      "step": 1093
    },
    {
      "epoch": 0.32584374243749886,
      "grad_norm": 0.7990564107894897,
      "learning_rate": 1.9921614073041847e-05,
      "loss": 1.1972,
      "step": 1094
    },
    {
      "epoch": 0.32614158863716747,
      "grad_norm": 0.775524377822876,
      "learning_rate": 1.9919939656943062e-05,
      "loss": 1.1823,
      "step": 1095
    },
    {
      "epoch": 0.32643943483683613,
      "grad_norm": 0.7752482891082764,
      "learning_rate": 1.9918247617450454e-05,
      "loss": 1.1952,
      "step": 1096
    },
    {
      "epoch": 0.3267372810365048,
      "grad_norm": 0.8057593703269958,
      "learning_rate": 1.9916537957570035e-05,
      "loss": 1.2117,
      "step": 1097
    },
    {
      "epoch": 0.3270351272361734,
      "grad_norm": 0.8050331473350525,
      "learning_rate": 1.991481068033912e-05,
      "loss": 1.2032,
      "step": 1098
    },
    {
      "epoch": 0.32733297343584206,
      "grad_norm": 0.7533349990844727,
      "learning_rate": 1.991306578882633e-05,
      "loss": 1.197,
      "step": 1099
    },
    {
      "epoch": 0.3276308196355107,
      "grad_norm": 0.7805649042129517,
      "learning_rate": 1.9911303286131574e-05,
      "loss": 1.2017,
      "step": 1100
    },
    {
      "epoch": 0.3279286658351794,
      "grad_norm": 0.8294554948806763,
      "learning_rate": 1.990952317538604e-05,
      "loss": 1.1911,
      "step": 1101
    },
    {
      "epoch": 0.328226512034848,
      "grad_norm": 0.7915886044502258,
      "learning_rate": 1.990772545975222e-05,
      "loss": 1.194,
      "step": 1102
    },
    {
      "epoch": 0.32852435823451664,
      "grad_norm": 0.7856479287147522,
      "learning_rate": 1.9905910142423853e-05,
      "loss": 1.1822,
      "step": 1103
    },
    {
      "epoch": 0.3288222044341853,
      "grad_norm": 0.7837520837783813,
      "learning_rate": 1.9904077226625978e-05,
      "loss": 1.1955,
      "step": 1104
    },
    {
      "epoch": 0.32912005063385397,
      "grad_norm": 0.7791110873222351,
      "learning_rate": 1.9902226715614876e-05,
      "loss": 1.1946,
      "step": 1105
    },
    {
      "epoch": 0.32941789683352257,
      "grad_norm": 0.8523234128952026,
      "learning_rate": 1.99003586126781e-05,
      "loss": 1.1992,
      "step": 1106
    },
    {
      "epoch": 0.32971574303319123,
      "grad_norm": 0.7853846549987793,
      "learning_rate": 1.989847292113445e-05,
      "loss": 1.2013,
      "step": 1107
    },
    {
      "epoch": 0.3300135892328599,
      "grad_norm": 0.7816670536994934,
      "learning_rate": 1.989656964433397e-05,
      "loss": 1.1815,
      "step": 1108
    },
    {
      "epoch": 0.33031143543252856,
      "grad_norm": 0.7886641621589661,
      "learning_rate": 1.989464878565796e-05,
      "loss": 1.1908,
      "step": 1109
    },
    {
      "epoch": 0.33060928163219716,
      "grad_norm": 0.7743871808052063,
      "learning_rate": 1.989271034851894e-05,
      "loss": 1.2073,
      "step": 1110
    },
    {
      "epoch": 0.3309071278318658,
      "grad_norm": 0.762353241443634,
      "learning_rate": 1.9890754336360666e-05,
      "loss": 1.1754,
      "step": 1111
    },
    {
      "epoch": 0.3312049740315345,
      "grad_norm": 0.8019553422927856,
      "learning_rate": 1.9888780752658115e-05,
      "loss": 1.1876,
      "step": 1112
    },
    {
      "epoch": 0.3315028202312031,
      "grad_norm": 0.766575038433075,
      "learning_rate": 1.9886789600917483e-05,
      "loss": 1.1791,
      "step": 1113
    },
    {
      "epoch": 0.33180066643087175,
      "grad_norm": 0.7778606414794922,
      "learning_rate": 1.9884780884676177e-05,
      "loss": 1.1893,
      "step": 1114
    },
    {
      "epoch": 0.3320985126305404,
      "grad_norm": 0.7692984938621521,
      "learning_rate": 1.9882754607502807e-05,
      "loss": 1.1733,
      "step": 1115
    },
    {
      "epoch": 0.33239635883020907,
      "grad_norm": 0.7668749094009399,
      "learning_rate": 1.988071077299718e-05,
      "loss": 1.1809,
      "step": 1116
    },
    {
      "epoch": 0.3326942050298777,
      "grad_norm": 0.7663444876670837,
      "learning_rate": 1.9878649384790294e-05,
      "loss": 1.1866,
      "step": 1117
    },
    {
      "epoch": 0.33299205122954634,
      "grad_norm": 0.7777450084686279,
      "learning_rate": 1.9876570446544335e-05,
      "loss": 1.2111,
      "step": 1118
    },
    {
      "epoch": 0.333289897429215,
      "grad_norm": 0.7870537042617798,
      "learning_rate": 1.9874473961952672e-05,
      "loss": 1.1877,
      "step": 1119
    },
    {
      "epoch": 0.33358774362888366,
      "grad_norm": 0.7963652014732361,
      "learning_rate": 1.987235993473984e-05,
      "loss": 1.1869,
      "step": 1120
    },
    {
      "epoch": 0.33388558982855226,
      "grad_norm": 0.7689020037651062,
      "learning_rate": 1.987022836866153e-05,
      "loss": 1.2045,
      "step": 1121
    },
    {
      "epoch": 0.3341834360282209,
      "grad_norm": 0.7844902276992798,
      "learning_rate": 1.9868079267504615e-05,
      "loss": 1.1895,
      "step": 1122
    },
    {
      "epoch": 0.3344812822278896,
      "grad_norm": 0.7462410926818848,
      "learning_rate": 1.9865912635087105e-05,
      "loss": 1.1991,
      "step": 1123
    },
    {
      "epoch": 0.3347791284275582,
      "grad_norm": 0.7612064480781555,
      "learning_rate": 1.9863728475258156e-05,
      "loss": 1.2006,
      "step": 1124
    },
    {
      "epoch": 0.33507697462722685,
      "grad_norm": 0.7978454232215881,
      "learning_rate": 1.9861526791898062e-05,
      "loss": 1.1908,
      "step": 1125
    },
    {
      "epoch": 0.3353748208268955,
      "grad_norm": 0.7676396369934082,
      "learning_rate": 1.9859307588918258e-05,
      "loss": 1.1918,
      "step": 1126
    },
    {
      "epoch": 0.3356726670265642,
      "grad_norm": 0.8240189552307129,
      "learning_rate": 1.9857070870261292e-05,
      "loss": 1.2032,
      "step": 1127
    },
    {
      "epoch": 0.3359705132262328,
      "grad_norm": 0.7722442746162415,
      "learning_rate": 1.9854816639900837e-05,
      "loss": 1.1919,
      "step": 1128
    },
    {
      "epoch": 0.33626835942590144,
      "grad_norm": 0.7727859020233154,
      "learning_rate": 1.985254490184167e-05,
      "loss": 1.1898,
      "step": 1129
    },
    {
      "epoch": 0.3365662056255701,
      "grad_norm": 0.8555111885070801,
      "learning_rate": 1.9850255660119683e-05,
      "loss": 1.1791,
      "step": 1130
    },
    {
      "epoch": 0.33686405182523876,
      "grad_norm": 0.7550113201141357,
      "learning_rate": 1.9847948918801857e-05,
      "loss": 1.1867,
      "step": 1131
    },
    {
      "epoch": 0.33716189802490737,
      "grad_norm": 0.8665043115615845,
      "learning_rate": 1.9845624681986254e-05,
      "loss": 1.2061,
      "step": 1132
    },
    {
      "epoch": 0.33745974422457603,
      "grad_norm": 0.7852857708930969,
      "learning_rate": 1.984328295380203e-05,
      "loss": 1.1856,
      "step": 1133
    },
    {
      "epoch": 0.3377575904242447,
      "grad_norm": 0.8599435091018677,
      "learning_rate": 1.984092373840942e-05,
      "loss": 1.187,
      "step": 1134
    },
    {
      "epoch": 0.33805543662391335,
      "grad_norm": 0.7684341073036194,
      "learning_rate": 1.9838547039999706e-05,
      "loss": 1.1766,
      "step": 1135
    },
    {
      "epoch": 0.33835328282358196,
      "grad_norm": 0.8140760660171509,
      "learning_rate": 1.9836152862795245e-05,
      "loss": 1.1621,
      "step": 1136
    },
    {
      "epoch": 0.3386511290232506,
      "grad_norm": 0.7564947009086609,
      "learning_rate": 1.9833741211049448e-05,
      "loss": 1.1736,
      "step": 1137
    },
    {
      "epoch": 0.3389489752229193,
      "grad_norm": 0.7635712623596191,
      "learning_rate": 1.983131208904676e-05,
      "loss": 1.1964,
      "step": 1138
    },
    {
      "epoch": 0.3392468214225879,
      "grad_norm": 0.8048486113548279,
      "learning_rate": 1.982886550110267e-05,
      "loss": 1.2033,
      "step": 1139
    },
    {
      "epoch": 0.33954466762225655,
      "grad_norm": 0.7376336455345154,
      "learning_rate": 1.9826401451563693e-05,
      "loss": 1.1902,
      "step": 1140
    },
    {
      "epoch": 0.3398425138219252,
      "grad_norm": 0.7756757140159607,
      "learning_rate": 1.982391994480737e-05,
      "loss": 1.1828,
      "step": 1141
    },
    {
      "epoch": 0.34014036002159387,
      "grad_norm": 0.733124852180481,
      "learning_rate": 1.982142098524225e-05,
      "loss": 1.1811,
      "step": 1142
    },
    {
      "epoch": 0.3404382062212625,
      "grad_norm": 0.7445728778839111,
      "learning_rate": 1.9818904577307897e-05,
      "loss": 1.1939,
      "step": 1143
    },
    {
      "epoch": 0.34073605242093113,
      "grad_norm": 0.7567268013954163,
      "learning_rate": 1.981637072547486e-05,
      "loss": 1.1853,
      "step": 1144
    },
    {
      "epoch": 0.3410338986205998,
      "grad_norm": 0.7796471118927002,
      "learning_rate": 1.9813819434244687e-05,
      "loss": 1.1839,
      "step": 1145
    },
    {
      "epoch": 0.34133174482026846,
      "grad_norm": 0.7899624705314636,
      "learning_rate": 1.981125070814991e-05,
      "loss": 1.1939,
      "step": 1146
    },
    {
      "epoch": 0.34162959101993706,
      "grad_norm": 0.8063474297523499,
      "learning_rate": 1.980866455175403e-05,
      "loss": 1.1881,
      "step": 1147
    },
    {
      "epoch": 0.3419274372196057,
      "grad_norm": 0.7845880389213562,
      "learning_rate": 1.9806060969651514e-05,
      "loss": 1.1788,
      "step": 1148
    },
    {
      "epoch": 0.3422252834192744,
      "grad_norm": 0.7957763671875,
      "learning_rate": 1.9803439966467792e-05,
      "loss": 1.1887,
      "step": 1149
    },
    {
      "epoch": 0.34252312961894305,
      "grad_norm": 0.7520830035209656,
      "learning_rate": 1.9800801546859238e-05,
      "loss": 1.174,
      "step": 1150
    },
    {
      "epoch": 0.34282097581861165,
      "grad_norm": 0.8565261363983154,
      "learning_rate": 1.9798145715513168e-05,
      "loss": 1.1893,
      "step": 1151
    },
    {
      "epoch": 0.3431188220182803,
      "grad_norm": 0.7698721885681152,
      "learning_rate": 1.9795472477147836e-05,
      "loss": 1.1869,
      "step": 1152
    },
    {
      "epoch": 0.343416668217949,
      "grad_norm": 0.866442859172821,
      "learning_rate": 1.9792781836512418e-05,
      "loss": 1.1845,
      "step": 1153
    },
    {
      "epoch": 0.3437145144176176,
      "grad_norm": 0.8051164746284485,
      "learning_rate": 1.9790073798387003e-05,
      "loss": 1.204,
      "step": 1154
    },
    {
      "epoch": 0.34401236061728624,
      "grad_norm": 0.9243982434272766,
      "learning_rate": 1.9787348367582586e-05,
      "loss": 1.1921,
      "step": 1155
    },
    {
      "epoch": 0.3443102068169549,
      "grad_norm": 0.781558632850647,
      "learning_rate": 1.9784605548941074e-05,
      "loss": 1.1927,
      "step": 1156
    },
    {
      "epoch": 0.34460805301662356,
      "grad_norm": 0.876884937286377,
      "learning_rate": 1.9781845347335253e-05,
      "loss": 1.1828,
      "step": 1157
    },
    {
      "epoch": 0.34490589921629217,
      "grad_norm": 0.7638490200042725,
      "learning_rate": 1.9779067767668794e-05,
      "loss": 1.1829,
      "step": 1158
    },
    {
      "epoch": 0.3452037454159608,
      "grad_norm": 0.8828395009040833,
      "learning_rate": 1.9776272814876235e-05,
      "loss": 1.1745,
      "step": 1159
    },
    {
      "epoch": 0.3455015916156295,
      "grad_norm": 0.7557709217071533,
      "learning_rate": 1.9773460493922994e-05,
      "loss": 1.2005,
      "step": 1160
    },
    {
      "epoch": 0.34579943781529815,
      "grad_norm": 0.8470264673233032,
      "learning_rate": 1.977063080980533e-05,
      "loss": 1.2138,
      "step": 1161
    },
    {
      "epoch": 0.34609728401496676,
      "grad_norm": 0.7927517890930176,
      "learning_rate": 1.9767783767550358e-05,
      "loss": 1.169,
      "step": 1162
    },
    {
      "epoch": 0.3463951302146354,
      "grad_norm": 0.8599880933761597,
      "learning_rate": 1.976491937221602e-05,
      "loss": 1.1773,
      "step": 1163
    },
    {
      "epoch": 0.3466929764143041,
      "grad_norm": 0.7831829190254211,
      "learning_rate": 1.9762037628891097e-05,
      "loss": 1.1747,
      "step": 1164
    },
    {
      "epoch": 0.3469908226139727,
      "grad_norm": 0.7971581220626831,
      "learning_rate": 1.975913854269519e-05,
      "loss": 1.1791,
      "step": 1165
    },
    {
      "epoch": 0.34728866881364134,
      "grad_norm": 0.8192868232727051,
      "learning_rate": 1.9756222118778704e-05,
      "loss": 1.1822,
      "step": 1166
    },
    {
      "epoch": 0.34758651501331,
      "grad_norm": 0.8377344012260437,
      "learning_rate": 1.975328836232285e-05,
      "loss": 1.1787,
      "step": 1167
    },
    {
      "epoch": 0.34788436121297867,
      "grad_norm": 0.7946770787239075,
      "learning_rate": 1.9750337278539623e-05,
      "loss": 1.201,
      "step": 1168
    },
    {
      "epoch": 0.34818220741264727,
      "grad_norm": 0.8571863174438477,
      "learning_rate": 1.974736887267182e-05,
      "loss": 1.1811,
      "step": 1169
    },
    {
      "epoch": 0.34848005361231593,
      "grad_norm": 0.7696424126625061,
      "learning_rate": 1.974438314999299e-05,
      "loss": 1.2087,
      "step": 1170
    },
    {
      "epoch": 0.3487778998119846,
      "grad_norm": 0.8254051208496094,
      "learning_rate": 1.974138011580746e-05,
      "loss": 1.1791,
      "step": 1171
    },
    {
      "epoch": 0.34907574601165325,
      "grad_norm": 0.8082626461982727,
      "learning_rate": 1.9738359775450313e-05,
      "loss": 1.1878,
      "step": 1172
    },
    {
      "epoch": 0.34937359221132186,
      "grad_norm": 0.79220050573349,
      "learning_rate": 1.9735322134287364e-05,
      "loss": 1.1742,
      "step": 1173
    },
    {
      "epoch": 0.3496714384109905,
      "grad_norm": 0.7713173627853394,
      "learning_rate": 1.9732267197715176e-05,
      "loss": 1.1755,
      "step": 1174
    },
    {
      "epoch": 0.3499692846106592,
      "grad_norm": 0.7721149325370789,
      "learning_rate": 1.972919497116104e-05,
      "loss": 1.1951,
      "step": 1175
    },
    {
      "epoch": 0.35026713081032784,
      "grad_norm": 0.7143570184707642,
      "learning_rate": 1.972610546008295e-05,
      "loss": 1.1679,
      "step": 1176
    },
    {
      "epoch": 0.35056497700999645,
      "grad_norm": 0.7622668743133545,
      "learning_rate": 1.9722998669969626e-05,
      "loss": 1.1818,
      "step": 1177
    },
    {
      "epoch": 0.3508628232096651,
      "grad_norm": 0.7875356078147888,
      "learning_rate": 1.971987460634046e-05,
      "loss": 1.1916,
      "step": 1178
    },
    {
      "epoch": 0.35116066940933377,
      "grad_norm": 0.7511767148971558,
      "learning_rate": 1.9716733274745566e-05,
      "loss": 1.185,
      "step": 1179
    },
    {
      "epoch": 0.3514585156090024,
      "grad_norm": 0.7700490951538086,
      "learning_rate": 1.97135746807657e-05,
      "loss": 1.1924,
      "step": 1180
    },
    {
      "epoch": 0.35175636180867104,
      "grad_norm": 0.8142291307449341,
      "learning_rate": 1.9710398830012313e-05,
      "loss": 1.2015,
      "step": 1181
    },
    {
      "epoch": 0.3520542080083397,
      "grad_norm": 0.7582557201385498,
      "learning_rate": 1.9707205728127496e-05,
      "loss": 1.1875,
      "step": 1182
    },
    {
      "epoch": 0.35235205420800836,
      "grad_norm": 0.7595452070236206,
      "learning_rate": 1.9703995380783993e-05,
      "loss": 1.1875,
      "step": 1183
    },
    {
      "epoch": 0.35264990040767696,
      "grad_norm": 0.7503029108047485,
      "learning_rate": 1.9700767793685195e-05,
      "loss": 1.176,
      "step": 1184
    },
    {
      "epoch": 0.3529477466073456,
      "grad_norm": 0.7393288612365723,
      "learning_rate": 1.9697522972565103e-05,
      "loss": 1.187,
      "step": 1185
    },
    {
      "epoch": 0.3532455928070143,
      "grad_norm": 0.749775230884552,
      "learning_rate": 1.9694260923188354e-05,
      "loss": 1.1896,
      "step": 1186
    },
    {
      "epoch": 0.35354343900668295,
      "grad_norm": 0.7662206292152405,
      "learning_rate": 1.969098165135018e-05,
      "loss": 1.1826,
      "step": 1187
    },
    {
      "epoch": 0.35384128520635155,
      "grad_norm": 0.7435895204544067,
      "learning_rate": 1.9687685162876406e-05,
      "loss": 1.1974,
      "step": 1188
    },
    {
      "epoch": 0.3541391314060202,
      "grad_norm": 0.7397310137748718,
      "learning_rate": 1.968437146362346e-05,
      "loss": 1.1899,
      "step": 1189
    },
    {
      "epoch": 0.3544369776056889,
      "grad_norm": 0.7301692366600037,
      "learning_rate": 1.968104055947833e-05,
      "loss": 1.1969,
      "step": 1190
    },
    {
      "epoch": 0.35473482380535754,
      "grad_norm": 0.77022784948349,
      "learning_rate": 1.967769245635858e-05,
      "loss": 1.1902,
      "step": 1191
    },
    {
      "epoch": 0.35503267000502614,
      "grad_norm": 0.7953547239303589,
      "learning_rate": 1.967432716021232e-05,
      "loss": 1.1952,
      "step": 1192
    },
    {
      "epoch": 0.3553305162046948,
      "grad_norm": 0.7217286229133606,
      "learning_rate": 1.9670944677018214e-05,
      "loss": 1.1697,
      "step": 1193
    },
    {
      "epoch": 0.35562836240436346,
      "grad_norm": 0.7339707612991333,
      "learning_rate": 1.9667545012785448e-05,
      "loss": 1.1792,
      "step": 1194
    },
    {
      "epoch": 0.35592620860403207,
      "grad_norm": 0.7872765064239502,
      "learning_rate": 1.9664128173553748e-05,
      "loss": 1.2019,
      "step": 1195
    },
    {
      "epoch": 0.35622405480370073,
      "grad_norm": 0.7579273581504822,
      "learning_rate": 1.9660694165393334e-05,
      "loss": 1.1908,
      "step": 1196
    },
    {
      "epoch": 0.3565219010033694,
      "grad_norm": 0.7430486679077148,
      "learning_rate": 1.9657242994404947e-05,
      "loss": 1.1607,
      "step": 1197
    },
    {
      "epoch": 0.35681974720303805,
      "grad_norm": 0.7504818439483643,
      "learning_rate": 1.9653774666719796e-05,
      "loss": 1.1696,
      "step": 1198
    },
    {
      "epoch": 0.35711759340270666,
      "grad_norm": 0.7376049757003784,
      "learning_rate": 1.9650289188499587e-05,
      "loss": 1.1785,
      "step": 1199
    },
    {
      "epoch": 0.3574154396023753,
      "grad_norm": 0.7408781051635742,
      "learning_rate": 1.9646786565936497e-05,
      "loss": 1.1826,
      "step": 1200
    },
    {
      "epoch": 0.357713285802044,
      "grad_norm": 0.7412210702896118,
      "learning_rate": 1.964326680525314e-05,
      "loss": 1.1934,
      "step": 1201
    },
    {
      "epoch": 0.35801113200171264,
      "grad_norm": 0.8186607956886292,
      "learning_rate": 1.963972991270261e-05,
      "loss": 1.1866,
      "step": 1202
    },
    {
      "epoch": 0.35830897820138125,
      "grad_norm": 0.7425365447998047,
      "learning_rate": 1.9636175894568397e-05,
      "loss": 1.1886,
      "step": 1203
    },
    {
      "epoch": 0.3586068244010499,
      "grad_norm": 0.8242449164390564,
      "learning_rate": 1.9632604757164456e-05,
      "loss": 1.2043,
      "step": 1204
    },
    {
      "epoch": 0.35890467060071857,
      "grad_norm": 0.7621359825134277,
      "learning_rate": 1.9629016506835122e-05,
      "loss": 1.1803,
      "step": 1205
    },
    {
      "epoch": 0.3592025168003872,
      "grad_norm": 0.7280339002609253,
      "learning_rate": 1.9625411149955156e-05,
      "loss": 1.1799,
      "step": 1206
    },
    {
      "epoch": 0.35950036300005583,
      "grad_norm": 0.7696136832237244,
      "learning_rate": 1.9621788692929695e-05,
      "loss": 1.1989,
      "step": 1207
    },
    {
      "epoch": 0.3597982091997245,
      "grad_norm": 0.7818377614021301,
      "learning_rate": 1.9618149142194262e-05,
      "loss": 1.204,
      "step": 1208
    },
    {
      "epoch": 0.36009605539939316,
      "grad_norm": 0.7287859916687012,
      "learning_rate": 1.9614492504214744e-05,
      "loss": 1.1701,
      "step": 1209
    },
    {
      "epoch": 0.36039390159906176,
      "grad_norm": 0.7436671853065491,
      "learning_rate": 1.9610818785487392e-05,
      "loss": 1.1818,
      "step": 1210
    },
    {
      "epoch": 0.3606917477987304,
      "grad_norm": 0.7257910966873169,
      "learning_rate": 1.9607127992538796e-05,
      "loss": 1.1847,
      "step": 1211
    },
    {
      "epoch": 0.3609895939983991,
      "grad_norm": 0.7491075992584229,
      "learning_rate": 1.960342013192588e-05,
      "loss": 1.1877,
      "step": 1212
    },
    {
      "epoch": 0.36128744019806774,
      "grad_norm": 0.7270704507827759,
      "learning_rate": 1.9599695210235886e-05,
      "loss": 1.1858,
      "step": 1213
    },
    {
      "epoch": 0.36158528639773635,
      "grad_norm": 0.7612900733947754,
      "learning_rate": 1.959595323408638e-05,
      "loss": 1.1863,
      "step": 1214
    },
    {
      "epoch": 0.361883132597405,
      "grad_norm": 0.7288168668746948,
      "learning_rate": 1.959219421012521e-05,
      "loss": 1.174,
      "step": 1215
    },
    {
      "epoch": 0.36218097879707367,
      "grad_norm": 0.753418505191803,
      "learning_rate": 1.958841814503052e-05,
      "loss": 1.182,
      "step": 1216
    },
    {
      "epoch": 0.36247882499674233,
      "grad_norm": 0.7423313856124878,
      "learning_rate": 1.9584625045510725e-05,
      "loss": 1.1952,
      "step": 1217
    },
    {
      "epoch": 0.36277667119641094,
      "grad_norm": 0.7235876321792603,
      "learning_rate": 1.9580814918304504e-05,
      "loss": 1.1965,
      "step": 1218
    },
    {
      "epoch": 0.3630745173960796,
      "grad_norm": 0.7418579459190369,
      "learning_rate": 1.9576987770180788e-05,
      "loss": 1.1826,
      "step": 1219
    },
    {
      "epoch": 0.36337236359574826,
      "grad_norm": 0.7338332533836365,
      "learning_rate": 1.957314360793874e-05,
      "loss": 1.1711,
      "step": 1220
    },
    {
      "epoch": 0.36367020979541687,
      "grad_norm": 0.7262481451034546,
      "learning_rate": 1.9569282438407763e-05,
      "loss": 1.18,
      "step": 1221
    },
    {
      "epoch": 0.3639680559950855,
      "grad_norm": 0.7585115432739258,
      "learning_rate": 1.956540426844746e-05,
      "loss": 1.1804,
      "step": 1222
    },
    {
      "epoch": 0.3642659021947542,
      "grad_norm": 0.7682564854621887,
      "learning_rate": 1.9561509104947643e-05,
      "loss": 1.1891,
      "step": 1223
    },
    {
      "epoch": 0.36456374839442285,
      "grad_norm": 0.7752809524536133,
      "learning_rate": 1.9557596954828315e-05,
      "loss": 1.2049,
      "step": 1224
    },
    {
      "epoch": 0.36486159459409145,
      "grad_norm": 0.7521039843559265,
      "learning_rate": 1.955366782503966e-05,
      "loss": 1.1908,
      "step": 1225
    },
    {
      "epoch": 0.3651594407937601,
      "grad_norm": 0.746833086013794,
      "learning_rate": 1.954972172256202e-05,
      "loss": 1.1887,
      "step": 1226
    },
    {
      "epoch": 0.3654572869934288,
      "grad_norm": 0.7554108500480652,
      "learning_rate": 1.9545758654405888e-05,
      "loss": 1.1844,
      "step": 1227
    },
    {
      "epoch": 0.36575513319309744,
      "grad_norm": 0.751042902469635,
      "learning_rate": 1.9541778627611908e-05,
      "loss": 1.1899,
      "step": 1228
    },
    {
      "epoch": 0.36605297939276604,
      "grad_norm": 0.7797791957855225,
      "learning_rate": 1.9537781649250848e-05,
      "loss": 1.1901,
      "step": 1229
    },
    {
      "epoch": 0.3663508255924347,
      "grad_norm": 0.762665867805481,
      "learning_rate": 1.9533767726423586e-05,
      "loss": 1.1694,
      "step": 1230
    },
    {
      "epoch": 0.36664867179210336,
      "grad_norm": 0.8164774775505066,
      "learning_rate": 1.9529736866261112e-05,
      "loss": 1.1858,
      "step": 1231
    },
    {
      "epoch": 0.366946517991772,
      "grad_norm": 0.785852313041687,
      "learning_rate": 1.9525689075924498e-05,
      "loss": 1.1833,
      "step": 1232
    },
    {
      "epoch": 0.36724436419144063,
      "grad_norm": 0.7918466329574585,
      "learning_rate": 1.9521624362604896e-05,
      "loss": 1.1882,
      "step": 1233
    },
    {
      "epoch": 0.3675422103911093,
      "grad_norm": 0.7896957397460938,
      "learning_rate": 1.9517542733523528e-05,
      "loss": 1.1604,
      "step": 1234
    },
    {
      "epoch": 0.36784005659077795,
      "grad_norm": 0.7514079809188843,
      "learning_rate": 1.951344419593166e-05,
      "loss": 1.1984,
      "step": 1235
    },
    {
      "epoch": 0.36813790279044656,
      "grad_norm": 0.7323447465896606,
      "learning_rate": 1.9509328757110598e-05,
      "loss": 1.1906,
      "step": 1236
    },
    {
      "epoch": 0.3684357489901152,
      "grad_norm": 0.7563832998275757,
      "learning_rate": 1.9505196424371685e-05,
      "loss": 1.1751,
      "step": 1237
    },
    {
      "epoch": 0.3687335951897839,
      "grad_norm": 0.7682973742485046,
      "learning_rate": 1.9501047205056262e-05,
      "loss": 1.1821,
      "step": 1238
    },
    {
      "epoch": 0.36903144138945254,
      "grad_norm": 0.7498724460601807,
      "learning_rate": 1.9496881106535675e-05,
      "loss": 1.1863,
      "step": 1239
    },
    {
      "epoch": 0.36932928758912115,
      "grad_norm": 0.7884436249732971,
      "learning_rate": 1.949269813621127e-05,
      "loss": 1.1916,
      "step": 1240
    },
    {
      "epoch": 0.3696271337887898,
      "grad_norm": 0.7508389949798584,
      "learning_rate": 1.9488498301514343e-05,
      "loss": 1.1903,
      "step": 1241
    },
    {
      "epoch": 0.36992497998845847,
      "grad_norm": 0.7556812167167664,
      "learning_rate": 1.9484281609906172e-05,
      "loss": 1.163,
      "step": 1242
    },
    {
      "epoch": 0.37022282618812713,
      "grad_norm": 0.780400276184082,
      "learning_rate": 1.948004806887797e-05,
      "loss": 1.1825,
      "step": 1243
    },
    {
      "epoch": 0.37052067238779574,
      "grad_norm": 0.7456673383712769,
      "learning_rate": 1.9475797685950885e-05,
      "loss": 1.2002,
      "step": 1244
    },
    {
      "epoch": 0.3708185185874644,
      "grad_norm": 0.7809370756149292,
      "learning_rate": 1.9471530468675995e-05,
      "loss": 1.1958,
      "step": 1245
    },
    {
      "epoch": 0.37111636478713306,
      "grad_norm": 0.7585657835006714,
      "learning_rate": 1.946724642463427e-05,
      "loss": 1.1854,
      "step": 1246
    },
    {
      "epoch": 0.37141421098680166,
      "grad_norm": 0.7737017273902893,
      "learning_rate": 1.946294556143659e-05,
      "loss": 1.1733,
      "step": 1247
    },
    {
      "epoch": 0.3717120571864703,
      "grad_norm": 0.7103421688079834,
      "learning_rate": 1.9458627886723706e-05,
      "loss": 1.1663,
      "step": 1248
    },
    {
      "epoch": 0.372009903386139,
      "grad_norm": 0.7692255973815918,
      "learning_rate": 1.945429340816624e-05,
      "loss": 1.1781,
      "step": 1249
    },
    {
      "epoch": 0.37230774958580765,
      "grad_norm": 0.7707586288452148,
      "learning_rate": 1.9449942133464654e-05,
      "loss": 1.1988,
      "step": 1250
    },
    {
      "epoch": 0.37260559578547625,
      "grad_norm": 0.7744660377502441,
      "learning_rate": 1.9445574070349272e-05,
      "loss": 1.207,
      "step": 1251
    },
    {
      "epoch": 0.3729034419851449,
      "grad_norm": 0.7319050431251526,
      "learning_rate": 1.9441189226580225e-05,
      "loss": 1.1836,
      "step": 1252
    },
    {
      "epoch": 0.3732012881848136,
      "grad_norm": 0.7696720361709595,
      "learning_rate": 1.9436787609947464e-05,
      "loss": 1.1934,
      "step": 1253
    },
    {
      "epoch": 0.37349913438448223,
      "grad_norm": 0.7679592370986938,
      "learning_rate": 1.9432369228270733e-05,
      "loss": 1.1753,
      "step": 1254
    },
    {
      "epoch": 0.37379698058415084,
      "grad_norm": 0.7620283961296082,
      "learning_rate": 1.942793408939957e-05,
      "loss": 1.1857,
      "step": 1255
    },
    {
      "epoch": 0.3740948267838195,
      "grad_norm": 0.7681359052658081,
      "learning_rate": 1.9423482201213275e-05,
      "loss": 1.1762,
      "step": 1256
    },
    {
      "epoch": 0.37439267298348816,
      "grad_norm": 0.7861395478248596,
      "learning_rate": 1.9419013571620907e-05,
      "loss": 1.1809,
      "step": 1257
    },
    {
      "epoch": 0.3746905191831568,
      "grad_norm": 0.8044225573539734,
      "learning_rate": 1.9414528208561262e-05,
      "loss": 1.1544,
      "step": 1258
    },
    {
      "epoch": 0.37498836538282543,
      "grad_norm": 0.8162577152252197,
      "learning_rate": 1.941002612000287e-05,
      "loss": 1.1721,
      "step": 1259
    },
    {
      "epoch": 0.3752862115824941,
      "grad_norm": 0.752648115158081,
      "learning_rate": 1.9405507313943975e-05,
      "loss": 1.1849,
      "step": 1260
    },
    {
      "epoch": 0.37558405778216275,
      "grad_norm": 0.7869867086410522,
      "learning_rate": 1.9400971798412514e-05,
      "loss": 1.1763,
      "step": 1261
    },
    {
      "epoch": 0.37588190398183136,
      "grad_norm": 0.8066076636314392,
      "learning_rate": 1.939641958146612e-05,
      "loss": 1.1999,
      "step": 1262
    },
    {
      "epoch": 0.3761797501815,
      "grad_norm": 0.7576519846916199,
      "learning_rate": 1.9391850671192092e-05,
      "loss": 1.1902,
      "step": 1263
    },
    {
      "epoch": 0.3764775963811687,
      "grad_norm": 0.7722829580307007,
      "learning_rate": 1.938726507570738e-05,
      "loss": 1.1837,
      "step": 1264
    },
    {
      "epoch": 0.37677544258083734,
      "grad_norm": 0.7318130731582642,
      "learning_rate": 1.9382662803158585e-05,
      "loss": 1.1779,
      "step": 1265
    },
    {
      "epoch": 0.37707328878050594,
      "grad_norm": 0.7827332615852356,
      "learning_rate": 1.937804386172193e-05,
      "loss": 1.1854,
      "step": 1266
    },
    {
      "epoch": 0.3773711349801746,
      "grad_norm": 0.7257670164108276,
      "learning_rate": 1.9373408259603254e-05,
      "loss": 1.1762,
      "step": 1267
    },
    {
      "epoch": 0.37766898117984327,
      "grad_norm": 0.7287430763244629,
      "learning_rate": 1.9368756005038e-05,
      "loss": 1.1858,
      "step": 1268
    },
    {
      "epoch": 0.3779668273795119,
      "grad_norm": 0.7346288561820984,
      "learning_rate": 1.936408710629118e-05,
      "loss": 1.2022,
      "step": 1269
    },
    {
      "epoch": 0.37826467357918053,
      "grad_norm": 0.7436503171920776,
      "learning_rate": 1.935940157165739e-05,
      "loss": 1.1863,
      "step": 1270
    },
    {
      "epoch": 0.3785625197788492,
      "grad_norm": 0.7002028226852417,
      "learning_rate": 1.9354699409460775e-05,
      "loss": 1.1744,
      "step": 1271
    },
    {
      "epoch": 0.37886036597851785,
      "grad_norm": 0.8091986179351807,
      "learning_rate": 1.9349980628055023e-05,
      "loss": 1.1928,
      "step": 1272
    },
    {
      "epoch": 0.3791582121781865,
      "grad_norm": 0.7486268877983093,
      "learning_rate": 1.9345245235823343e-05,
      "loss": 1.1889,
      "step": 1273
    },
    {
      "epoch": 0.3794560583778551,
      "grad_norm": 0.7134963870048523,
      "learning_rate": 1.9340493241178452e-05,
      "loss": 1.167,
      "step": 1274
    },
    {
      "epoch": 0.3797539045775238,
      "grad_norm": 0.749902069568634,
      "learning_rate": 1.933572465256257e-05,
      "loss": 1.171,
      "step": 1275
    },
    {
      "epoch": 0.38005175077719244,
      "grad_norm": 0.7200993895530701,
      "learning_rate": 1.9330939478447392e-05,
      "loss": 1.1794,
      "step": 1276
    },
    {
      "epoch": 0.38034959697686105,
      "grad_norm": 0.7357526421546936,
      "learning_rate": 1.932613772733408e-05,
      "loss": 1.1748,
      "step": 1277
    },
    {
      "epoch": 0.3806474431765297,
      "grad_norm": 0.7499547004699707,
      "learning_rate": 1.9321319407753244e-05,
      "loss": 1.176,
      "step": 1278
    },
    {
      "epoch": 0.38094528937619837,
      "grad_norm": 0.756248414516449,
      "learning_rate": 1.9316484528264932e-05,
      "loss": 1.1883,
      "step": 1279
    },
    {
      "epoch": 0.38124313557586703,
      "grad_norm": 0.7588652968406677,
      "learning_rate": 1.9311633097458608e-05,
      "loss": 1.1764,
      "step": 1280
    },
    {
      "epoch": 0.38154098177553564,
      "grad_norm": 0.7738041877746582,
      "learning_rate": 1.930676512395315e-05,
      "loss": 1.162,
      "step": 1281
    },
    {
      "epoch": 0.3818388279752043,
      "grad_norm": 0.7325159311294556,
      "learning_rate": 1.9301880616396803e-05,
      "loss": 1.1795,
      "step": 1282
    },
    {
      "epoch": 0.38213667417487296,
      "grad_norm": 0.7577996253967285,
      "learning_rate": 1.929697958346722e-05,
      "loss": 1.1746,
      "step": 1283
    },
    {
      "epoch": 0.3824345203745416,
      "grad_norm": 0.7975197434425354,
      "learning_rate": 1.9292062033871374e-05,
      "loss": 1.1821,
      "step": 1284
    },
    {
      "epoch": 0.3827323665742102,
      "grad_norm": 0.7376799583435059,
      "learning_rate": 1.928712797634561e-05,
      "loss": 1.1833,
      "step": 1285
    },
    {
      "epoch": 0.3830302127738789,
      "grad_norm": 0.7908791303634644,
      "learning_rate": 1.9282177419655586e-05,
      "loss": 1.194,
      "step": 1286
    },
    {
      "epoch": 0.38332805897354755,
      "grad_norm": 0.6974128484725952,
      "learning_rate": 1.9277210372596278e-05,
      "loss": 1.1738,
      "step": 1287
    },
    {
      "epoch": 0.38362590517321615,
      "grad_norm": 0.7993919253349304,
      "learning_rate": 1.9272226843991956e-05,
      "loss": 1.1752,
      "step": 1288
    },
    {
      "epoch": 0.3839237513728848,
      "grad_norm": 0.7306740880012512,
      "learning_rate": 1.9267226842696174e-05,
      "loss": 1.1867,
      "step": 1289
    },
    {
      "epoch": 0.3842215975725535,
      "grad_norm": 0.8096683025360107,
      "learning_rate": 1.926221037759174e-05,
      "loss": 1.1858,
      "step": 1290
    },
    {
      "epoch": 0.38451944377222214,
      "grad_norm": 0.7774428725242615,
      "learning_rate": 1.925717745759072e-05,
      "loss": 1.1853,
      "step": 1291
    },
    {
      "epoch": 0.38481728997189074,
      "grad_norm": 0.7809367775917053,
      "learning_rate": 1.9252128091634418e-05,
      "loss": 1.1958,
      "step": 1292
    },
    {
      "epoch": 0.3851151361715594,
      "grad_norm": 0.7895611524581909,
      "learning_rate": 1.9247062288693342e-05,
      "loss": 1.1673,
      "step": 1293
    },
    {
      "epoch": 0.38541298237122806,
      "grad_norm": 0.8459852337837219,
      "learning_rate": 1.924198005776721e-05,
      "loss": 1.1711,
      "step": 1294
    },
    {
      "epoch": 0.3857108285708967,
      "grad_norm": 0.7680491209030151,
      "learning_rate": 1.9236881407884928e-05,
      "loss": 1.1947,
      "step": 1295
    },
    {
      "epoch": 0.38600867477056533,
      "grad_norm": 0.8136183023452759,
      "learning_rate": 1.9231766348104556e-05,
      "loss": 1.1838,
      "step": 1296
    },
    {
      "epoch": 0.386306520970234,
      "grad_norm": 0.750747799873352,
      "learning_rate": 1.922663488751333e-05,
      "loss": 1.1709,
      "step": 1297
    },
    {
      "epoch": 0.38660436716990265,
      "grad_norm": 0.777517557144165,
      "learning_rate": 1.9221487035227602e-05,
      "loss": 1.1894,
      "step": 1298
    },
    {
      "epoch": 0.3869022133695713,
      "grad_norm": 0.7362677454948425,
      "learning_rate": 1.9216322800392855e-05,
      "loss": 1.1737,
      "step": 1299
    },
    {
      "epoch": 0.3872000595692399,
      "grad_norm": 0.7473068237304688,
      "learning_rate": 1.9211142192183683e-05,
      "loss": 1.16,
      "step": 1300
    },
    {
      "epoch": 0.3874979057689086,
      "grad_norm": 0.7427869439125061,
      "learning_rate": 1.920594521980375e-05,
      "loss": 1.1938,
      "step": 1301
    },
    {
      "epoch": 0.38779575196857724,
      "grad_norm": 0.7564520239830017,
      "learning_rate": 1.920073189248581e-05,
      "loss": 1.1845,
      "step": 1302
    },
    {
      "epoch": 0.38809359816824585,
      "grad_norm": 0.7463542819023132,
      "learning_rate": 1.9195502219491663e-05,
      "loss": 1.1829,
      "step": 1303
    },
    {
      "epoch": 0.3883914443679145,
      "grad_norm": 0.7871155738830566,
      "learning_rate": 1.9190256210112155e-05,
      "loss": 1.1805,
      "step": 1304
    },
    {
      "epoch": 0.38868929056758317,
      "grad_norm": 0.7801784873008728,
      "learning_rate": 1.918499387366714e-05,
      "loss": 1.1845,
      "step": 1305
    },
    {
      "epoch": 0.38898713676725183,
      "grad_norm": 0.7150505185127258,
      "learning_rate": 1.9179715219505498e-05,
      "loss": 1.1935,
      "step": 1306
    },
    {
      "epoch": 0.38928498296692043,
      "grad_norm": 0.7189215421676636,
      "learning_rate": 1.9174420257005085e-05,
      "loss": 1.1833,
      "step": 1307
    },
    {
      "epoch": 0.3895828291665891,
      "grad_norm": 0.733676552772522,
      "learning_rate": 1.9169108995572735e-05,
      "loss": 1.1705,
      "step": 1308
    },
    {
      "epoch": 0.38988067536625776,
      "grad_norm": 0.7118922472000122,
      "learning_rate": 1.916378144464423e-05,
      "loss": 1.1805,
      "step": 1309
    },
    {
      "epoch": 0.3901785215659264,
      "grad_norm": 0.7454823851585388,
      "learning_rate": 1.9158437613684305e-05,
      "loss": 1.1741,
      "step": 1310
    },
    {
      "epoch": 0.390476367765595,
      "grad_norm": 0.7496629357337952,
      "learning_rate": 1.9153077512186605e-05,
      "loss": 1.1747,
      "step": 1311
    },
    {
      "epoch": 0.3907742139652637,
      "grad_norm": 0.7565000057220459,
      "learning_rate": 1.914770114967369e-05,
      "loss": 1.1737,
      "step": 1312
    },
    {
      "epoch": 0.39107206016493234,
      "grad_norm": 0.7621682286262512,
      "learning_rate": 1.9142308535697005e-05,
      "loss": 1.1826,
      "step": 1313
    },
    {
      "epoch": 0.391369906364601,
      "grad_norm": 0.7408864498138428,
      "learning_rate": 1.9136899679836863e-05,
      "loss": 1.1973,
      "step": 1314
    },
    {
      "epoch": 0.3916677525642696,
      "grad_norm": 0.7508994340896606,
      "learning_rate": 1.9131474591702438e-05,
      "loss": 1.1726,
      "step": 1315
    },
    {
      "epoch": 0.39196559876393827,
      "grad_norm": 0.7480928301811218,
      "learning_rate": 1.9126033280931734e-05,
      "loss": 1.17,
      "step": 1316
    },
    {
      "epoch": 0.39226344496360693,
      "grad_norm": 0.7746785283088684,
      "learning_rate": 1.9120575757191584e-05,
      "loss": 1.1924,
      "step": 1317
    },
    {
      "epoch": 0.39256129116327554,
      "grad_norm": 0.6911002993583679,
      "learning_rate": 1.911510203017762e-05,
      "loss": 1.1703,
      "step": 1318
    },
    {
      "epoch": 0.3928591373629442,
      "grad_norm": 0.7377850413322449,
      "learning_rate": 1.9109612109614263e-05,
      "loss": 1.1752,
      "step": 1319
    },
    {
      "epoch": 0.39315698356261286,
      "grad_norm": 0.7119675874710083,
      "learning_rate": 1.9104106005254696e-05,
      "loss": 1.1689,
      "step": 1320
    },
    {
      "epoch": 0.3934548297622815,
      "grad_norm": 0.7650914788246155,
      "learning_rate": 1.909858372688086e-05,
      "loss": 1.1813,
      "step": 1321
    },
    {
      "epoch": 0.3937526759619501,
      "grad_norm": 0.7208782434463501,
      "learning_rate": 1.909304528430343e-05,
      "loss": 1.1733,
      "step": 1322
    },
    {
      "epoch": 0.3940505221616188,
      "grad_norm": 0.7546354532241821,
      "learning_rate": 1.9087490687361794e-05,
      "loss": 1.1744,
      "step": 1323
    },
    {
      "epoch": 0.39434836836128745,
      "grad_norm": 0.7382991909980774,
      "learning_rate": 1.908191994592404e-05,
      "loss": 1.1815,
      "step": 1324
    },
    {
      "epoch": 0.3946462145609561,
      "grad_norm": 0.7719926238059998,
      "learning_rate": 1.9076333069886943e-05,
      "loss": 1.1791,
      "step": 1325
    },
    {
      "epoch": 0.3949440607606247,
      "grad_norm": 0.7703098058700562,
      "learning_rate": 1.9070730069175936e-05,
      "loss": 1.1752,
      "step": 1326
    },
    {
      "epoch": 0.3952419069602934,
      "grad_norm": 0.7358701825141907,
      "learning_rate": 1.9065110953745098e-05,
      "loss": 1.1768,
      "step": 1327
    },
    {
      "epoch": 0.39553975315996204,
      "grad_norm": 0.7509908080101013,
      "learning_rate": 1.9059475733577147e-05,
      "loss": 1.1722,
      "step": 1328
    },
    {
      "epoch": 0.39583759935963064,
      "grad_norm": 0.7682883143424988,
      "learning_rate": 1.9053824418683395e-05,
      "loss": 1.1883,
      "step": 1329
    },
    {
      "epoch": 0.3961354455592993,
      "grad_norm": 0.7478414177894592,
      "learning_rate": 1.9048157019103758e-05,
      "loss": 1.1916,
      "step": 1330
    },
    {
      "epoch": 0.39643329175896797,
      "grad_norm": 0.7387676239013672,
      "learning_rate": 1.9042473544906733e-05,
      "loss": 1.1704,
      "step": 1331
    },
    {
      "epoch": 0.3967311379586366,
      "grad_norm": 0.7943106293678284,
      "learning_rate": 1.903677400618936e-05,
      "loss": 1.1789,
      "step": 1332
    },
    {
      "epoch": 0.39702898415830523,
      "grad_norm": 0.7633237242698669,
      "learning_rate": 1.9031058413077233e-05,
      "loss": 1.1721,
      "step": 1333
    },
    {
      "epoch": 0.3973268303579739,
      "grad_norm": 0.7671225666999817,
      "learning_rate": 1.902532677572446e-05,
      "loss": 1.1782,
      "step": 1334
    },
    {
      "epoch": 0.39762467655764255,
      "grad_norm": 0.764367401599884,
      "learning_rate": 1.9019579104313645e-05,
      "loss": 1.1692,
      "step": 1335
    },
    {
      "epoch": 0.3979225227573112,
      "grad_norm": 0.7977184057235718,
      "learning_rate": 1.9013815409055895e-05,
      "loss": 1.16,
      "step": 1336
    },
    {
      "epoch": 0.3982203689569798,
      "grad_norm": 0.7282764911651611,
      "learning_rate": 1.9008035700190774e-05,
      "loss": 1.1721,
      "step": 1337
    },
    {
      "epoch": 0.3985182151566485,
      "grad_norm": 0.7930206656455994,
      "learning_rate": 1.9002239987986296e-05,
      "loss": 1.1671,
      "step": 1338
    },
    {
      "epoch": 0.39881606135631714,
      "grad_norm": 0.7978341579437256,
      "learning_rate": 1.8996428282738906e-05,
      "loss": 1.1778,
      "step": 1339
    },
    {
      "epoch": 0.3991139075559858,
      "grad_norm": 0.7553421854972839,
      "learning_rate": 1.899060059477346e-05,
      "loss": 1.1568,
      "step": 1340
    },
    {
      "epoch": 0.3994117537556544,
      "grad_norm": 0.7776437401771545,
      "learning_rate": 1.8984756934443215e-05,
      "loss": 1.1649,
      "step": 1341
    },
    {
      "epoch": 0.39970959995532307,
      "grad_norm": 0.8138219714164734,
      "learning_rate": 1.89788973121298e-05,
      "loss": 1.181,
      "step": 1342
    },
    {
      "epoch": 0.40000744615499173,
      "grad_norm": 0.8876965045928955,
      "learning_rate": 1.8973021738243193e-05,
      "loss": 1.1844,
      "step": 1343
    },
    {
      "epoch": 0.40030529235466034,
      "grad_norm": 0.7946558594703674,
      "learning_rate": 1.896713022322173e-05,
      "loss": 1.184,
      "step": 1344
    },
    {
      "epoch": 0.400603138554329,
      "grad_norm": 0.8262109756469727,
      "learning_rate": 1.8961222777532048e-05,
      "loss": 1.196,
      "step": 1345
    },
    {
      "epoch": 0.40090098475399766,
      "grad_norm": 0.8196043968200684,
      "learning_rate": 1.8955299411669093e-05,
      "loss": 1.1704,
      "step": 1346
    },
    {
      "epoch": 0.4011988309536663,
      "grad_norm": 0.8154504299163818,
      "learning_rate": 1.89493601361561e-05,
      "loss": 1.1753,
      "step": 1347
    },
    {
      "epoch": 0.4014966771533349,
      "grad_norm": 0.738422155380249,
      "learning_rate": 1.8943404961544565e-05,
      "loss": 1.1867,
      "step": 1348
    },
    {
      "epoch": 0.4017945233530036,
      "grad_norm": 0.8594563603401184,
      "learning_rate": 1.8937433898414223e-05,
      "loss": 1.1456,
      "step": 1349
    },
    {
      "epoch": 0.40209236955267225,
      "grad_norm": 0.7644219994544983,
      "learning_rate": 1.8931446957373045e-05,
      "loss": 1.173,
      "step": 1350
    },
    {
      "epoch": 0.4023902157523409,
      "grad_norm": 0.838342010974884,
      "learning_rate": 1.8925444149057206e-05,
      "loss": 1.1775,
      "step": 1351
    },
    {
      "epoch": 0.4026880619520095,
      "grad_norm": 0.7222621440887451,
      "learning_rate": 1.8919425484131072e-05,
      "loss": 1.1749,
      "step": 1352
    },
    {
      "epoch": 0.4029859081516782,
      "grad_norm": 0.8797105550765991,
      "learning_rate": 1.8913390973287176e-05,
      "loss": 1.1763,
      "step": 1353
    },
    {
      "epoch": 0.40328375435134683,
      "grad_norm": 0.727583110332489,
      "learning_rate": 1.8907340627246204e-05,
      "loss": 1.1676,
      "step": 1354
    },
    {
      "epoch": 0.4035816005510155,
      "grad_norm": 0.7924251556396484,
      "learning_rate": 1.890127445675698e-05,
      "loss": 1.1772,
      "step": 1355
    },
    {
      "epoch": 0.4038794467506841,
      "grad_norm": 0.7086780667304993,
      "learning_rate": 1.8895192472596425e-05,
      "loss": 1.155,
      "step": 1356
    },
    {
      "epoch": 0.40417729295035276,
      "grad_norm": 0.77681964635849,
      "learning_rate": 1.8889094685569577e-05,
      "loss": 1.1566,
      "step": 1357
    },
    {
      "epoch": 0.4044751391500214,
      "grad_norm": 0.7347881197929382,
      "learning_rate": 1.8882981106509528e-05,
      "loss": 1.1652,
      "step": 1358
    },
    {
      "epoch": 0.40477298534969003,
      "grad_norm": 0.8370055556297302,
      "learning_rate": 1.8876851746277434e-05,
      "loss": 1.1792,
      "step": 1359
    },
    {
      "epoch": 0.4050708315493587,
      "grad_norm": 0.731926441192627,
      "learning_rate": 1.8870706615762492e-05,
      "loss": 1.1832,
      "step": 1360
    },
    {
      "epoch": 0.40536867774902735,
      "grad_norm": 0.7681272029876709,
      "learning_rate": 1.8864545725881908e-05,
      "loss": 1.19,
      "step": 1361
    },
    {
      "epoch": 0.405666523948696,
      "grad_norm": 0.7270525693893433,
      "learning_rate": 1.8858369087580887e-05,
      "loss": 1.1846,
      "step": 1362
    },
    {
      "epoch": 0.4059643701483646,
      "grad_norm": 0.7651084661483765,
      "learning_rate": 1.8852176711832614e-05,
      "loss": 1.1605,
      "step": 1363
    },
    {
      "epoch": 0.4062622163480333,
      "grad_norm": 0.7167275547981262,
      "learning_rate": 1.884596860963824e-05,
      "loss": 1.1748,
      "step": 1364
    },
    {
      "epoch": 0.40656006254770194,
      "grad_norm": 0.7927423715591431,
      "learning_rate": 1.8839744792026837e-05,
      "loss": 1.1955,
      "step": 1365
    },
    {
      "epoch": 0.4068579087473706,
      "grad_norm": 0.7079978585243225,
      "learning_rate": 1.883350527005541e-05,
      "loss": 1.1826,
      "step": 1366
    },
    {
      "epoch": 0.4071557549470392,
      "grad_norm": 0.7415812611579895,
      "learning_rate": 1.8827250054808864e-05,
      "loss": 1.1777,
      "step": 1367
    },
    {
      "epoch": 0.40745360114670787,
      "grad_norm": 0.7239328622817993,
      "learning_rate": 1.8820979157399976e-05,
      "loss": 1.1736,
      "step": 1368
    },
    {
      "epoch": 0.4077514473463765,
      "grad_norm": 0.749142587184906,
      "learning_rate": 1.8814692588969387e-05,
      "loss": 1.1853,
      "step": 1369
    },
    {
      "epoch": 0.4080492935460452,
      "grad_norm": 0.73469477891922,
      "learning_rate": 1.8808390360685586e-05,
      "loss": 1.1639,
      "step": 1370
    },
    {
      "epoch": 0.4083471397457138,
      "grad_norm": 0.7581102848052979,
      "learning_rate": 1.8802072483744867e-05,
      "loss": 1.1798,
      "step": 1371
    },
    {
      "epoch": 0.40864498594538246,
      "grad_norm": 0.8163647651672363,
      "learning_rate": 1.8795738969371343e-05,
      "loss": 1.1855,
      "step": 1372
    },
    {
      "epoch": 0.4089428321450511,
      "grad_norm": 0.7544882893562317,
      "learning_rate": 1.8789389828816894e-05,
      "loss": 1.1727,
      "step": 1373
    },
    {
      "epoch": 0.4092406783447197,
      "grad_norm": 0.7269321084022522,
      "learning_rate": 1.8783025073361162e-05,
      "loss": 1.1931,
      "step": 1374
    },
    {
      "epoch": 0.4095385245443884,
      "grad_norm": 0.7968947887420654,
      "learning_rate": 1.877664471431154e-05,
      "loss": 1.1846,
      "step": 1375
    },
    {
      "epoch": 0.40983637074405704,
      "grad_norm": 0.7836090922355652,
      "learning_rate": 1.8770248763003135e-05,
      "loss": 1.1854,
      "step": 1376
    },
    {
      "epoch": 0.4101342169437257,
      "grad_norm": 0.7331004738807678,
      "learning_rate": 1.876383723079875e-05,
      "loss": 1.1574,
      "step": 1377
    },
    {
      "epoch": 0.4104320631433943,
      "grad_norm": 0.7248095870018005,
      "learning_rate": 1.875741012908888e-05,
      "loss": 1.1529,
      "step": 1378
    },
    {
      "epoch": 0.41072990934306297,
      "grad_norm": 0.7834470272064209,
      "learning_rate": 1.8750967469291666e-05,
      "loss": 1.1792,
      "step": 1379
    },
    {
      "epoch": 0.41102775554273163,
      "grad_norm": 0.7055449485778809,
      "learning_rate": 1.8744509262852902e-05,
      "loss": 1.1638,
      "step": 1380
    },
    {
      "epoch": 0.4113256017424003,
      "grad_norm": 0.7226158976554871,
      "learning_rate": 1.873803552124599e-05,
      "loss": 1.1783,
      "step": 1381
    },
    {
      "epoch": 0.4116234479420689,
      "grad_norm": 0.7454487681388855,
      "learning_rate": 1.8731546255971948e-05,
      "loss": 1.1767,
      "step": 1382
    },
    {
      "epoch": 0.41192129414173756,
      "grad_norm": 0.7255661487579346,
      "learning_rate": 1.8725041478559354e-05,
      "loss": 1.1712,
      "step": 1383
    },
    {
      "epoch": 0.4122191403414062,
      "grad_norm": 0.74687260389328,
      "learning_rate": 1.8718521200564352e-05,
      "loss": 1.1871,
      "step": 1384
    },
    {
      "epoch": 0.4125169865410748,
      "grad_norm": 0.7747193574905396,
      "learning_rate": 1.8711985433570628e-05,
      "loss": 1.1811,
      "step": 1385
    },
    {
      "epoch": 0.4128148327407435,
      "grad_norm": 0.7165244817733765,
      "learning_rate": 1.8705434189189374e-05,
      "loss": 1.1791,
      "step": 1386
    },
    {
      "epoch": 0.41311267894041215,
      "grad_norm": 0.7255867719650269,
      "learning_rate": 1.86988674790593e-05,
      "loss": 1.1785,
      "step": 1387
    },
    {
      "epoch": 0.4134105251400808,
      "grad_norm": 0.7561094164848328,
      "learning_rate": 1.869228531484656e-05,
      "loss": 1.1481,
      "step": 1388
    },
    {
      "epoch": 0.4137083713397494,
      "grad_norm": 0.7655609250068665,
      "learning_rate": 1.8685687708244794e-05,
      "loss": 1.1777,
      "step": 1389
    },
    {
      "epoch": 0.4140062175394181,
      "grad_norm": 0.7376701831817627,
      "learning_rate": 1.8679074670975056e-05,
      "loss": 1.183,
      "step": 1390
    },
    {
      "epoch": 0.41430406373908674,
      "grad_norm": 0.7706101536750793,
      "learning_rate": 1.8672446214785824e-05,
      "loss": 1.1734,
      "step": 1391
    },
    {
      "epoch": 0.4146019099387554,
      "grad_norm": 0.7527652978897095,
      "learning_rate": 1.8665802351452966e-05,
      "loss": 1.1562,
      "step": 1392
    },
    {
      "epoch": 0.414899756138424,
      "grad_norm": 0.744774580001831,
      "learning_rate": 1.865914309277972e-05,
      "loss": 1.1811,
      "step": 1393
    },
    {
      "epoch": 0.41519760233809266,
      "grad_norm": 0.7727804183959961,
      "learning_rate": 1.8652468450596673e-05,
      "loss": 1.1892,
      "step": 1394
    },
    {
      "epoch": 0.4154954485377613,
      "grad_norm": 0.7415773868560791,
      "learning_rate": 1.8645778436761748e-05,
      "loss": 1.186,
      "step": 1395
    },
    {
      "epoch": 0.41579329473743,
      "grad_norm": 0.7378939986228943,
      "learning_rate": 1.8639073063160172e-05,
      "loss": 1.1809,
      "step": 1396
    },
    {
      "epoch": 0.4160911409370986,
      "grad_norm": 0.7451414465904236,
      "learning_rate": 1.863235234170446e-05,
      "loss": 1.1652,
      "step": 1397
    },
    {
      "epoch": 0.41638898713676725,
      "grad_norm": 0.7288876175880432,
      "learning_rate": 1.8625616284334405e-05,
      "loss": 1.1931,
      "step": 1398
    },
    {
      "epoch": 0.4166868333364359,
      "grad_norm": 0.7764760255813599,
      "learning_rate": 1.8618864903017018e-05,
      "loss": 1.1919,
      "step": 1399
    },
    {
      "epoch": 0.4169846795361045,
      "grad_norm": 0.764079213142395,
      "learning_rate": 1.861209820974656e-05,
      "loss": 1.1729,
      "step": 1400
    },
    {
      "epoch": 0.4172825257357732,
      "grad_norm": 0.7303447723388672,
      "learning_rate": 1.8605316216544485e-05,
      "loss": 1.1863,
      "step": 1401
    },
    {
      "epoch": 0.41758037193544184,
      "grad_norm": 0.7388205528259277,
      "learning_rate": 1.8598518935459424e-05,
      "loss": 1.1887,
      "step": 1402
    },
    {
      "epoch": 0.4178782181351105,
      "grad_norm": 0.7869564890861511,
      "learning_rate": 1.859170637856718e-05,
      "loss": 1.1742,
      "step": 1403
    },
    {
      "epoch": 0.4181760643347791,
      "grad_norm": 0.7114114761352539,
      "learning_rate": 1.8584878557970677e-05,
      "loss": 1.155,
      "step": 1404
    },
    {
      "epoch": 0.41847391053444777,
      "grad_norm": 0.7448421716690063,
      "learning_rate": 1.857803548579997e-05,
      "loss": 1.1844,
      "step": 1405
    },
    {
      "epoch": 0.41877175673411643,
      "grad_norm": 0.7448183298110962,
      "learning_rate": 1.8571177174212214e-05,
      "loss": 1.1785,
      "step": 1406
    },
    {
      "epoch": 0.4190696029337851,
      "grad_norm": 0.7525942921638489,
      "learning_rate": 1.8564303635391617e-05,
      "loss": 1.1645,
      "step": 1407
    },
    {
      "epoch": 0.4193674491334537,
      "grad_norm": 0.7103204131126404,
      "learning_rate": 1.8557414881549453e-05,
      "loss": 1.1829,
      "step": 1408
    },
    {
      "epoch": 0.41966529533312236,
      "grad_norm": 0.7342169284820557,
      "learning_rate": 1.855051092492403e-05,
      "loss": 1.1786,
      "step": 1409
    },
    {
      "epoch": 0.419963141532791,
      "grad_norm": 0.7336074113845825,
      "learning_rate": 1.8543591777780653e-05,
      "loss": 1.1612,
      "step": 1410
    },
    {
      "epoch": 0.4202609877324597,
      "grad_norm": 0.7504079937934875,
      "learning_rate": 1.853665745241162e-05,
      "loss": 1.1614,
      "step": 1411
    },
    {
      "epoch": 0.4205588339321283,
      "grad_norm": 0.7435110807418823,
      "learning_rate": 1.8529707961136202e-05,
      "loss": 1.1613,
      "step": 1412
    },
    {
      "epoch": 0.42085668013179695,
      "grad_norm": 0.7113166451454163,
      "learning_rate": 1.8522743316300597e-05,
      "loss": 1.1738,
      "step": 1413
    },
    {
      "epoch": 0.4211545263314656,
      "grad_norm": 0.7583416700363159,
      "learning_rate": 1.8515763530277938e-05,
      "loss": 1.1842,
      "step": 1414
    },
    {
      "epoch": 0.4214523725311342,
      "grad_norm": 0.732340395450592,
      "learning_rate": 1.8508768615468248e-05,
      "loss": 1.177,
      "step": 1415
    },
    {
      "epoch": 0.4217502187308029,
      "grad_norm": 0.7636895775794983,
      "learning_rate": 1.8501758584298436e-05,
      "loss": 1.1723,
      "step": 1416
    },
    {
      "epoch": 0.42204806493047153,
      "grad_norm": 0.730125904083252,
      "learning_rate": 1.8494733449222254e-05,
      "loss": 1.1826,
      "step": 1417
    },
    {
      "epoch": 0.4223459111301402,
      "grad_norm": 0.7261401414871216,
      "learning_rate": 1.8487693222720297e-05,
      "loss": 1.1753,
      "step": 1418
    },
    {
      "epoch": 0.4226437573298088,
      "grad_norm": 0.7306705117225647,
      "learning_rate": 1.848063791729997e-05,
      "loss": 1.1772,
      "step": 1419
    },
    {
      "epoch": 0.42294160352947746,
      "grad_norm": 0.7335963845252991,
      "learning_rate": 1.8473567545495464e-05,
      "loss": 1.1878,
      "step": 1420
    },
    {
      "epoch": 0.4232394497291461,
      "grad_norm": 0.707231342792511,
      "learning_rate": 1.846648211986774e-05,
      "loss": 1.1779,
      "step": 1421
    },
    {
      "epoch": 0.4235372959288148,
      "grad_norm": 0.7129309773445129,
      "learning_rate": 1.8459381653004495e-05,
      "loss": 1.1618,
      "step": 1422
    },
    {
      "epoch": 0.4238351421284834,
      "grad_norm": 0.7050052881240845,
      "learning_rate": 1.8452266157520156e-05,
      "loss": 1.1829,
      "step": 1423
    },
    {
      "epoch": 0.42413298832815205,
      "grad_norm": 0.7216488122940063,
      "learning_rate": 1.8445135646055844e-05,
      "loss": 1.1814,
      "step": 1424
    },
    {
      "epoch": 0.4244308345278207,
      "grad_norm": 0.7108482718467712,
      "learning_rate": 1.8437990131279364e-05,
      "loss": 1.1778,
      "step": 1425
    },
    {
      "epoch": 0.4247286807274893,
      "grad_norm": 0.7019352912902832,
      "learning_rate": 1.8430829625885166e-05,
      "loss": 1.167,
      "step": 1426
    },
    {
      "epoch": 0.425026526927158,
      "grad_norm": 0.713994026184082,
      "learning_rate": 1.8423654142594333e-05,
      "loss": 1.1693,
      "step": 1427
    },
    {
      "epoch": 0.42532437312682664,
      "grad_norm": 0.7363821864128113,
      "learning_rate": 1.8416463694154564e-05,
      "loss": 1.1713,
      "step": 1428
    },
    {
      "epoch": 0.4256222193264953,
      "grad_norm": 0.7393612861633301,
      "learning_rate": 1.8409258293340146e-05,
      "loss": 1.1634,
      "step": 1429
    },
    {
      "epoch": 0.4259200655261639,
      "grad_norm": 0.7492763996124268,
      "learning_rate": 1.8402037952951916e-05,
      "loss": 1.1785,
      "step": 1430
    },
    {
      "epoch": 0.42621791172583257,
      "grad_norm": 0.7383298277854919,
      "learning_rate": 1.8394802685817262e-05,
      "loss": 1.1877,
      "step": 1431
    },
    {
      "epoch": 0.4265157579255012,
      "grad_norm": 0.7454927563667297,
      "learning_rate": 1.8387552504790097e-05,
      "loss": 1.1637,
      "step": 1432
    },
    {
      "epoch": 0.4268136041251699,
      "grad_norm": 0.7243757247924805,
      "learning_rate": 1.8380287422750812e-05,
      "loss": 1.1813,
      "step": 1433
    },
    {
      "epoch": 0.4271114503248385,
      "grad_norm": 0.7181768417358398,
      "learning_rate": 1.8373007452606283e-05,
      "loss": 1.1813,
      "step": 1434
    },
    {
      "epoch": 0.42740929652450715,
      "grad_norm": 0.6943856477737427,
      "learning_rate": 1.8365712607289835e-05,
      "loss": 1.1769,
      "step": 1435
    },
    {
      "epoch": 0.4277071427241758,
      "grad_norm": 0.7518101930618286,
      "learning_rate": 1.8358402899761218e-05,
      "loss": 1.1779,
      "step": 1436
    },
    {
      "epoch": 0.4280049889238445,
      "grad_norm": 0.7290526628494263,
      "learning_rate": 1.8351078343006583e-05,
      "loss": 1.1819,
      "step": 1437
    },
    {
      "epoch": 0.4283028351235131,
      "grad_norm": 0.7149537205696106,
      "learning_rate": 1.834373895003846e-05,
      "loss": 1.1742,
      "step": 1438
    },
    {
      "epoch": 0.42860068132318174,
      "grad_norm": 0.7344151735305786,
      "learning_rate": 1.8336384733895748e-05,
      "loss": 1.1722,
      "step": 1439
    },
    {
      "epoch": 0.4288985275228504,
      "grad_norm": 0.739539384841919,
      "learning_rate": 1.8329015707643665e-05,
      "loss": 1.1696,
      "step": 1440
    },
    {
      "epoch": 0.429196373722519,
      "grad_norm": 0.7256489992141724,
      "learning_rate": 1.8321631884373753e-05,
      "loss": 1.1832,
      "step": 1441
    },
    {
      "epoch": 0.42949421992218767,
      "grad_norm": 0.7507062554359436,
      "learning_rate": 1.831423327720383e-05,
      "loss": 1.156,
      "step": 1442
    },
    {
      "epoch": 0.42979206612185633,
      "grad_norm": 0.7506979703903198,
      "learning_rate": 1.830681989927799e-05,
      "loss": 1.1851,
      "step": 1443
    },
    {
      "epoch": 0.430089912321525,
      "grad_norm": 0.7234819531440735,
      "learning_rate": 1.829939176376656e-05,
      "loss": 1.1705,
      "step": 1444
    },
    {
      "epoch": 0.4303877585211936,
      "grad_norm": 0.7429983615875244,
      "learning_rate": 1.8291948883866095e-05,
      "loss": 1.1885,
      "step": 1445
    },
    {
      "epoch": 0.43068560472086226,
      "grad_norm": 0.6876159310340881,
      "learning_rate": 1.8284491272799327e-05,
      "loss": 1.1711,
      "step": 1446
    },
    {
      "epoch": 0.4309834509205309,
      "grad_norm": 0.7321100234985352,
      "learning_rate": 1.8277018943815173e-05,
      "loss": 1.1427,
      "step": 1447
    },
    {
      "epoch": 0.4312812971201996,
      "grad_norm": 0.7354869246482849,
      "learning_rate": 1.82695319101887e-05,
      "loss": 1.1853,
      "step": 1448
    },
    {
      "epoch": 0.4315791433198682,
      "grad_norm": 0.7836506962776184,
      "learning_rate": 1.826203018522108e-05,
      "loss": 1.1701,
      "step": 1449
    },
    {
      "epoch": 0.43187698951953685,
      "grad_norm": 0.7324841022491455,
      "learning_rate": 1.82545137822396e-05,
      "loss": 1.1746,
      "step": 1450
    },
    {
      "epoch": 0.4321748357192055,
      "grad_norm": 0.7387935519218445,
      "learning_rate": 1.824698271459763e-05,
      "loss": 1.2069,
      "step": 1451
    },
    {
      "epoch": 0.43247268191887417,
      "grad_norm": 0.7598223686218262,
      "learning_rate": 1.823943699567457e-05,
      "loss": 1.1683,
      "step": 1452
    },
    {
      "epoch": 0.4327705281185428,
      "grad_norm": 0.7654528021812439,
      "learning_rate": 1.8231876638875873e-05,
      "loss": 1.1643,
      "step": 1453
    },
    {
      "epoch": 0.43306837431821144,
      "grad_norm": 0.7328842282295227,
      "learning_rate": 1.822430165763298e-05,
      "loss": 1.1701,
      "step": 1454
    },
    {
      "epoch": 0.4333662205178801,
      "grad_norm": 0.8112918138504028,
      "learning_rate": 1.8216712065403318e-05,
      "loss": 1.1908,
      "step": 1455
    },
    {
      "epoch": 0.4336640667175487,
      "grad_norm": 0.7139400839805603,
      "learning_rate": 1.8209107875670278e-05,
      "loss": 1.1966,
      "step": 1456
    },
    {
      "epoch": 0.43396191291721736,
      "grad_norm": 0.7561649680137634,
      "learning_rate": 1.8201489101943178e-05,
      "loss": 1.1883,
      "step": 1457
    },
    {
      "epoch": 0.434259759116886,
      "grad_norm": 0.725346565246582,
      "learning_rate": 1.8193855757757242e-05,
      "loss": 1.173,
      "step": 1458
    },
    {
      "epoch": 0.4345576053165547,
      "grad_norm": 0.7543953061103821,
      "learning_rate": 1.818620785667359e-05,
      "loss": 1.1861,
      "step": 1459
    },
    {
      "epoch": 0.4348554515162233,
      "grad_norm": 0.713070809841156,
      "learning_rate": 1.8178545412279194e-05,
      "loss": 1.1774,
      "step": 1460
    },
    {
      "epoch": 0.43515329771589195,
      "grad_norm": 0.7358941435813904,
      "learning_rate": 1.8170868438186862e-05,
      "loss": 1.1613,
      "step": 1461
    },
    {
      "epoch": 0.4354511439155606,
      "grad_norm": 0.7270810008049011,
      "learning_rate": 1.816317694803523e-05,
      "loss": 1.167,
      "step": 1462
    },
    {
      "epoch": 0.4357489901152293,
      "grad_norm": 0.7366968989372253,
      "learning_rate": 1.81554709554887e-05,
      "loss": 1.1916,
      "step": 1463
    },
    {
      "epoch": 0.4360468363148979,
      "grad_norm": 0.7492762804031372,
      "learning_rate": 1.8147750474237456e-05,
      "loss": 1.1658,
      "step": 1464
    },
    {
      "epoch": 0.43634468251456654,
      "grad_norm": 0.7266594171524048,
      "learning_rate": 1.814001551799742e-05,
      "loss": 1.173,
      "step": 1465
    },
    {
      "epoch": 0.4366425287142352,
      "grad_norm": 0.7668642997741699,
      "learning_rate": 1.8132266100510217e-05,
      "loss": 1.1601,
      "step": 1466
    },
    {
      "epoch": 0.4369403749139038,
      "grad_norm": 0.7565826773643494,
      "learning_rate": 1.8124502235543174e-05,
      "loss": 1.1564,
      "step": 1467
    },
    {
      "epoch": 0.43723822111357247,
      "grad_norm": 0.7129597067832947,
      "learning_rate": 1.8116723936889294e-05,
      "loss": 1.1773,
      "step": 1468
    },
    {
      "epoch": 0.43753606731324113,
      "grad_norm": 0.8002951145172119,
      "learning_rate": 1.8108931218367198e-05,
      "loss": 1.1823,
      "step": 1469
    },
    {
      "epoch": 0.4378339135129098,
      "grad_norm": 0.7371430397033691,
      "learning_rate": 1.8101124093821144e-05,
      "loss": 1.1611,
      "step": 1470
    },
    {
      "epoch": 0.4381317597125784,
      "grad_norm": 0.7776963710784912,
      "learning_rate": 1.809330257712098e-05,
      "loss": 1.1776,
      "step": 1471
    },
    {
      "epoch": 0.43842960591224706,
      "grad_norm": 0.6851039528846741,
      "learning_rate": 1.8085466682162116e-05,
      "loss": 1.1736,
      "step": 1472
    },
    {
      "epoch": 0.4387274521119157,
      "grad_norm": 0.7555819153785706,
      "learning_rate": 1.807761642286551e-05,
      "loss": 1.1837,
      "step": 1473
    },
    {
      "epoch": 0.4390252983115844,
      "grad_norm": 0.7075178027153015,
      "learning_rate": 1.806975181317764e-05,
      "loss": 1.1565,
      "step": 1474
    },
    {
      "epoch": 0.439323144511253,
      "grad_norm": 0.7452244758605957,
      "learning_rate": 1.806187286707048e-05,
      "loss": 1.158,
      "step": 1475
    },
    {
      "epoch": 0.43962099071092164,
      "grad_norm": 0.7333664298057556,
      "learning_rate": 1.8053979598541473e-05,
      "loss": 1.1823,
      "step": 1476
    },
    {
      "epoch": 0.4399188369105903,
      "grad_norm": 0.7648371458053589,
      "learning_rate": 1.8046072021613496e-05,
      "loss": 1.1878,
      "step": 1477
    },
    {
      "epoch": 0.44021668311025897,
      "grad_norm": 0.754614531993866,
      "learning_rate": 1.8038150150334856e-05,
      "loss": 1.1594,
      "step": 1478
    },
    {
      "epoch": 0.44051452930992757,
      "grad_norm": 0.7922732830047607,
      "learning_rate": 1.803021399877926e-05,
      "loss": 1.1477,
      "step": 1479
    },
    {
      "epoch": 0.44081237550959623,
      "grad_norm": 0.8391221165657043,
      "learning_rate": 1.8022263581045775e-05,
      "loss": 1.1696,
      "step": 1480
    },
    {
      "epoch": 0.4411102217092649,
      "grad_norm": 0.7397218942642212,
      "learning_rate": 1.8014298911258813e-05,
      "loss": 1.1674,
      "step": 1481
    },
    {
      "epoch": 0.4414080679089335,
      "grad_norm": 0.8336960673332214,
      "learning_rate": 1.800632000356811e-05,
      "loss": 1.1698,
      "step": 1482
    },
    {
      "epoch": 0.44170591410860216,
      "grad_norm": 0.7441984415054321,
      "learning_rate": 1.79983268721487e-05,
      "loss": 1.1725,
      "step": 1483
    },
    {
      "epoch": 0.4420037603082708,
      "grad_norm": 0.7343475818634033,
      "learning_rate": 1.799031953120087e-05,
      "loss": 1.1581,
      "step": 1484
    },
    {
      "epoch": 0.4423016065079395,
      "grad_norm": 0.7276702523231506,
      "learning_rate": 1.798229799495018e-05,
      "loss": 1.1656,
      "step": 1485
    },
    {
      "epoch": 0.4425994527076081,
      "grad_norm": 0.7294262647628784,
      "learning_rate": 1.7974262277647376e-05,
      "loss": 1.1689,
      "step": 1486
    },
    {
      "epoch": 0.44289729890727675,
      "grad_norm": 0.7017423510551453,
      "learning_rate": 1.7966212393568418e-05,
      "loss": 1.1568,
      "step": 1487
    },
    {
      "epoch": 0.4431951451069454,
      "grad_norm": 0.7039442658424377,
      "learning_rate": 1.7958148357014424e-05,
      "loss": 1.1535,
      "step": 1488
    },
    {
      "epoch": 0.44349299130661407,
      "grad_norm": 0.7390060424804688,
      "learning_rate": 1.795007018231167e-05,
      "loss": 1.1561,
      "step": 1489
    },
    {
      "epoch": 0.4437908375062827,
      "grad_norm": 0.7517692446708679,
      "learning_rate": 1.794197788381153e-05,
      "loss": 1.179,
      "step": 1490
    },
    {
      "epoch": 0.44408868370595134,
      "grad_norm": 0.7245727181434631,
      "learning_rate": 1.7933871475890484e-05,
      "loss": 1.1638,
      "step": 1491
    },
    {
      "epoch": 0.44438652990562,
      "grad_norm": 0.7294563055038452,
      "learning_rate": 1.7925750972950074e-05,
      "loss": 1.1744,
      "step": 1492
    },
    {
      "epoch": 0.44468437610528866,
      "grad_norm": 0.7747794389724731,
      "learning_rate": 1.791761638941688e-05,
      "loss": 1.1818,
      "step": 1493
    },
    {
      "epoch": 0.44498222230495726,
      "grad_norm": 0.7011266946792603,
      "learning_rate": 1.7909467739742494e-05,
      "loss": 1.1736,
      "step": 1494
    },
    {
      "epoch": 0.4452800685046259,
      "grad_norm": 0.7643789052963257,
      "learning_rate": 1.790130503840351e-05,
      "loss": 1.1682,
      "step": 1495
    },
    {
      "epoch": 0.4455779147042946,
      "grad_norm": 0.7668893337249756,
      "learning_rate": 1.789312829990147e-05,
      "loss": 1.1825,
      "step": 1496
    },
    {
      "epoch": 0.4458757609039632,
      "grad_norm": 0.7049111723899841,
      "learning_rate": 1.7884937538762872e-05,
      "loss": 1.1857,
      "step": 1497
    },
    {
      "epoch": 0.44617360710363185,
      "grad_norm": 0.7034857869148254,
      "learning_rate": 1.7876732769539108e-05,
      "loss": 1.1547,
      "step": 1498
    },
    {
      "epoch": 0.4464714533033005,
      "grad_norm": 0.7338677644729614,
      "learning_rate": 1.786851400680647e-05,
      "loss": 1.1659,
      "step": 1499
    },
    {
      "epoch": 0.4467692995029692,
      "grad_norm": 0.7163870334625244,
      "learning_rate": 1.7860281265166097e-05,
      "loss": 1.1669,
      "step": 1500
    },
    {
      "epoch": 0.4470671457026378,
      "grad_norm": 0.7136034369468689,
      "learning_rate": 1.7852034559243977e-05,
      "loss": 1.169,
      "step": 1501
    },
    {
      "epoch": 0.44736499190230644,
      "grad_norm": 0.7424219846725464,
      "learning_rate": 1.7843773903690896e-05,
      "loss": 1.1597,
      "step": 1502
    },
    {
      "epoch": 0.4476628381019751,
      "grad_norm": 0.7345615029335022,
      "learning_rate": 1.7835499313182426e-05,
      "loss": 1.1615,
      "step": 1503
    },
    {
      "epoch": 0.44796068430164376,
      "grad_norm": 0.7402758002281189,
      "learning_rate": 1.7827210802418894e-05,
      "loss": 1.1656,
      "step": 1504
    },
    {
      "epoch": 0.44825853050131237,
      "grad_norm": 0.6944360733032227,
      "learning_rate": 1.7818908386125358e-05,
      "loss": 1.166,
      "step": 1505
    },
    {
      "epoch": 0.44855637670098103,
      "grad_norm": 0.7249569296836853,
      "learning_rate": 1.7810592079051586e-05,
      "loss": 1.177,
      "step": 1506
    },
    {
      "epoch": 0.4488542229006497,
      "grad_norm": 0.7216330766677856,
      "learning_rate": 1.780226189597201e-05,
      "loss": 1.143,
      "step": 1507
    },
    {
      "epoch": 0.4491520691003183,
      "grad_norm": 0.6777287125587463,
      "learning_rate": 1.7793917851685725e-05,
      "loss": 1.1558,
      "step": 1508
    },
    {
      "epoch": 0.44944991529998696,
      "grad_norm": 0.7455222010612488,
      "learning_rate": 1.7785559961016452e-05,
      "loss": 1.1769,
      "step": 1509
    },
    {
      "epoch": 0.4497477614996556,
      "grad_norm": 0.6941633224487305,
      "learning_rate": 1.7777188238812503e-05,
      "loss": 1.1696,
      "step": 1510
    },
    {
      "epoch": 0.4500456076993243,
      "grad_norm": 0.7095164060592651,
      "learning_rate": 1.7768802699946767e-05,
      "loss": 1.161,
      "step": 1511
    },
    {
      "epoch": 0.4503434538989929,
      "grad_norm": 0.6937164664268494,
      "learning_rate": 1.776040335931668e-05,
      "loss": 1.1816,
      "step": 1512
    },
    {
      "epoch": 0.45064130009866155,
      "grad_norm": 0.7406914234161377,
      "learning_rate": 1.7751990231844195e-05,
      "loss": 1.1661,
      "step": 1513
    },
    {
      "epoch": 0.4509391462983302,
      "grad_norm": 0.695449709892273,
      "learning_rate": 1.774356333247576e-05,
      "loss": 1.1821,
      "step": 1514
    },
    {
      "epoch": 0.45123699249799887,
      "grad_norm": 0.706566333770752,
      "learning_rate": 1.7735122676182288e-05,
      "loss": 1.176,
      "step": 1515
    },
    {
      "epoch": 0.4515348386976675,
      "grad_norm": 0.7057842016220093,
      "learning_rate": 1.7726668277959137e-05,
      "loss": 1.1582,
      "step": 1516
    },
    {
      "epoch": 0.45183268489733613,
      "grad_norm": 0.6889585852622986,
      "learning_rate": 1.771820015282607e-05,
      "loss": 1.1677,
      "step": 1517
    },
    {
      "epoch": 0.4521305310970048,
      "grad_norm": 0.7395607829093933,
      "learning_rate": 1.7709718315827246e-05,
      "loss": 1.1578,
      "step": 1518
    },
    {
      "epoch": 0.45242837729667346,
      "grad_norm": 0.7199034690856934,
      "learning_rate": 1.7701222782031177e-05,
      "loss": 1.1649,
      "step": 1519
    },
    {
      "epoch": 0.45272622349634206,
      "grad_norm": 0.6963168978691101,
      "learning_rate": 1.7692713566530712e-05,
      "loss": 1.1629,
      "step": 1520
    },
    {
      "epoch": 0.4530240696960107,
      "grad_norm": 0.7118576765060425,
      "learning_rate": 1.7684190684443003e-05,
      "loss": 1.1676,
      "step": 1521
    },
    {
      "epoch": 0.4533219158956794,
      "grad_norm": 0.7198029160499573,
      "learning_rate": 1.7675654150909485e-05,
      "loss": 1.1817,
      "step": 1522
    },
    {
      "epoch": 0.453619762095348,
      "grad_norm": 0.7912507057189941,
      "learning_rate": 1.7667103981095844e-05,
      "loss": 1.1702,
      "step": 1523
    },
    {
      "epoch": 0.45391760829501665,
      "grad_norm": 0.7488912343978882,
      "learning_rate": 1.7658540190191992e-05,
      "loss": 1.1897,
      "step": 1524
    },
    {
      "epoch": 0.4542154544946853,
      "grad_norm": 0.7358275651931763,
      "learning_rate": 1.7649962793412036e-05,
      "loss": 1.1699,
      "step": 1525
    },
    {
      "epoch": 0.45451330069435397,
      "grad_norm": 0.7243742942810059,
      "learning_rate": 1.7641371805994266e-05,
      "loss": 1.1632,
      "step": 1526
    },
    {
      "epoch": 0.4548111468940226,
      "grad_norm": 0.8163059949874878,
      "learning_rate": 1.7632767243201104e-05,
      "loss": 1.1743,
      "step": 1527
    },
    {
      "epoch": 0.45510899309369124,
      "grad_norm": 0.7063804864883423,
      "learning_rate": 1.7624149120319092e-05,
      "loss": 1.1834,
      "step": 1528
    },
    {
      "epoch": 0.4554068392933599,
      "grad_norm": 0.7055456042289734,
      "learning_rate": 1.7615517452658873e-05,
      "loss": 1.1294,
      "step": 1529
    },
    {
      "epoch": 0.45570468549302856,
      "grad_norm": 0.7424860596656799,
      "learning_rate": 1.760687225555514e-05,
      "loss": 1.1674,
      "step": 1530
    },
    {
      "epoch": 0.45600253169269717,
      "grad_norm": 0.7328478693962097,
      "learning_rate": 1.7598213544366624e-05,
      "loss": 1.1727,
      "step": 1531
    },
    {
      "epoch": 0.4563003778923658,
      "grad_norm": 0.7771193385124207,
      "learning_rate": 1.758954133447607e-05,
      "loss": 1.1628,
      "step": 1532
    },
    {
      "epoch": 0.4565982240920345,
      "grad_norm": 0.7342403531074524,
      "learning_rate": 1.7580855641290203e-05,
      "loss": 1.1697,
      "step": 1533
    },
    {
      "epoch": 0.45689607029170315,
      "grad_norm": 0.7574604153633118,
      "learning_rate": 1.7572156480239702e-05,
      "loss": 1.1614,
      "step": 1534
    },
    {
      "epoch": 0.45719391649137175,
      "grad_norm": 0.6881849765777588,
      "learning_rate": 1.756344386677917e-05,
      "loss": 1.1531,
      "step": 1535
    },
    {
      "epoch": 0.4574917626910404,
      "grad_norm": 0.7295597791671753,
      "learning_rate": 1.755471781638711e-05,
      "loss": 1.1762,
      "step": 1536
    },
    {
      "epoch": 0.4577896088907091,
      "grad_norm": 0.7741530537605286,
      "learning_rate": 1.754597834456589e-05,
      "loss": 1.1655,
      "step": 1537
    },
    {
      "epoch": 0.4580874550903777,
      "grad_norm": 0.6921853423118591,
      "learning_rate": 1.753722546684174e-05,
      "loss": 1.1655,
      "step": 1538
    },
    {
      "epoch": 0.45838530129004634,
      "grad_norm": 0.7381775975227356,
      "learning_rate": 1.7528459198764688e-05,
      "loss": 1.1573,
      "step": 1539
    },
    {
      "epoch": 0.458683147489715,
      "grad_norm": 0.678419291973114,
      "learning_rate": 1.751967955590856e-05,
      "loss": 1.1683,
      "step": 1540
    },
    {
      "epoch": 0.45898099368938367,
      "grad_norm": 0.7309117913246155,
      "learning_rate": 1.751088655387094e-05,
      "loss": 1.1879,
      "step": 1541
    },
    {
      "epoch": 0.45927883988905227,
      "grad_norm": 0.7030755877494812,
      "learning_rate": 1.7502080208273152e-05,
      "loss": 1.163,
      "step": 1542
    },
    {
      "epoch": 0.45957668608872093,
      "grad_norm": 0.6959496736526489,
      "learning_rate": 1.749326053476022e-05,
      "loss": 1.1646,
      "step": 1543
    },
    {
      "epoch": 0.4598745322883896,
      "grad_norm": 0.6901364922523499,
      "learning_rate": 1.7484427549000836e-05,
      "loss": 1.1702,
      "step": 1544
    },
    {
      "epoch": 0.46017237848805825,
      "grad_norm": 0.7237486839294434,
      "learning_rate": 1.7475581266687362e-05,
      "loss": 1.1748,
      "step": 1545
    },
    {
      "epoch": 0.46047022468772686,
      "grad_norm": 0.7074930667877197,
      "learning_rate": 1.7466721703535765e-05,
      "loss": 1.163,
      "step": 1546
    },
    {
      "epoch": 0.4607680708873955,
      "grad_norm": 0.716848611831665,
      "learning_rate": 1.7457848875285618e-05,
      "loss": 1.1485,
      "step": 1547
    },
    {
      "epoch": 0.4610659170870642,
      "grad_norm": 0.7023268342018127,
      "learning_rate": 1.744896279770006e-05,
      "loss": 1.1753,
      "step": 1548
    },
    {
      "epoch": 0.4613637632867328,
      "grad_norm": 0.7246884703636169,
      "learning_rate": 1.7440063486565755e-05,
      "loss": 1.1456,
      "step": 1549
    },
    {
      "epoch": 0.46166160948640145,
      "grad_norm": 0.7001634240150452,
      "learning_rate": 1.7431150957692896e-05,
      "loss": 1.1692,
      "step": 1550
    },
    {
      "epoch": 0.4619594556860701,
      "grad_norm": 0.7504196763038635,
      "learning_rate": 1.7422225226915138e-05,
      "loss": 1.1742,
      "step": 1551
    },
    {
      "epoch": 0.46225730188573877,
      "grad_norm": 0.7059693336486816,
      "learning_rate": 1.741328631008961e-05,
      "loss": 1.1702,
      "step": 1552
    },
    {
      "epoch": 0.4625551480854074,
      "grad_norm": 0.7596415281295776,
      "learning_rate": 1.7404334223096852e-05,
      "loss": 1.1674,
      "step": 1553
    },
    {
      "epoch": 0.46285299428507604,
      "grad_norm": 0.7759929299354553,
      "learning_rate": 1.7395368981840804e-05,
      "loss": 1.1573,
      "step": 1554
    },
    {
      "epoch": 0.4631508404847447,
      "grad_norm": 0.7179751396179199,
      "learning_rate": 1.7386390602248787e-05,
      "loss": 1.1793,
      "step": 1555
    },
    {
      "epoch": 0.46344868668441336,
      "grad_norm": 0.7769747972488403,
      "learning_rate": 1.737739910027145e-05,
      "loss": 1.1727,
      "step": 1556
    },
    {
      "epoch": 0.46374653288408196,
      "grad_norm": 0.707478404045105,
      "learning_rate": 1.7368394491882757e-05,
      "loss": 1.1654,
      "step": 1557
    },
    {
      "epoch": 0.4640443790837506,
      "grad_norm": 0.715491771697998,
      "learning_rate": 1.7359376793079963e-05,
      "loss": 1.1654,
      "step": 1558
    },
    {
      "epoch": 0.4643422252834193,
      "grad_norm": 0.7617186903953552,
      "learning_rate": 1.735034601988357e-05,
      "loss": 1.1621,
      "step": 1559
    },
    {
      "epoch": 0.46464007148308795,
      "grad_norm": 0.7010698318481445,
      "learning_rate": 1.734130218833732e-05,
      "loss": 1.1699,
      "step": 1560
    },
    {
      "epoch": 0.46493791768275655,
      "grad_norm": 0.7178402543067932,
      "learning_rate": 1.7332245314508137e-05,
      "loss": 1.1662,
      "step": 1561
    },
    {
      "epoch": 0.4652357638824252,
      "grad_norm": 0.7282794117927551,
      "learning_rate": 1.7323175414486125e-05,
      "loss": 1.1825,
      "step": 1562
    },
    {
      "epoch": 0.4655336100820939,
      "grad_norm": 0.7237359881401062,
      "learning_rate": 1.731409250438453e-05,
      "loss": 1.1789,
      "step": 1563
    },
    {
      "epoch": 0.4658314562817625,
      "grad_norm": 0.691853940486908,
      "learning_rate": 1.7304996600339718e-05,
      "loss": 1.1671,
      "step": 1564
    },
    {
      "epoch": 0.46612930248143114,
      "grad_norm": 0.6950562596321106,
      "learning_rate": 1.7295887718511123e-05,
      "loss": 1.1865,
      "step": 1565
    },
    {
      "epoch": 0.4664271486810998,
      "grad_norm": 0.7177675366401672,
      "learning_rate": 1.7286765875081243e-05,
      "loss": 1.1646,
      "step": 1566
    },
    {
      "epoch": 0.46672499488076846,
      "grad_norm": 0.6940879821777344,
      "learning_rate": 1.727763108625561e-05,
      "loss": 1.1588,
      "step": 1567
    },
    {
      "epoch": 0.46702284108043707,
      "grad_norm": 0.7044256329536438,
      "learning_rate": 1.726848336826275e-05,
      "loss": 1.1561,
      "step": 1568
    },
    {
      "epoch": 0.46732068728010573,
      "grad_norm": 0.7212578654289246,
      "learning_rate": 1.725932273735415e-05,
      "loss": 1.1781,
      "step": 1569
    },
    {
      "epoch": 0.4676185334797744,
      "grad_norm": 0.6851856708526611,
      "learning_rate": 1.7250149209804247e-05,
      "loss": 1.1721,
      "step": 1570
    },
    {
      "epoch": 0.46791637967944305,
      "grad_norm": 0.6951323747634888,
      "learning_rate": 1.7240962801910387e-05,
      "loss": 1.1454,
      "step": 1571
    },
    {
      "epoch": 0.46821422587911166,
      "grad_norm": 0.7196218371391296,
      "learning_rate": 1.7231763529992796e-05,
      "loss": 1.1642,
      "step": 1572
    },
    {
      "epoch": 0.4685120720787803,
      "grad_norm": 0.7157500982284546,
      "learning_rate": 1.7222551410394564e-05,
      "loss": 1.1589,
      "step": 1573
    },
    {
      "epoch": 0.468809918278449,
      "grad_norm": 0.7000492215156555,
      "learning_rate": 1.721332645948159e-05,
      "loss": 1.138,
      "step": 1574
    },
    {
      "epoch": 0.46910776447811764,
      "grad_norm": 0.6970203518867493,
      "learning_rate": 1.720408869364258e-05,
      "loss": 1.1613,
      "step": 1575
    },
    {
      "epoch": 0.46940561067778624,
      "grad_norm": 0.6970378160476685,
      "learning_rate": 1.7194838129289006e-05,
      "loss": 1.1836,
      "step": 1576
    },
    {
      "epoch": 0.4697034568774549,
      "grad_norm": 0.7003931403160095,
      "learning_rate": 1.718557478285507e-05,
      "loss": 1.1583,
      "step": 1577
    },
    {
      "epoch": 0.47000130307712357,
      "grad_norm": 0.6910634636878967,
      "learning_rate": 1.717629867079769e-05,
      "loss": 1.1688,
      "step": 1578
    },
    {
      "epoch": 0.47029914927679217,
      "grad_norm": 0.6990684270858765,
      "learning_rate": 1.7167009809596457e-05,
      "loss": 1.1677,
      "step": 1579
    },
    {
      "epoch": 0.47059699547646083,
      "grad_norm": 0.679383397102356,
      "learning_rate": 1.7157708215753615e-05,
      "loss": 1.1529,
      "step": 1580
    },
    {
      "epoch": 0.4708948416761295,
      "grad_norm": 0.7120311260223389,
      "learning_rate": 1.7148393905794036e-05,
      "loss": 1.1519,
      "step": 1581
    },
    {
      "epoch": 0.47119268787579816,
      "grad_norm": 0.7229364514350891,
      "learning_rate": 1.7139066896265162e-05,
      "loss": 1.171,
      "step": 1582
    },
    {
      "epoch": 0.47149053407546676,
      "grad_norm": 0.6965063214302063,
      "learning_rate": 1.7129727203737018e-05,
      "loss": 1.147,
      "step": 1583
    },
    {
      "epoch": 0.4717883802751354,
      "grad_norm": 0.7071117162704468,
      "learning_rate": 1.712037484480215e-05,
      "loss": 1.1735,
      "step": 1584
    },
    {
      "epoch": 0.4720862264748041,
      "grad_norm": 0.7141831517219543,
      "learning_rate": 1.7111009836075604e-05,
      "loss": 1.1722,
      "step": 1585
    },
    {
      "epoch": 0.47238407267447274,
      "grad_norm": 0.7130959630012512,
      "learning_rate": 1.710163219419491e-05,
      "loss": 1.1547,
      "step": 1586
    },
    {
      "epoch": 0.47268191887414135,
      "grad_norm": 0.7031484246253967,
      "learning_rate": 1.7092241935820036e-05,
      "loss": 1.1466,
      "step": 1587
    },
    {
      "epoch": 0.47297976507381,
      "grad_norm": 0.725978672504425,
      "learning_rate": 1.7082839077633363e-05,
      "loss": 1.1822,
      "step": 1588
    },
    {
      "epoch": 0.47327761127347867,
      "grad_norm": 0.7073671817779541,
      "learning_rate": 1.707342363633966e-05,
      "loss": 1.1578,
      "step": 1589
    },
    {
      "epoch": 0.4735754574731473,
      "grad_norm": 0.715928852558136,
      "learning_rate": 1.706399562866604e-05,
      "loss": 1.1543,
      "step": 1590
    },
    {
      "epoch": 0.47387330367281594,
      "grad_norm": 0.700507402420044,
      "learning_rate": 1.7054555071361954e-05,
      "loss": 1.1513,
      "step": 1591
    },
    {
      "epoch": 0.4741711498724846,
      "grad_norm": 0.6934207081794739,
      "learning_rate": 1.7045101981199144e-05,
      "loss": 1.1559,
      "step": 1592
    },
    {
      "epoch": 0.47446899607215326,
      "grad_norm": 0.6781994700431824,
      "learning_rate": 1.7035636374971618e-05,
      "loss": 1.1629,
      "step": 1593
    },
    {
      "epoch": 0.47476684227182187,
      "grad_norm": 0.7108228206634521,
      "learning_rate": 1.7026158269495612e-05,
      "loss": 1.169,
      "step": 1594
    },
    {
      "epoch": 0.4750646884714905,
      "grad_norm": 0.7082929015159607,
      "learning_rate": 1.701666768160958e-05,
      "loss": 1.1598,
      "step": 1595
    },
    {
      "epoch": 0.4753625346711592,
      "grad_norm": 0.7200596332550049,
      "learning_rate": 1.700716462817414e-05,
      "loss": 1.1557,
      "step": 1596
    },
    {
      "epoch": 0.47566038087082785,
      "grad_norm": 0.726783037185669,
      "learning_rate": 1.6997649126072064e-05,
      "loss": 1.1793,
      "step": 1597
    },
    {
      "epoch": 0.47595822707049645,
      "grad_norm": 0.703177809715271,
      "learning_rate": 1.6988121192208237e-05,
      "loss": 1.1758,
      "step": 1598
    },
    {
      "epoch": 0.4762560732701651,
      "grad_norm": 0.7500465512275696,
      "learning_rate": 1.6978580843509635e-05,
      "loss": 1.1713,
      "step": 1599
    },
    {
      "epoch": 0.4765539194698338,
      "grad_norm": 0.7249962687492371,
      "learning_rate": 1.6969028096925275e-05,
      "loss": 1.186,
      "step": 1600
    },
    {
      "epoch": 0.47685176566950244,
      "grad_norm": 0.7242220044136047,
      "learning_rate": 1.6959462969426215e-05,
      "loss": 1.1591,
      "step": 1601
    },
    {
      "epoch": 0.47714961186917104,
      "grad_norm": 0.672217845916748,
      "learning_rate": 1.6949885478005497e-05,
      "loss": 1.1611,
      "step": 1602
    },
    {
      "epoch": 0.4774474580688397,
      "grad_norm": 0.6888530254364014,
      "learning_rate": 1.6940295639678147e-05,
      "loss": 1.1683,
      "step": 1603
    },
    {
      "epoch": 0.47774530426850836,
      "grad_norm": 0.7481662034988403,
      "learning_rate": 1.69306934714811e-05,
      "loss": 1.165,
      "step": 1604
    },
    {
      "epoch": 0.47804315046817697,
      "grad_norm": 0.6949246525764465,
      "learning_rate": 1.6921078990473216e-05,
      "loss": 1.1569,
      "step": 1605
    },
    {
      "epoch": 0.47834099666784563,
      "grad_norm": 0.6799510717391968,
      "learning_rate": 1.6911452213735223e-05,
      "loss": 1.1688,
      "step": 1606
    },
    {
      "epoch": 0.4786388428675143,
      "grad_norm": 0.6815329194068909,
      "learning_rate": 1.6901813158369686e-05,
      "loss": 1.1567,
      "step": 1607
    },
    {
      "epoch": 0.47893668906718295,
      "grad_norm": 0.6943604350090027,
      "learning_rate": 1.6892161841501e-05,
      "loss": 1.1611,
      "step": 1608
    },
    {
      "epoch": 0.47923453526685156,
      "grad_norm": 0.7006758451461792,
      "learning_rate": 1.6882498280275322e-05,
      "loss": 1.1695,
      "step": 1609
    },
    {
      "epoch": 0.4795323814665202,
      "grad_norm": 0.7200384736061096,
      "learning_rate": 1.6872822491860583e-05,
      "loss": 1.1508,
      "step": 1610
    },
    {
      "epoch": 0.4798302276661889,
      "grad_norm": 0.6886991262435913,
      "learning_rate": 1.686313449344642e-05,
      "loss": 1.1581,
      "step": 1611
    },
    {
      "epoch": 0.48012807386585754,
      "grad_norm": 0.7256563901901245,
      "learning_rate": 1.6853434302244175e-05,
      "loss": 1.1531,
      "step": 1612
    },
    {
      "epoch": 0.48042592006552615,
      "grad_norm": 0.718068540096283,
      "learning_rate": 1.684372193548684e-05,
      "loss": 1.1579,
      "step": 1613
    },
    {
      "epoch": 0.4807237662651948,
      "grad_norm": 0.7266986966133118,
      "learning_rate": 1.6833997410429046e-05,
      "loss": 1.1543,
      "step": 1614
    },
    {
      "epoch": 0.48102161246486347,
      "grad_norm": 0.7555680871009827,
      "learning_rate": 1.682426074434702e-05,
      "loss": 1.1671,
      "step": 1615
    },
    {
      "epoch": 0.48131945866453213,
      "grad_norm": 0.75299072265625,
      "learning_rate": 1.681451195453856e-05,
      "loss": 1.1721,
      "step": 1616
    },
    {
      "epoch": 0.48161730486420073,
      "grad_norm": 0.6986310482025146,
      "learning_rate": 1.6804751058322994e-05,
      "loss": 1.1539,
      "step": 1617
    },
    {
      "epoch": 0.4819151510638694,
      "grad_norm": 0.7601458430290222,
      "learning_rate": 1.6794978073041176e-05,
      "loss": 1.1789,
      "step": 1618
    },
    {
      "epoch": 0.48221299726353806,
      "grad_norm": 0.7315093278884888,
      "learning_rate": 1.6785193016055415e-05,
      "loss": 1.1777,
      "step": 1619
    },
    {
      "epoch": 0.48251084346320666,
      "grad_norm": 0.7502493262290955,
      "learning_rate": 1.6775395904749486e-05,
      "loss": 1.161,
      "step": 1620
    },
    {
      "epoch": 0.4828086896628753,
      "grad_norm": 0.7133820056915283,
      "learning_rate": 1.676558675652857e-05,
      "loss": 1.172,
      "step": 1621
    },
    {
      "epoch": 0.483106535862544,
      "grad_norm": 0.7319211363792419,
      "learning_rate": 1.6755765588819226e-05,
      "loss": 1.1574,
      "step": 1622
    },
    {
      "epoch": 0.48340438206221265,
      "grad_norm": 0.7203944325447083,
      "learning_rate": 1.674593241906938e-05,
      "loss": 1.1749,
      "step": 1623
    },
    {
      "epoch": 0.48370222826188125,
      "grad_norm": 0.7421601414680481,
      "learning_rate": 1.6736087264748267e-05,
      "loss": 1.1594,
      "step": 1624
    },
    {
      "epoch": 0.4840000744615499,
      "grad_norm": 0.6968820691108704,
      "learning_rate": 1.6726230143346433e-05,
      "loss": 1.1748,
      "step": 1625
    },
    {
      "epoch": 0.4842979206612186,
      "grad_norm": 0.7087630033493042,
      "learning_rate": 1.6716361072375657e-05,
      "loss": 1.1568,
      "step": 1626
    },
    {
      "epoch": 0.48459576686088723,
      "grad_norm": 0.7110248804092407,
      "learning_rate": 1.6706480069368968e-05,
      "loss": 1.1466,
      "step": 1627
    },
    {
      "epoch": 0.48489361306055584,
      "grad_norm": 0.7071283459663391,
      "learning_rate": 1.6696587151880586e-05,
      "loss": 1.1531,
      "step": 1628
    },
    {
      "epoch": 0.4851914592602245,
      "grad_norm": 0.6930647492408752,
      "learning_rate": 1.6686682337485897e-05,
      "loss": 1.1639,
      "step": 1629
    },
    {
      "epoch": 0.48548930545989316,
      "grad_norm": 0.6989061832427979,
      "learning_rate": 1.667676564378143e-05,
      "loss": 1.153,
      "step": 1630
    },
    {
      "epoch": 0.48578715165956177,
      "grad_norm": 0.6959564089775085,
      "learning_rate": 1.66668370883848e-05,
      "loss": 1.1712,
      "step": 1631
    },
    {
      "epoch": 0.4860849978592304,
      "grad_norm": 0.7126989364624023,
      "learning_rate": 1.6656896688934717e-05,
      "loss": 1.1646,
      "step": 1632
    },
    {
      "epoch": 0.4863828440588991,
      "grad_norm": 0.7302117347717285,
      "learning_rate": 1.6646944463090922e-05,
      "loss": 1.1575,
      "step": 1633
    },
    {
      "epoch": 0.48668069025856775,
      "grad_norm": 0.7120747566223145,
      "learning_rate": 1.6636980428534163e-05,
      "loss": 1.172,
      "step": 1634
    },
    {
      "epoch": 0.48697853645823636,
      "grad_norm": 0.6849662661552429,
      "learning_rate": 1.6627004602966176e-05,
      "loss": 1.1823,
      "step": 1635
    },
    {
      "epoch": 0.487276382657905,
      "grad_norm": 0.7090687155723572,
      "learning_rate": 1.661701700410963e-05,
      "loss": 1.1578,
      "step": 1636
    },
    {
      "epoch": 0.4875742288575737,
      "grad_norm": 0.7485430836677551,
      "learning_rate": 1.6607017649708133e-05,
      "loss": 1.1545,
      "step": 1637
    },
    {
      "epoch": 0.48787207505724234,
      "grad_norm": 0.6739163994789124,
      "learning_rate": 1.6597006557526156e-05,
      "loss": 1.1716,
      "step": 1638
    },
    {
      "epoch": 0.48816992125691094,
      "grad_norm": 0.6751441359519958,
      "learning_rate": 1.6586983745349033e-05,
      "loss": 1.1439,
      "step": 1639
    },
    {
      "epoch": 0.4884677674565796,
      "grad_norm": 0.7457461953163147,
      "learning_rate": 1.6576949230982918e-05,
      "loss": 1.1558,
      "step": 1640
    },
    {
      "epoch": 0.48876561365624827,
      "grad_norm": 0.7506716251373291,
      "learning_rate": 1.6566903032254754e-05,
      "loss": 1.1527,
      "step": 1641
    },
    {
      "epoch": 0.4890634598559169,
      "grad_norm": 0.7162187099456787,
      "learning_rate": 1.6556845167012238e-05,
      "loss": 1.158,
      "step": 1642
    },
    {
      "epoch": 0.48936130605558553,
      "grad_norm": 0.7258857488632202,
      "learning_rate": 1.6546775653123803e-05,
      "loss": 1.1574,
      "step": 1643
    },
    {
      "epoch": 0.4896591522552542,
      "grad_norm": 0.6966115832328796,
      "learning_rate": 1.6536694508478565e-05,
      "loss": 1.1625,
      "step": 1644
    },
    {
      "epoch": 0.48995699845492285,
      "grad_norm": 0.6851395964622498,
      "learning_rate": 1.6526601750986314e-05,
      "loss": 1.1581,
      "step": 1645
    },
    {
      "epoch": 0.49025484465459146,
      "grad_norm": 0.6830716729164124,
      "learning_rate": 1.651649739857746e-05,
      "loss": 1.1691,
      "step": 1646
    },
    {
      "epoch": 0.4905526908542601,
      "grad_norm": 0.7018412351608276,
      "learning_rate": 1.6506381469203025e-05,
      "loss": 1.1578,
      "step": 1647
    },
    {
      "epoch": 0.4908505370539288,
      "grad_norm": 0.7244733572006226,
      "learning_rate": 1.6496253980834586e-05,
      "loss": 1.1511,
      "step": 1648
    },
    {
      "epoch": 0.49114838325359744,
      "grad_norm": 0.7336034774780273,
      "learning_rate": 1.648611495146426e-05,
      "loss": 1.1665,
      "step": 1649
    },
    {
      "epoch": 0.49144622945326605,
      "grad_norm": 0.7560194730758667,
      "learning_rate": 1.647596439910467e-05,
      "loss": 1.1901,
      "step": 1650
    },
    {
      "epoch": 0.4917440756529347,
      "grad_norm": 0.7002713680267334,
      "learning_rate": 1.6465802341788903e-05,
      "loss": 1.1605,
      "step": 1651
    },
    {
      "epoch": 0.49204192185260337,
      "grad_norm": 0.752888560295105,
      "learning_rate": 1.6455628797570494e-05,
      "loss": 1.1718,
      "step": 1652
    },
    {
      "epoch": 0.49233976805227203,
      "grad_norm": 0.7282813787460327,
      "learning_rate": 1.644544378452338e-05,
      "loss": 1.1735,
      "step": 1653
    },
    {
      "epoch": 0.49263761425194064,
      "grad_norm": 0.7468515634536743,
      "learning_rate": 1.6435247320741873e-05,
      "loss": 1.1672,
      "step": 1654
    },
    {
      "epoch": 0.4929354604516093,
      "grad_norm": 0.8140262961387634,
      "learning_rate": 1.6425039424340633e-05,
      "loss": 1.1668,
      "step": 1655
    },
    {
      "epoch": 0.49323330665127796,
      "grad_norm": 0.7048836946487427,
      "learning_rate": 1.6414820113454624e-05,
      "loss": 1.1582,
      "step": 1656
    },
    {
      "epoch": 0.4935311528509466,
      "grad_norm": 0.7551459074020386,
      "learning_rate": 1.6404589406239094e-05,
      "loss": 1.1564,
      "step": 1657
    },
    {
      "epoch": 0.4938289990506152,
      "grad_norm": 0.7169157862663269,
      "learning_rate": 1.639434732086953e-05,
      "loss": 1.1416,
      "step": 1658
    },
    {
      "epoch": 0.4941268452502839,
      "grad_norm": 0.7279033660888672,
      "learning_rate": 1.6384093875541642e-05,
      "loss": 1.1718,
      "step": 1659
    },
    {
      "epoch": 0.49442469144995255,
      "grad_norm": 0.7511667013168335,
      "learning_rate": 1.637382908847132e-05,
      "loss": 1.1791,
      "step": 1660
    },
    {
      "epoch": 0.49472253764962115,
      "grad_norm": 0.7214314341545105,
      "learning_rate": 1.636355297789459e-05,
      "loss": 1.1789,
      "step": 1661
    },
    {
      "epoch": 0.4950203838492898,
      "grad_norm": 0.7100359201431274,
      "learning_rate": 1.635326556206762e-05,
      "loss": 1.1776,
      "step": 1662
    },
    {
      "epoch": 0.4953182300489585,
      "grad_norm": 0.7099316120147705,
      "learning_rate": 1.6342966859266637e-05,
      "loss": 1.1718,
      "step": 1663
    },
    {
      "epoch": 0.49561607624862714,
      "grad_norm": 0.7423000931739807,
      "learning_rate": 1.6332656887787937e-05,
      "loss": 1.158,
      "step": 1664
    },
    {
      "epoch": 0.49591392244829574,
      "grad_norm": 0.6679733395576477,
      "learning_rate": 1.632233566594783e-05,
      "loss": 1.1563,
      "step": 1665
    },
    {
      "epoch": 0.4962117686479644,
      "grad_norm": 0.7420492172241211,
      "learning_rate": 1.631200321208261e-05,
      "loss": 1.1656,
      "step": 1666
    },
    {
      "epoch": 0.49650961484763306,
      "grad_norm": 0.672215461730957,
      "learning_rate": 1.6301659544548528e-05,
      "loss": 1.1591,
      "step": 1667
    },
    {
      "epoch": 0.4968074610473017,
      "grad_norm": 0.7169597744941711,
      "learning_rate": 1.629130468172176e-05,
      "loss": 1.152,
      "step": 1668
    },
    {
      "epoch": 0.49710530724697033,
      "grad_norm": 0.6964256763458252,
      "learning_rate": 1.6280938641998366e-05,
      "loss": 1.1615,
      "step": 1669
    },
    {
      "epoch": 0.497403153446639,
      "grad_norm": 0.6891978979110718,
      "learning_rate": 1.6270561443794262e-05,
      "loss": 1.1661,
      "step": 1670
    },
    {
      "epoch": 0.49770099964630765,
      "grad_norm": 0.6962587833404541,
      "learning_rate": 1.6260173105545198e-05,
      "loss": 1.1546,
      "step": 1671
    },
    {
      "epoch": 0.49799884584597626,
      "grad_norm": 0.7045391798019409,
      "learning_rate": 1.6249773645706698e-05,
      "loss": 1.1528,
      "step": 1672
    },
    {
      "epoch": 0.4982966920456449,
      "grad_norm": 0.7105042934417725,
      "learning_rate": 1.623936308275406e-05,
      "loss": 1.1761,
      "step": 1673
    },
    {
      "epoch": 0.4985945382453136,
      "grad_norm": 0.6865647435188293,
      "learning_rate": 1.6228941435182294e-05,
      "loss": 1.1665,
      "step": 1674
    },
    {
      "epoch": 0.49889238444498224,
      "grad_norm": 0.7247456908226013,
      "learning_rate": 1.621850872150612e-05,
      "loss": 1.1624,
      "step": 1675
    },
    {
      "epoch": 0.49919023064465085,
      "grad_norm": 0.6827667355537415,
      "learning_rate": 1.6208064960259897e-05,
      "loss": 1.1538,
      "step": 1676
    },
    {
      "epoch": 0.4994880768443195,
      "grad_norm": 0.7096077799797058,
      "learning_rate": 1.6197610169997625e-05,
      "loss": 1.162,
      "step": 1677
    },
    {
      "epoch": 0.49978592304398817,
      "grad_norm": 0.726831316947937,
      "learning_rate": 1.6187144369292894e-05,
      "loss": 1.1354,
      "step": 1678
    },
    {
      "epoch": 0.5000837692436568,
      "grad_norm": 0.7002685070037842,
      "learning_rate": 1.6176667576738852e-05,
      "loss": 1.1528,
      "step": 1679
    },
    {
      "epoch": 0.5003816154433255,
      "grad_norm": 0.7245243191719055,
      "learning_rate": 1.6166179810948182e-05,
      "loss": 1.1582,
      "step": 1680
    },
    {
      "epoch": 0.5006794616429942,
      "grad_norm": 0.753537118434906,
      "learning_rate": 1.615568109055305e-05,
      "loss": 1.1626,
      "step": 1681
    },
    {
      "epoch": 0.5009773078426627,
      "grad_norm": 0.7027132511138916,
      "learning_rate": 1.614517143420509e-05,
      "loss": 1.1502,
      "step": 1682
    },
    {
      "epoch": 0.5012751540423314,
      "grad_norm": 0.7363595962524414,
      "learning_rate": 1.613465086057537e-05,
      "loss": 1.1498,
      "step": 1683
    },
    {
      "epoch": 0.501573000242,
      "grad_norm": 0.6997918486595154,
      "learning_rate": 1.6124119388354343e-05,
      "loss": 1.1603,
      "step": 1684
    },
    {
      "epoch": 0.5018708464416687,
      "grad_norm": 0.716607391834259,
      "learning_rate": 1.611357703625183e-05,
      "loss": 1.1601,
      "step": 1685
    },
    {
      "epoch": 0.5021686926413373,
      "grad_norm": 0.7098117470741272,
      "learning_rate": 1.6103023822996982e-05,
      "loss": 1.1782,
      "step": 1686
    },
    {
      "epoch": 0.502466538841006,
      "grad_norm": 0.6941043734550476,
      "learning_rate": 1.6092459767338236e-05,
      "loss": 1.1656,
      "step": 1687
    },
    {
      "epoch": 0.5027643850406747,
      "grad_norm": 0.6923449039459229,
      "learning_rate": 1.6081884888043307e-05,
      "loss": 1.183,
      "step": 1688
    },
    {
      "epoch": 0.5030622312403432,
      "grad_norm": 0.7082279920578003,
      "learning_rate": 1.6071299203899127e-05,
      "loss": 1.1655,
      "step": 1689
    },
    {
      "epoch": 0.5033600774400119,
      "grad_norm": 0.7139174342155457,
      "learning_rate": 1.6060702733711823e-05,
      "loss": 1.1696,
      "step": 1690
    },
    {
      "epoch": 0.5036579236396805,
      "grad_norm": 0.6981754899024963,
      "learning_rate": 1.605009549630669e-05,
      "loss": 1.172,
      "step": 1691
    },
    {
      "epoch": 0.5039557698393492,
      "grad_norm": 0.69107985496521,
      "learning_rate": 1.6039477510528155e-05,
      "loss": 1.1581,
      "step": 1692
    },
    {
      "epoch": 0.5042536160390179,
      "grad_norm": 0.6882843971252441,
      "learning_rate": 1.6028848795239725e-05,
      "loss": 1.1456,
      "step": 1693
    },
    {
      "epoch": 0.5045514622386865,
      "grad_norm": 0.7075868248939514,
      "learning_rate": 1.6018209369323983e-05,
      "loss": 1.147,
      "step": 1694
    },
    {
      "epoch": 0.5048493084383552,
      "grad_norm": 0.7074382305145264,
      "learning_rate": 1.6007559251682532e-05,
      "loss": 1.1496,
      "step": 1695
    },
    {
      "epoch": 0.5051471546380238,
      "grad_norm": 0.6998595595359802,
      "learning_rate": 1.5996898461235976e-05,
      "loss": 1.1478,
      "step": 1696
    },
    {
      "epoch": 0.5054450008376924,
      "grad_norm": 0.7426585555076599,
      "learning_rate": 1.598622701692388e-05,
      "loss": 1.1746,
      "step": 1697
    },
    {
      "epoch": 0.505742847037361,
      "grad_norm": 0.6913571953773499,
      "learning_rate": 1.5975544937704724e-05,
      "loss": 1.151,
      "step": 1698
    },
    {
      "epoch": 0.5060406932370297,
      "grad_norm": 0.7123813629150391,
      "learning_rate": 1.59648522425559e-05,
      "loss": 1.1639,
      "step": 1699
    },
    {
      "epoch": 0.5063385394366984,
      "grad_norm": 0.7160590887069702,
      "learning_rate": 1.5954148950473642e-05,
      "loss": 1.1517,
      "step": 1700
    },
    {
      "epoch": 0.506636385636367,
      "grad_norm": 0.7349366545677185,
      "learning_rate": 1.5943435080473025e-05,
      "loss": 1.1668,
      "step": 1701
    },
    {
      "epoch": 0.5069342318360357,
      "grad_norm": 0.7529348134994507,
      "learning_rate": 1.5932710651587902e-05,
      "loss": 1.1339,
      "step": 1702
    },
    {
      "epoch": 0.5072320780357044,
      "grad_norm": 0.7347894906997681,
      "learning_rate": 1.59219756828709e-05,
      "loss": 1.166,
      "step": 1703
    },
    {
      "epoch": 0.5075299242353729,
      "grad_norm": 0.7522632479667664,
      "learning_rate": 1.591123019339336e-05,
      "loss": 1.1509,
      "step": 1704
    },
    {
      "epoch": 0.5078277704350416,
      "grad_norm": 0.732912003993988,
      "learning_rate": 1.5900474202245315e-05,
      "loss": 1.146,
      "step": 1705
    },
    {
      "epoch": 0.5081256166347102,
      "grad_norm": 0.7385713458061218,
      "learning_rate": 1.588970772853546e-05,
      "loss": 1.1687,
      "step": 1706
    },
    {
      "epoch": 0.5084234628343789,
      "grad_norm": 0.7328245043754578,
      "learning_rate": 1.5878930791391106e-05,
      "loss": 1.1706,
      "step": 1707
    },
    {
      "epoch": 0.5087213090340476,
      "grad_norm": 0.7300674319267273,
      "learning_rate": 1.5868143409958162e-05,
      "loss": 1.1682,
      "step": 1708
    },
    {
      "epoch": 0.5090191552337162,
      "grad_norm": 0.6966181993484497,
      "learning_rate": 1.5857345603401082e-05,
      "loss": 1.1586,
      "step": 1709
    },
    {
      "epoch": 0.5093170014333849,
      "grad_norm": 0.6993558406829834,
      "learning_rate": 1.5846537390902845e-05,
      "loss": 1.1724,
      "step": 1710
    },
    {
      "epoch": 0.5096148476330535,
      "grad_norm": 0.7228909730911255,
      "learning_rate": 1.583571879166492e-05,
      "loss": 1.1731,
      "step": 1711
    },
    {
      "epoch": 0.5099126938327221,
      "grad_norm": 0.6940150856971741,
      "learning_rate": 1.5824889824907225e-05,
      "loss": 1.1545,
      "step": 1712
    },
    {
      "epoch": 0.5102105400323907,
      "grad_norm": 0.6965219974517822,
      "learning_rate": 1.5814050509868093e-05,
      "loss": 1.1786,
      "step": 1713
    },
    {
      "epoch": 0.5105083862320594,
      "grad_norm": 0.6784839630126953,
      "learning_rate": 1.5803200865804253e-05,
      "loss": 1.1556,
      "step": 1714
    },
    {
      "epoch": 0.5108062324317281,
      "grad_norm": 0.6994674801826477,
      "learning_rate": 1.579234091199077e-05,
      "loss": 1.1682,
      "step": 1715
    },
    {
      "epoch": 0.5111040786313967,
      "grad_norm": 0.6772380471229553,
      "learning_rate": 1.578147066772104e-05,
      "loss": 1.1518,
      "step": 1716
    },
    {
      "epoch": 0.5114019248310654,
      "grad_norm": 0.6807464957237244,
      "learning_rate": 1.5770590152306723e-05,
      "loss": 1.148,
      "step": 1717
    },
    {
      "epoch": 0.511699771030734,
      "grad_norm": 0.6975213885307312,
      "learning_rate": 1.5759699385077744e-05,
      "loss": 1.1718,
      "step": 1718
    },
    {
      "epoch": 0.5119976172304026,
      "grad_norm": 0.6641414761543274,
      "learning_rate": 1.5748798385382226e-05,
      "loss": 1.1661,
      "step": 1719
    },
    {
      "epoch": 0.5122954634300713,
      "grad_norm": 0.6973991990089417,
      "learning_rate": 1.573788717258648e-05,
      "loss": 1.1679,
      "step": 1720
    },
    {
      "epoch": 0.5125933096297399,
      "grad_norm": 0.6800613403320312,
      "learning_rate": 1.5726965766074962e-05,
      "loss": 1.1569,
      "step": 1721
    },
    {
      "epoch": 0.5128911558294086,
      "grad_norm": 0.706687867641449,
      "learning_rate": 1.571603418525023e-05,
      "loss": 1.1601,
      "step": 1722
    },
    {
      "epoch": 0.5131890020290772,
      "grad_norm": 0.6794283986091614,
      "learning_rate": 1.5705092449532922e-05,
      "loss": 1.1351,
      "step": 1723
    },
    {
      "epoch": 0.5134868482287459,
      "grad_norm": 0.7175126671791077,
      "learning_rate": 1.5694140578361722e-05,
      "loss": 1.1513,
      "step": 1724
    },
    {
      "epoch": 0.5137846944284146,
      "grad_norm": 0.7025666236877441,
      "learning_rate": 1.5683178591193306e-05,
      "loss": 1.1499,
      "step": 1725
    },
    {
      "epoch": 0.5140825406280832,
      "grad_norm": 0.6926100850105286,
      "learning_rate": 1.5672206507502337e-05,
      "loss": 1.1213,
      "step": 1726
    },
    {
      "epoch": 0.5143803868277518,
      "grad_norm": 0.7152054309844971,
      "learning_rate": 1.566122434678141e-05,
      "loss": 1.1283,
      "step": 1727
    },
    {
      "epoch": 0.5146782330274204,
      "grad_norm": 0.6804320216178894,
      "learning_rate": 1.565023212854101e-05,
      "loss": 1.1424,
      "step": 1728
    },
    {
      "epoch": 0.5149760792270891,
      "grad_norm": 0.7043558359146118,
      "learning_rate": 1.5639229872309512e-05,
      "loss": 1.1612,
      "step": 1729
    },
    {
      "epoch": 0.5152739254267578,
      "grad_norm": 0.7164785265922546,
      "learning_rate": 1.5628217597633112e-05,
      "loss": 1.1474,
      "step": 1730
    },
    {
      "epoch": 0.5155717716264264,
      "grad_norm": 0.7089158296585083,
      "learning_rate": 1.56171953240758e-05,
      "loss": 1.1712,
      "step": 1731
    },
    {
      "epoch": 0.5158696178260951,
      "grad_norm": 0.6917033195495605,
      "learning_rate": 1.560616307121934e-05,
      "loss": 1.1608,
      "step": 1732
    },
    {
      "epoch": 0.5161674640257637,
      "grad_norm": 0.6996780633926392,
      "learning_rate": 1.5595120858663215e-05,
      "loss": 1.1546,
      "step": 1733
    },
    {
      "epoch": 0.5164653102254323,
      "grad_norm": 0.6914119124412537,
      "learning_rate": 1.5584068706024612e-05,
      "loss": 1.1603,
      "step": 1734
    },
    {
      "epoch": 0.516763156425101,
      "grad_norm": 0.6981446146965027,
      "learning_rate": 1.5573006632938364e-05,
      "loss": 1.148,
      "step": 1735
    },
    {
      "epoch": 0.5170610026247696,
      "grad_norm": 0.7029121518135071,
      "learning_rate": 1.5561934659056947e-05,
      "loss": 1.1688,
      "step": 1736
    },
    {
      "epoch": 0.5173588488244383,
      "grad_norm": 0.719550371170044,
      "learning_rate": 1.5550852804050412e-05,
      "loss": 1.1446,
      "step": 1737
    },
    {
      "epoch": 0.5176566950241069,
      "grad_norm": 0.6796586513519287,
      "learning_rate": 1.5539761087606364e-05,
      "loss": 1.1495,
      "step": 1738
    },
    {
      "epoch": 0.5179545412237756,
      "grad_norm": 0.7405799031257629,
      "learning_rate": 1.5528659529429935e-05,
      "loss": 1.1545,
      "step": 1739
    },
    {
      "epoch": 0.5182523874234443,
      "grad_norm": 0.7103970050811768,
      "learning_rate": 1.551754814924374e-05,
      "loss": 1.1471,
      "step": 1740
    },
    {
      "epoch": 0.5185502336231128,
      "grad_norm": 0.7027060985565186,
      "learning_rate": 1.5506426966787836e-05,
      "loss": 1.1708,
      "step": 1741
    },
    {
      "epoch": 0.5188480798227815,
      "grad_norm": 0.7215583324432373,
      "learning_rate": 1.5495296001819704e-05,
      "loss": 1.1489,
      "step": 1742
    },
    {
      "epoch": 0.5191459260224501,
      "grad_norm": 0.7007563710212708,
      "learning_rate": 1.5484155274114197e-05,
      "loss": 1.132,
      "step": 1743
    },
    {
      "epoch": 0.5194437722221188,
      "grad_norm": 0.6875362396240234,
      "learning_rate": 1.547300480346352e-05,
      "loss": 1.1523,
      "step": 1744
    },
    {
      "epoch": 0.5197416184217875,
      "grad_norm": 0.6781772375106812,
      "learning_rate": 1.546184460967718e-05,
      "loss": 1.1517,
      "step": 1745
    },
    {
      "epoch": 0.5200394646214561,
      "grad_norm": 0.7340065240859985,
      "learning_rate": 1.545067471258196e-05,
      "loss": 1.1579,
      "step": 1746
    },
    {
      "epoch": 0.5203373108211248,
      "grad_norm": 0.7310569286346436,
      "learning_rate": 1.543949513202188e-05,
      "loss": 1.1819,
      "step": 1747
    },
    {
      "epoch": 0.5206351570207934,
      "grad_norm": 0.696808934211731,
      "learning_rate": 1.5428305887858166e-05,
      "loss": 1.1573,
      "step": 1748
    },
    {
      "epoch": 0.520933003220462,
      "grad_norm": 0.6835257411003113,
      "learning_rate": 1.541710699996921e-05,
      "loss": 1.1724,
      "step": 1749
    },
    {
      "epoch": 0.5212308494201306,
      "grad_norm": 0.6591678261756897,
      "learning_rate": 1.540589848825054e-05,
      "loss": 1.1357,
      "step": 1750
    },
    {
      "epoch": 0.5215286956197993,
      "grad_norm": 0.6938487887382507,
      "learning_rate": 1.5394680372614773e-05,
      "loss": 1.1568,
      "step": 1751
    },
    {
      "epoch": 0.521826541819468,
      "grad_norm": 0.6758677959442139,
      "learning_rate": 1.5383452672991598e-05,
      "loss": 1.1504,
      "step": 1752
    },
    {
      "epoch": 0.5221243880191366,
      "grad_norm": 0.6768665313720703,
      "learning_rate": 1.5372215409327726e-05,
      "loss": 1.1565,
      "step": 1753
    },
    {
      "epoch": 0.5224222342188053,
      "grad_norm": 0.7244126200675964,
      "learning_rate": 1.5360968601586858e-05,
      "loss": 1.1594,
      "step": 1754
    },
    {
      "epoch": 0.522720080418474,
      "grad_norm": 0.714651346206665,
      "learning_rate": 1.5349712269749656e-05,
      "loss": 1.1528,
      "step": 1755
    },
    {
      "epoch": 0.5230179266181425,
      "grad_norm": 0.7285259962081909,
      "learning_rate": 1.5338446433813695e-05,
      "loss": 1.148,
      "step": 1756
    },
    {
      "epoch": 0.5233157728178112,
      "grad_norm": 0.6866744756698608,
      "learning_rate": 1.5327171113793437e-05,
      "loss": 1.1402,
      "step": 1757
    },
    {
      "epoch": 0.5236136190174798,
      "grad_norm": 0.7133128046989441,
      "learning_rate": 1.531588632972019e-05,
      "loss": 1.1547,
      "step": 1758
    },
    {
      "epoch": 0.5239114652171485,
      "grad_norm": 0.7262802124023438,
      "learning_rate": 1.5304592101642087e-05,
      "loss": 1.1632,
      "step": 1759
    },
    {
      "epoch": 0.5242093114168171,
      "grad_norm": 0.7285698056221008,
      "learning_rate": 1.5293288449624033e-05,
      "loss": 1.1592,
      "step": 1760
    },
    {
      "epoch": 0.5245071576164858,
      "grad_norm": 0.7387452721595764,
      "learning_rate": 1.528197539374767e-05,
      "loss": 1.1522,
      "step": 1761
    },
    {
      "epoch": 0.5248050038161545,
      "grad_norm": 0.7010663747787476,
      "learning_rate": 1.5270652954111344e-05,
      "loss": 1.1545,
      "step": 1762
    },
    {
      "epoch": 0.5251028500158231,
      "grad_norm": 0.7384028434753418,
      "learning_rate": 1.5259321150830082e-05,
      "loss": 1.1419,
      "step": 1763
    },
    {
      "epoch": 0.5254006962154917,
      "grad_norm": 0.7450021505355835,
      "learning_rate": 1.5247980004035547e-05,
      "loss": 1.1687,
      "step": 1764
    },
    {
      "epoch": 0.5256985424151603,
      "grad_norm": 0.6992811560630798,
      "learning_rate": 1.523662953387599e-05,
      "loss": 1.1444,
      "step": 1765
    },
    {
      "epoch": 0.525996388614829,
      "grad_norm": 0.7160733342170715,
      "learning_rate": 1.5225269760516233e-05,
      "loss": 1.1562,
      "step": 1766
    },
    {
      "epoch": 0.5262942348144977,
      "grad_norm": 0.7006943821907043,
      "learning_rate": 1.5213900704137627e-05,
      "loss": 1.1604,
      "step": 1767
    },
    {
      "epoch": 0.5265920810141663,
      "grad_norm": 0.6929695010185242,
      "learning_rate": 1.5202522384938012e-05,
      "loss": 1.1586,
      "step": 1768
    },
    {
      "epoch": 0.526889927213835,
      "grad_norm": 0.6889545321464539,
      "learning_rate": 1.519113482313168e-05,
      "loss": 1.1687,
      "step": 1769
    },
    {
      "epoch": 0.5271877734135036,
      "grad_norm": 0.7950661182403564,
      "learning_rate": 1.5179738038949354e-05,
      "loss": 1.145,
      "step": 1770
    },
    {
      "epoch": 0.5274856196131722,
      "grad_norm": 0.685127854347229,
      "learning_rate": 1.516833205263813e-05,
      "loss": 1.1598,
      "step": 1771
    },
    {
      "epoch": 0.5277834658128409,
      "grad_norm": 0.7172015309333801,
      "learning_rate": 1.5156916884461457e-05,
      "loss": 1.1573,
      "step": 1772
    },
    {
      "epoch": 0.5280813120125095,
      "grad_norm": 0.7026886940002441,
      "learning_rate": 1.51454925546991e-05,
      "loss": 1.1612,
      "step": 1773
    },
    {
      "epoch": 0.5283791582121782,
      "grad_norm": 0.6845586895942688,
      "learning_rate": 1.5134059083647086e-05,
      "loss": 1.1472,
      "step": 1774
    },
    {
      "epoch": 0.5286770044118468,
      "grad_norm": 0.6596636176109314,
      "learning_rate": 1.5122616491617706e-05,
      "loss": 1.1496,
      "step": 1775
    },
    {
      "epoch": 0.5289748506115155,
      "grad_norm": 0.7205066084861755,
      "learning_rate": 1.5111164798939433e-05,
      "loss": 1.1511,
      "step": 1776
    },
    {
      "epoch": 0.5292726968111842,
      "grad_norm": 0.7409471869468689,
      "learning_rate": 1.5099704025956911e-05,
      "loss": 1.157,
      "step": 1777
    },
    {
      "epoch": 0.5295705430108528,
      "grad_norm": 0.710748016834259,
      "learning_rate": 1.5088234193030933e-05,
      "loss": 1.1736,
      "step": 1778
    },
    {
      "epoch": 0.5298683892105214,
      "grad_norm": 0.7299618124961853,
      "learning_rate": 1.5076755320538368e-05,
      "loss": 1.1444,
      "step": 1779
    },
    {
      "epoch": 0.53016623541019,
      "grad_norm": 0.69326251745224,
      "learning_rate": 1.5065267428872152e-05,
      "loss": 1.154,
      "step": 1780
    },
    {
      "epoch": 0.5304640816098587,
      "grad_norm": 0.70351642370224,
      "learning_rate": 1.5053770538441246e-05,
      "loss": 1.1579,
      "step": 1781
    },
    {
      "epoch": 0.5307619278095274,
      "grad_norm": 0.7217586636543274,
      "learning_rate": 1.504226466967059e-05,
      "loss": 1.1574,
      "step": 1782
    },
    {
      "epoch": 0.531059774009196,
      "grad_norm": 0.6905164122581482,
      "learning_rate": 1.503074984300109e-05,
      "loss": 1.1725,
      "step": 1783
    },
    {
      "epoch": 0.5313576202088647,
      "grad_norm": 0.7088436484336853,
      "learning_rate": 1.5019226078889543e-05,
      "loss": 1.1507,
      "step": 1784
    },
    {
      "epoch": 0.5316554664085333,
      "grad_norm": 0.7002828121185303,
      "learning_rate": 1.5007693397808652e-05,
      "loss": 1.1625,
      "step": 1785
    },
    {
      "epoch": 0.5319533126082019,
      "grad_norm": 0.7343918681144714,
      "learning_rate": 1.4996151820246936e-05,
      "loss": 1.1388,
      "step": 1786
    },
    {
      "epoch": 0.5322511588078706,
      "grad_norm": 0.7142190933227539,
      "learning_rate": 1.4984601366708735e-05,
      "loss": 1.149,
      "step": 1787
    },
    {
      "epoch": 0.5325490050075392,
      "grad_norm": 0.7306362986564636,
      "learning_rate": 1.4973042057714154e-05,
      "loss": 1.154,
      "step": 1788
    },
    {
      "epoch": 0.5328468512072079,
      "grad_norm": 0.6774846315383911,
      "learning_rate": 1.496147391379902e-05,
      "loss": 1.1644,
      "step": 1789
    },
    {
      "epoch": 0.5331446974068765,
      "grad_norm": 0.6809706091880798,
      "learning_rate": 1.4949896955514877e-05,
      "loss": 1.1268,
      "step": 1790
    },
    {
      "epoch": 0.5334425436065452,
      "grad_norm": 0.722968578338623,
      "learning_rate": 1.4938311203428912e-05,
      "loss": 1.1598,
      "step": 1791
    },
    {
      "epoch": 0.5337403898062139,
      "grad_norm": 0.6972602009773254,
      "learning_rate": 1.4926716678123933e-05,
      "loss": 1.1513,
      "step": 1792
    },
    {
      "epoch": 0.5340382360058825,
      "grad_norm": 0.6924199461936951,
      "learning_rate": 1.4915113400198349e-05,
      "loss": 1.1428,
      "step": 1793
    },
    {
      "epoch": 0.5343360822055511,
      "grad_norm": 0.7179045677185059,
      "learning_rate": 1.4903501390266104e-05,
      "loss": 1.1649,
      "step": 1794
    },
    {
      "epoch": 0.5346339284052197,
      "grad_norm": 0.7365652918815613,
      "learning_rate": 1.4891880668956661e-05,
      "loss": 1.151,
      "step": 1795
    },
    {
      "epoch": 0.5349317746048884,
      "grad_norm": 0.690294623374939,
      "learning_rate": 1.4880251256914964e-05,
      "loss": 1.1453,
      "step": 1796
    },
    {
      "epoch": 0.535229620804557,
      "grad_norm": 0.7115709781646729,
      "learning_rate": 1.4868613174801382e-05,
      "loss": 1.1495,
      "step": 1797
    },
    {
      "epoch": 0.5355274670042257,
      "grad_norm": 0.691786527633667,
      "learning_rate": 1.4856966443291705e-05,
      "loss": 1.1435,
      "step": 1798
    },
    {
      "epoch": 0.5358253132038944,
      "grad_norm": 0.7360423803329468,
      "learning_rate": 1.4845311083077076e-05,
      "loss": 1.1576,
      "step": 1799
    },
    {
      "epoch": 0.536123159403563,
      "grad_norm": 0.6791401505470276,
      "learning_rate": 1.4833647114863978e-05,
      "loss": 1.1473,
      "step": 1800
    },
    {
      "epoch": 0.5364210056032316,
      "grad_norm": 0.7760128974914551,
      "learning_rate": 1.4821974559374172e-05,
      "loss": 1.1644,
      "step": 1801
    },
    {
      "epoch": 0.5367188518029002,
      "grad_norm": 0.6775657534599304,
      "learning_rate": 1.481029343734469e-05,
      "loss": 1.131,
      "step": 1802
    },
    {
      "epoch": 0.5370166980025689,
      "grad_norm": 0.6920539736747742,
      "learning_rate": 1.4798603769527775e-05,
      "loss": 1.1522,
      "step": 1803
    },
    {
      "epoch": 0.5373145442022376,
      "grad_norm": 0.686362087726593,
      "learning_rate": 1.4786905576690852e-05,
      "loss": 1.1542,
      "step": 1804
    },
    {
      "epoch": 0.5376123904019062,
      "grad_norm": 0.6946977972984314,
      "learning_rate": 1.4775198879616495e-05,
      "loss": 1.147,
      "step": 1805
    },
    {
      "epoch": 0.5379102366015749,
      "grad_norm": 0.6781960725784302,
      "learning_rate": 1.476348369910238e-05,
      "loss": 1.1447,
      "step": 1806
    },
    {
      "epoch": 0.5382080828012435,
      "grad_norm": 0.6997618079185486,
      "learning_rate": 1.4751760055961267e-05,
      "loss": 1.1554,
      "step": 1807
    },
    {
      "epoch": 0.5385059290009122,
      "grad_norm": 0.6656970977783203,
      "learning_rate": 1.4740027971020936e-05,
      "loss": 1.1629,
      "step": 1808
    },
    {
      "epoch": 0.5388037752005808,
      "grad_norm": 0.7032130360603333,
      "learning_rate": 1.4728287465124169e-05,
      "loss": 1.1612,
      "step": 1809
    },
    {
      "epoch": 0.5391016214002494,
      "grad_norm": 0.6848834753036499,
      "learning_rate": 1.4716538559128715e-05,
      "loss": 1.1361,
      "step": 1810
    },
    {
      "epoch": 0.5393994675999181,
      "grad_norm": 0.7095216512680054,
      "learning_rate": 1.4704781273907234e-05,
      "loss": 1.156,
      "step": 1811
    },
    {
      "epoch": 0.5396973137995867,
      "grad_norm": 0.6689624786376953,
      "learning_rate": 1.4693015630347287e-05,
      "loss": 1.1476,
      "step": 1812
    },
    {
      "epoch": 0.5399951599992554,
      "grad_norm": 0.6961745619773865,
      "learning_rate": 1.4681241649351275e-05,
      "loss": 1.1641,
      "step": 1813
    },
    {
      "epoch": 0.5402930061989241,
      "grad_norm": 0.70126873254776,
      "learning_rate": 1.4669459351836404e-05,
      "loss": 1.1828,
      "step": 1814
    },
    {
      "epoch": 0.5405908523985927,
      "grad_norm": 0.6869078874588013,
      "learning_rate": 1.4657668758734673e-05,
      "loss": 1.1546,
      "step": 1815
    },
    {
      "epoch": 0.5408886985982613,
      "grad_norm": 0.6669703125953674,
      "learning_rate": 1.4645869890992804e-05,
      "loss": 1.1292,
      "step": 1816
    },
    {
      "epoch": 0.5411865447979299,
      "grad_norm": 0.6907333135604858,
      "learning_rate": 1.4634062769572225e-05,
      "loss": 1.159,
      "step": 1817
    },
    {
      "epoch": 0.5414843909975986,
      "grad_norm": 0.6744605302810669,
      "learning_rate": 1.4622247415449028e-05,
      "loss": 1.157,
      "step": 1818
    },
    {
      "epoch": 0.5417822371972673,
      "grad_norm": 0.7096845507621765,
      "learning_rate": 1.4610423849613926e-05,
      "loss": 1.1569,
      "step": 1819
    },
    {
      "epoch": 0.5420800833969359,
      "grad_norm": 0.673180878162384,
      "learning_rate": 1.4598592093072228e-05,
      "loss": 1.1529,
      "step": 1820
    },
    {
      "epoch": 0.5423779295966046,
      "grad_norm": 0.7228463292121887,
      "learning_rate": 1.4586752166843786e-05,
      "loss": 1.1704,
      "step": 1821
    },
    {
      "epoch": 0.5426757757962732,
      "grad_norm": 0.6856001019477844,
      "learning_rate": 1.4574904091962973e-05,
      "loss": 1.1477,
      "step": 1822
    },
    {
      "epoch": 0.5429736219959418,
      "grad_norm": 0.6953288316726685,
      "learning_rate": 1.4563047889478633e-05,
      "loss": 1.1511,
      "step": 1823
    },
    {
      "epoch": 0.5432714681956105,
      "grad_norm": 0.7183958292007446,
      "learning_rate": 1.4551183580454056e-05,
      "loss": 1.1588,
      "step": 1824
    },
    {
      "epoch": 0.5435693143952791,
      "grad_norm": 0.7849579453468323,
      "learning_rate": 1.4539311185966929e-05,
      "loss": 1.1396,
      "step": 1825
    },
    {
      "epoch": 0.5438671605949478,
      "grad_norm": 0.7594764232635498,
      "learning_rate": 1.4527430727109299e-05,
      "loss": 1.1721,
      "step": 1826
    },
    {
      "epoch": 0.5441650067946164,
      "grad_norm": 0.7227590084075928,
      "learning_rate": 1.4515542224987552e-05,
      "loss": 1.1467,
      "step": 1827
    },
    {
      "epoch": 0.5444628529942851,
      "grad_norm": 0.6811076402664185,
      "learning_rate": 1.450364570072235e-05,
      "loss": 1.1536,
      "step": 1828
    },
    {
      "epoch": 0.5447606991939538,
      "grad_norm": 0.6939821243286133,
      "learning_rate": 1.4491741175448617e-05,
      "loss": 1.1342,
      "step": 1829
    },
    {
      "epoch": 0.5450585453936224,
      "grad_norm": 0.7158921360969543,
      "learning_rate": 1.4479828670315486e-05,
      "loss": 1.1414,
      "step": 1830
    },
    {
      "epoch": 0.545356391593291,
      "grad_norm": 0.7603862881660461,
      "learning_rate": 1.446790820648627e-05,
      "loss": 1.1577,
      "step": 1831
    },
    {
      "epoch": 0.5456542377929596,
      "grad_norm": 0.7004614472389221,
      "learning_rate": 1.4455979805138416e-05,
      "loss": 1.1533,
      "step": 1832
    },
    {
      "epoch": 0.5459520839926283,
      "grad_norm": 0.7094352841377258,
      "learning_rate": 1.4444043487463479e-05,
      "loss": 1.1541,
      "step": 1833
    },
    {
      "epoch": 0.546249930192297,
      "grad_norm": 0.7315611839294434,
      "learning_rate": 1.4432099274667073e-05,
      "loss": 1.1347,
      "step": 1834
    },
    {
      "epoch": 0.5465477763919656,
      "grad_norm": 0.7053688764572144,
      "learning_rate": 1.442014718796884e-05,
      "loss": 1.1745,
      "step": 1835
    },
    {
      "epoch": 0.5468456225916343,
      "grad_norm": 0.7605271339416504,
      "learning_rate": 1.440818724860241e-05,
      "loss": 1.1583,
      "step": 1836
    },
    {
      "epoch": 0.5471434687913029,
      "grad_norm": 0.7082123756408691,
      "learning_rate": 1.4396219477815363e-05,
      "loss": 1.1514,
      "step": 1837
    },
    {
      "epoch": 0.5474413149909715,
      "grad_norm": 0.6848958730697632,
      "learning_rate": 1.4384243896869194e-05,
      "loss": 1.1415,
      "step": 1838
    },
    {
      "epoch": 0.5477391611906401,
      "grad_norm": 0.6896035671234131,
      "learning_rate": 1.4372260527039275e-05,
      "loss": 1.1618,
      "step": 1839
    },
    {
      "epoch": 0.5480370073903088,
      "grad_norm": 0.6861293911933899,
      "learning_rate": 1.4360269389614809e-05,
      "loss": 1.1433,
      "step": 1840
    },
    {
      "epoch": 0.5483348535899775,
      "grad_norm": 0.6944071054458618,
      "learning_rate": 1.4348270505898802e-05,
      "loss": 1.1569,
      "step": 1841
    },
    {
      "epoch": 0.5486326997896461,
      "grad_norm": 0.6917097568511963,
      "learning_rate": 1.4336263897208025e-05,
      "loss": 1.148,
      "step": 1842
    },
    {
      "epoch": 0.5489305459893148,
      "grad_norm": 0.6852682828903198,
      "learning_rate": 1.4324249584872969e-05,
      "loss": 1.1565,
      "step": 1843
    },
    {
      "epoch": 0.5492283921889834,
      "grad_norm": 0.6747717261314392,
      "learning_rate": 1.431222759023781e-05,
      "loss": 1.1569,
      "step": 1844
    },
    {
      "epoch": 0.5495262383886521,
      "grad_norm": 0.701622486114502,
      "learning_rate": 1.4300197934660378e-05,
      "loss": 1.1487,
      "step": 1845
    },
    {
      "epoch": 0.5498240845883207,
      "grad_norm": 0.7015729546546936,
      "learning_rate": 1.4288160639512105e-05,
      "loss": 1.1523,
      "step": 1846
    },
    {
      "epoch": 0.5501219307879893,
      "grad_norm": 0.6769354939460754,
      "learning_rate": 1.4276115726178003e-05,
      "loss": 1.1541,
      "step": 1847
    },
    {
      "epoch": 0.550419776987658,
      "grad_norm": 0.7126910090446472,
      "learning_rate": 1.4264063216056609e-05,
      "loss": 1.1629,
      "step": 1848
    },
    {
      "epoch": 0.5507176231873266,
      "grad_norm": 0.6819556951522827,
      "learning_rate": 1.4252003130559971e-05,
      "loss": 1.1468,
      "step": 1849
    },
    {
      "epoch": 0.5510154693869953,
      "grad_norm": 0.6864358186721802,
      "learning_rate": 1.4239935491113579e-05,
      "loss": 1.1517,
      "step": 1850
    },
    {
      "epoch": 0.551313315586664,
      "grad_norm": 0.6681830286979675,
      "learning_rate": 1.4227860319156353e-05,
      "loss": 1.1484,
      "step": 1851
    },
    {
      "epoch": 0.5516111617863326,
      "grad_norm": 0.6848908066749573,
      "learning_rate": 1.4215777636140589e-05,
      "loss": 1.1432,
      "step": 1852
    },
    {
      "epoch": 0.5519090079860012,
      "grad_norm": 0.6935808658599854,
      "learning_rate": 1.420368746353193e-05,
      "loss": 1.1516,
      "step": 1853
    },
    {
      "epoch": 0.5522068541856698,
      "grad_norm": 0.7087704539299011,
      "learning_rate": 1.4191589822809326e-05,
      "loss": 1.1525,
      "step": 1854
    },
    {
      "epoch": 0.5525047003853385,
      "grad_norm": 0.7263489961624146,
      "learning_rate": 1.4179484735464995e-05,
      "loss": 1.1322,
      "step": 1855
    },
    {
      "epoch": 0.5528025465850072,
      "grad_norm": 0.6936025023460388,
      "learning_rate": 1.4167372223004379e-05,
      "loss": 1.1471,
      "step": 1856
    },
    {
      "epoch": 0.5531003927846758,
      "grad_norm": 0.6898125410079956,
      "learning_rate": 1.4155252306946117e-05,
      "loss": 1.1403,
      "step": 1857
    },
    {
      "epoch": 0.5533982389843445,
      "grad_norm": 0.7064741253852844,
      "learning_rate": 1.4143125008821996e-05,
      "loss": 1.1565,
      "step": 1858
    },
    {
      "epoch": 0.5536960851840131,
      "grad_norm": 0.7068637609481812,
      "learning_rate": 1.4130990350176924e-05,
      "loss": 1.1475,
      "step": 1859
    },
    {
      "epoch": 0.5539939313836818,
      "grad_norm": 0.6765069961547852,
      "learning_rate": 1.411884835256888e-05,
      "loss": 1.1497,
      "step": 1860
    },
    {
      "epoch": 0.5542917775833504,
      "grad_norm": 0.6807036399841309,
      "learning_rate": 1.4106699037568878e-05,
      "loss": 1.1413,
      "step": 1861
    },
    {
      "epoch": 0.554589623783019,
      "grad_norm": 0.7522012591362,
      "learning_rate": 1.4094542426760947e-05,
      "loss": 1.1681,
      "step": 1862
    },
    {
      "epoch": 0.5548874699826877,
      "grad_norm": 0.7028205394744873,
      "learning_rate": 1.4082378541742057e-05,
      "loss": 1.1659,
      "step": 1863
    },
    {
      "epoch": 0.5551853161823563,
      "grad_norm": 0.6792667508125305,
      "learning_rate": 1.4070207404122118e-05,
      "loss": 1.1466,
      "step": 1864
    },
    {
      "epoch": 0.555483162382025,
      "grad_norm": 0.6783607006072998,
      "learning_rate": 1.4058029035523917e-05,
      "loss": 1.1497,
      "step": 1865
    },
    {
      "epoch": 0.5557810085816937,
      "grad_norm": 0.6885132193565369,
      "learning_rate": 1.4045843457583086e-05,
      "loss": 1.1658,
      "step": 1866
    },
    {
      "epoch": 0.5560788547813623,
      "grad_norm": 0.7093797922134399,
      "learning_rate": 1.4033650691948068e-05,
      "loss": 1.1536,
      "step": 1867
    },
    {
      "epoch": 0.5563767009810309,
      "grad_norm": 0.6795626282691956,
      "learning_rate": 1.4021450760280076e-05,
      "loss": 1.1448,
      "step": 1868
    },
    {
      "epoch": 0.5566745471806995,
      "grad_norm": 0.7001522779464722,
      "learning_rate": 1.400924368425305e-05,
      "loss": 1.1613,
      "step": 1869
    },
    {
      "epoch": 0.5569723933803682,
      "grad_norm": 0.7052515149116516,
      "learning_rate": 1.3997029485553622e-05,
      "loss": 1.1385,
      "step": 1870
    },
    {
      "epoch": 0.5572702395800369,
      "grad_norm": 0.7053892016410828,
      "learning_rate": 1.3984808185881086e-05,
      "loss": 1.1617,
      "step": 1871
    },
    {
      "epoch": 0.5575680857797055,
      "grad_norm": 0.6819269061088562,
      "learning_rate": 1.3972579806947347e-05,
      "loss": 1.1602,
      "step": 1872
    },
    {
      "epoch": 0.5578659319793742,
      "grad_norm": 0.7349790334701538,
      "learning_rate": 1.396034437047688e-05,
      "loss": 1.1327,
      "step": 1873
    },
    {
      "epoch": 0.5581637781790428,
      "grad_norm": 0.6757753491401672,
      "learning_rate": 1.394810189820671e-05,
      "loss": 1.1342,
      "step": 1874
    },
    {
      "epoch": 0.5584616243787115,
      "grad_norm": 0.7225736379623413,
      "learning_rate": 1.3935852411886347e-05,
      "loss": 1.1642,
      "step": 1875
    },
    {
      "epoch": 0.55875947057838,
      "grad_norm": 0.6793896555900574,
      "learning_rate": 1.392359593327778e-05,
      "loss": 1.1528,
      "step": 1876
    },
    {
      "epoch": 0.5590573167780487,
      "grad_norm": 0.6859183311462402,
      "learning_rate": 1.3911332484155406e-05,
      "loss": 1.136,
      "step": 1877
    },
    {
      "epoch": 0.5593551629777174,
      "grad_norm": 0.687907874584198,
      "learning_rate": 1.3899062086306008e-05,
      "loss": 1.1432,
      "step": 1878
    },
    {
      "epoch": 0.559653009177386,
      "grad_norm": 0.690032422542572,
      "learning_rate": 1.388678476152872e-05,
      "loss": 1.1548,
      "step": 1879
    },
    {
      "epoch": 0.5599508553770547,
      "grad_norm": 0.72669517993927,
      "learning_rate": 1.3874500531634975e-05,
      "loss": 1.1598,
      "step": 1880
    },
    {
      "epoch": 0.5602487015767234,
      "grad_norm": 0.7053661942481995,
      "learning_rate": 1.3862209418448476e-05,
      "loss": 1.1616,
      "step": 1881
    },
    {
      "epoch": 0.560546547776392,
      "grad_norm": 0.708357572555542,
      "learning_rate": 1.384991144380516e-05,
      "loss": 1.1424,
      "step": 1882
    },
    {
      "epoch": 0.5608443939760606,
      "grad_norm": 0.6785634756088257,
      "learning_rate": 1.3837606629553143e-05,
      "loss": 1.1429,
      "step": 1883
    },
    {
      "epoch": 0.5611422401757292,
      "grad_norm": 0.7114861607551575,
      "learning_rate": 1.3825294997552703e-05,
      "loss": 1.1539,
      "step": 1884
    },
    {
      "epoch": 0.5614400863753979,
      "grad_norm": 0.7003664374351501,
      "learning_rate": 1.381297656967622e-05,
      "loss": 1.1597,
      "step": 1885
    },
    {
      "epoch": 0.5617379325750665,
      "grad_norm": 0.724639356136322,
      "learning_rate": 1.3800651367808158e-05,
      "loss": 1.1637,
      "step": 1886
    },
    {
      "epoch": 0.5620357787747352,
      "grad_norm": 0.7091076970100403,
      "learning_rate": 1.3788319413845004e-05,
      "loss": 1.1444,
      "step": 1887
    },
    {
      "epoch": 0.5623336249744039,
      "grad_norm": 0.673536479473114,
      "learning_rate": 1.3775980729695247e-05,
      "loss": 1.127,
      "step": 1888
    },
    {
      "epoch": 0.5626314711740725,
      "grad_norm": 0.7382071018218994,
      "learning_rate": 1.3763635337279335e-05,
      "loss": 1.1537,
      "step": 1889
    },
    {
      "epoch": 0.5629293173737412,
      "grad_norm": 0.6829398274421692,
      "learning_rate": 1.375128325852963e-05,
      "loss": 1.1327,
      "step": 1890
    },
    {
      "epoch": 0.5632271635734097,
      "grad_norm": 0.686085045337677,
      "learning_rate": 1.373892451539037e-05,
      "loss": 1.1527,
      "step": 1891
    },
    {
      "epoch": 0.5635250097730784,
      "grad_norm": 0.6910931468009949,
      "learning_rate": 1.3726559129817641e-05,
      "loss": 1.1565,
      "step": 1892
    },
    {
      "epoch": 0.5638228559727471,
      "grad_norm": 0.7061626315116882,
      "learning_rate": 1.3714187123779316e-05,
      "loss": 1.1554,
      "step": 1893
    },
    {
      "epoch": 0.5641207021724157,
      "grad_norm": 0.7075145840644836,
      "learning_rate": 1.3701808519255051e-05,
      "loss": 1.1464,
      "step": 1894
    },
    {
      "epoch": 0.5644185483720844,
      "grad_norm": 0.6991412043571472,
      "learning_rate": 1.3689423338236201e-05,
      "loss": 1.154,
      "step": 1895
    },
    {
      "epoch": 0.564716394571753,
      "grad_norm": 0.6882572174072266,
      "learning_rate": 1.367703160272582e-05,
      "loss": 1.1468,
      "step": 1896
    },
    {
      "epoch": 0.5650142407714217,
      "grad_norm": 0.6746490597724915,
      "learning_rate": 1.3664633334738601e-05,
      "loss": 1.1353,
      "step": 1897
    },
    {
      "epoch": 0.5653120869710903,
      "grad_norm": 0.687214732170105,
      "learning_rate": 1.3652228556300844e-05,
      "loss": 1.1413,
      "step": 1898
    },
    {
      "epoch": 0.5656099331707589,
      "grad_norm": 0.7113034725189209,
      "learning_rate": 1.3639817289450412e-05,
      "loss": 1.1494,
      "step": 1899
    },
    {
      "epoch": 0.5659077793704276,
      "grad_norm": 0.7852702736854553,
      "learning_rate": 1.36273995562367e-05,
      "loss": 1.1462,
      "step": 1900
    },
    {
      "epoch": 0.5662056255700962,
      "grad_norm": 0.7082939743995667,
      "learning_rate": 1.3614975378720592e-05,
      "loss": 1.1518,
      "step": 1901
    },
    {
      "epoch": 0.5665034717697649,
      "grad_norm": 0.7555058002471924,
      "learning_rate": 1.3602544778974407e-05,
      "loss": 1.1362,
      "step": 1902
    },
    {
      "epoch": 0.5668013179694336,
      "grad_norm": 0.693461000919342,
      "learning_rate": 1.3590107779081894e-05,
      "loss": 1.1501,
      "step": 1903
    },
    {
      "epoch": 0.5670991641691022,
      "grad_norm": 0.766847550868988,
      "learning_rate": 1.3577664401138158e-05,
      "loss": 1.1563,
      "step": 1904
    },
    {
      "epoch": 0.5673970103687708,
      "grad_norm": 0.7231875658035278,
      "learning_rate": 1.3565214667249633e-05,
      "loss": 1.1578,
      "step": 1905
    },
    {
      "epoch": 0.5676948565684394,
      "grad_norm": 0.6955324411392212,
      "learning_rate": 1.3552758599534062e-05,
      "loss": 1.1613,
      "step": 1906
    },
    {
      "epoch": 0.5679927027681081,
      "grad_norm": 0.7389485239982605,
      "learning_rate": 1.354029622012042e-05,
      "loss": 1.1503,
      "step": 1907
    },
    {
      "epoch": 0.5682905489677768,
      "grad_norm": 0.7032195925712585,
      "learning_rate": 1.3527827551148906e-05,
      "loss": 1.15,
      "step": 1908
    },
    {
      "epoch": 0.5685883951674454,
      "grad_norm": 0.7320900559425354,
      "learning_rate": 1.3515352614770895e-05,
      "loss": 1.1702,
      "step": 1909
    },
    {
      "epoch": 0.5688862413671141,
      "grad_norm": 0.6840920448303223,
      "learning_rate": 1.350287143314889e-05,
      "loss": 1.157,
      "step": 1910
    },
    {
      "epoch": 0.5691840875667827,
      "grad_norm": 0.699867844581604,
      "learning_rate": 1.3490384028456492e-05,
      "loss": 1.1445,
      "step": 1911
    },
    {
      "epoch": 0.5694819337664514,
      "grad_norm": 0.6893196105957031,
      "learning_rate": 1.3477890422878357e-05,
      "loss": 1.1443,
      "step": 1912
    },
    {
      "epoch": 0.56977977996612,
      "grad_norm": 0.6987179517745972,
      "learning_rate": 1.3465390638610154e-05,
      "loss": 1.1429,
      "step": 1913
    },
    {
      "epoch": 0.5700776261657886,
      "grad_norm": 0.6830858588218689,
      "learning_rate": 1.3452884697858538e-05,
      "loss": 1.1392,
      "step": 1914
    },
    {
      "epoch": 0.5703754723654573,
      "grad_norm": 0.6821770668029785,
      "learning_rate": 1.3440372622841091e-05,
      "loss": 1.1496,
      "step": 1915
    },
    {
      "epoch": 0.5706733185651259,
      "grad_norm": 0.7186193466186523,
      "learning_rate": 1.3427854435786303e-05,
      "loss": 1.1538,
      "step": 1916
    },
    {
      "epoch": 0.5709711647647946,
      "grad_norm": 0.7171904444694519,
      "learning_rate": 1.341533015893351e-05,
      "loss": 1.1412,
      "step": 1917
    },
    {
      "epoch": 0.5712690109644633,
      "grad_norm": 0.6976035237312317,
      "learning_rate": 1.3402799814532881e-05,
      "loss": 1.1495,
      "step": 1918
    },
    {
      "epoch": 0.5715668571641319,
      "grad_norm": 0.6590944528579712,
      "learning_rate": 1.3390263424845353e-05,
      "loss": 1.1479,
      "step": 1919
    },
    {
      "epoch": 0.5718647033638005,
      "grad_norm": 0.706636905670166,
      "learning_rate": 1.3377721012142603e-05,
      "loss": 1.1337,
      "step": 1920
    },
    {
      "epoch": 0.5721625495634691,
      "grad_norm": 0.7061787843704224,
      "learning_rate": 1.3365172598707023e-05,
      "loss": 1.1659,
      "step": 1921
    },
    {
      "epoch": 0.5724603957631378,
      "grad_norm": 0.6637413501739502,
      "learning_rate": 1.3352618206831647e-05,
      "loss": 1.1582,
      "step": 1922
    },
    {
      "epoch": 0.5727582419628064,
      "grad_norm": 0.7116951942443848,
      "learning_rate": 1.3340057858820141e-05,
      "loss": 1.1556,
      "step": 1923
    },
    {
      "epoch": 0.5730560881624751,
      "grad_norm": 0.6938287615776062,
      "learning_rate": 1.3327491576986749e-05,
      "loss": 1.1504,
      "step": 1924
    },
    {
      "epoch": 0.5733539343621438,
      "grad_norm": 0.6729810833930969,
      "learning_rate": 1.3314919383656256e-05,
      "loss": 1.1458,
      "step": 1925
    },
    {
      "epoch": 0.5736517805618124,
      "grad_norm": 0.6771924495697021,
      "learning_rate": 1.3302341301163952e-05,
      "loss": 1.1397,
      "step": 1926
    },
    {
      "epoch": 0.5739496267614811,
      "grad_norm": 0.7212462425231934,
      "learning_rate": 1.3289757351855587e-05,
      "loss": 1.1492,
      "step": 1927
    },
    {
      "epoch": 0.5742474729611496,
      "grad_norm": 0.7020722031593323,
      "learning_rate": 1.3277167558087338e-05,
      "loss": 1.1605,
      "step": 1928
    },
    {
      "epoch": 0.5745453191608183,
      "grad_norm": 0.6964832544326782,
      "learning_rate": 1.326457194222576e-05,
      "loss": 1.1478,
      "step": 1929
    },
    {
      "epoch": 0.574843165360487,
      "grad_norm": 0.6605055332183838,
      "learning_rate": 1.325197052664775e-05,
      "loss": 1.1368,
      "step": 1930
    },
    {
      "epoch": 0.5751410115601556,
      "grad_norm": 0.7050206661224365,
      "learning_rate": 1.3239363333740517e-05,
      "loss": 1.1661,
      "step": 1931
    },
    {
      "epoch": 0.5754388577598243,
      "grad_norm": 0.6759475469589233,
      "learning_rate": 1.3226750385901521e-05,
      "loss": 1.1475,
      "step": 1932
    },
    {
      "epoch": 0.575736703959493,
      "grad_norm": 0.671838641166687,
      "learning_rate": 1.3214131705538463e-05,
      "loss": 1.1604,
      "step": 1933
    },
    {
      "epoch": 0.5760345501591616,
      "grad_norm": 0.6771221160888672,
      "learning_rate": 1.320150731506921e-05,
      "loss": 1.1449,
      "step": 1934
    },
    {
      "epoch": 0.5763323963588302,
      "grad_norm": 0.6752444505691528,
      "learning_rate": 1.3188877236921789e-05,
      "loss": 1.1379,
      "step": 1935
    },
    {
      "epoch": 0.5766302425584988,
      "grad_norm": 0.6848403215408325,
      "learning_rate": 1.3176241493534321e-05,
      "loss": 1.1348,
      "step": 1936
    },
    {
      "epoch": 0.5769280887581675,
      "grad_norm": 0.6968139410018921,
      "learning_rate": 1.3163600107354992e-05,
      "loss": 1.1501,
      "step": 1937
    },
    {
      "epoch": 0.5772259349578361,
      "grad_norm": 0.6898945569992065,
      "learning_rate": 1.3150953100842023e-05,
      "loss": 1.1539,
      "step": 1938
    },
    {
      "epoch": 0.5775237811575048,
      "grad_norm": 0.680107057094574,
      "learning_rate": 1.3138300496463606e-05,
      "loss": 1.1431,
      "step": 1939
    },
    {
      "epoch": 0.5778216273571735,
      "grad_norm": 0.6442410349845886,
      "learning_rate": 1.3125642316697889e-05,
      "loss": 1.1175,
      "step": 1940
    },
    {
      "epoch": 0.5781194735568421,
      "grad_norm": 0.7069491147994995,
      "learning_rate": 1.3112978584032923e-05,
      "loss": 1.1399,
      "step": 1941
    },
    {
      "epoch": 0.5784173197565108,
      "grad_norm": 0.6884000301361084,
      "learning_rate": 1.3100309320966613e-05,
      "loss": 1.1409,
      "step": 1942
    },
    {
      "epoch": 0.5787151659561793,
      "grad_norm": 0.7137901782989502,
      "learning_rate": 1.3087634550006707e-05,
      "loss": 1.1604,
      "step": 1943
    },
    {
      "epoch": 0.579013012155848,
      "grad_norm": 0.697763204574585,
      "learning_rate": 1.3074954293670726e-05,
      "loss": 1.1597,
      "step": 1944
    },
    {
      "epoch": 0.5793108583555167,
      "grad_norm": 0.7183982729911804,
      "learning_rate": 1.3062268574485937e-05,
      "loss": 1.1542,
      "step": 1945
    },
    {
      "epoch": 0.5796087045551853,
      "grad_norm": 0.6606554985046387,
      "learning_rate": 1.3049577414989318e-05,
      "loss": 1.1488,
      "step": 1946
    },
    {
      "epoch": 0.579906550754854,
      "grad_norm": 0.6886841654777527,
      "learning_rate": 1.3036880837727503e-05,
      "loss": 1.1404,
      "step": 1947
    },
    {
      "epoch": 0.5802043969545226,
      "grad_norm": 0.6678399443626404,
      "learning_rate": 1.3024178865256764e-05,
      "loss": 1.1416,
      "step": 1948
    },
    {
      "epoch": 0.5805022431541913,
      "grad_norm": 0.7201886177062988,
      "learning_rate": 1.3011471520142946e-05,
      "loss": 1.1615,
      "step": 1949
    },
    {
      "epoch": 0.5808000893538598,
      "grad_norm": 0.7163878083229065,
      "learning_rate": 1.2998758824961442e-05,
      "loss": 1.1335,
      "step": 1950
    },
    {
      "epoch": 0.5810979355535285,
      "grad_norm": 0.6653960347175598,
      "learning_rate": 1.2986040802297153e-05,
      "loss": 1.1421,
      "step": 1951
    },
    {
      "epoch": 0.5813957817531972,
      "grad_norm": 0.6978620290756226,
      "learning_rate": 1.297331747474444e-05,
      "loss": 1.1407,
      "step": 1952
    },
    {
      "epoch": 0.5816936279528658,
      "grad_norm": 0.6879580616950989,
      "learning_rate": 1.2960588864907096e-05,
      "loss": 1.1406,
      "step": 1953
    },
    {
      "epoch": 0.5819914741525345,
      "grad_norm": 0.6925548315048218,
      "learning_rate": 1.2947854995398286e-05,
      "loss": 1.1473,
      "step": 1954
    },
    {
      "epoch": 0.5822893203522032,
      "grad_norm": 0.6857414245605469,
      "learning_rate": 1.2935115888840532e-05,
      "loss": 1.1418,
      "step": 1955
    },
    {
      "epoch": 0.5825871665518718,
      "grad_norm": 0.6739926338195801,
      "learning_rate": 1.2922371567865652e-05,
      "loss": 1.1523,
      "step": 1956
    },
    {
      "epoch": 0.5828850127515405,
      "grad_norm": 0.6618935465812683,
      "learning_rate": 1.290962205511473e-05,
      "loss": 1.1344,
      "step": 1957
    },
    {
      "epoch": 0.583182858951209,
      "grad_norm": 0.7284050583839417,
      "learning_rate": 1.2896867373238075e-05,
      "loss": 1.1553,
      "step": 1958
    },
    {
      "epoch": 0.5834807051508777,
      "grad_norm": 0.7016205191612244,
      "learning_rate": 1.2884107544895175e-05,
      "loss": 1.1563,
      "step": 1959
    },
    {
      "epoch": 0.5837785513505463,
      "grad_norm": 0.6993693113327026,
      "learning_rate": 1.287134259275467e-05,
      "loss": 1.1382,
      "step": 1960
    },
    {
      "epoch": 0.584076397550215,
      "grad_norm": 0.676010251045227,
      "learning_rate": 1.2858572539494289e-05,
      "loss": 1.1378,
      "step": 1961
    },
    {
      "epoch": 0.5843742437498837,
      "grad_norm": 0.6895237565040588,
      "learning_rate": 1.2845797407800834e-05,
      "loss": 1.1443,
      "step": 1962
    },
    {
      "epoch": 0.5846720899495523,
      "grad_norm": 0.7140636444091797,
      "learning_rate": 1.283301722037013e-05,
      "loss": 1.1362,
      "step": 1963
    },
    {
      "epoch": 0.584969936149221,
      "grad_norm": 0.6820301413536072,
      "learning_rate": 1.2820231999906974e-05,
      "loss": 1.1405,
      "step": 1964
    },
    {
      "epoch": 0.5852677823488895,
      "grad_norm": 0.6772931814193726,
      "learning_rate": 1.2807441769125116e-05,
      "loss": 1.136,
      "step": 1965
    },
    {
      "epoch": 0.5855656285485582,
      "grad_norm": 0.656184196472168,
      "learning_rate": 1.2794646550747196e-05,
      "loss": 1.1366,
      "step": 1966
    },
    {
      "epoch": 0.5858634747482269,
      "grad_norm": 0.7068488597869873,
      "learning_rate": 1.2781846367504725e-05,
      "loss": 1.1502,
      "step": 1967
    },
    {
      "epoch": 0.5861613209478955,
      "grad_norm": 0.7021655440330505,
      "learning_rate": 1.2769041242138027e-05,
      "loss": 1.1467,
      "step": 1968
    },
    {
      "epoch": 0.5864591671475642,
      "grad_norm": 0.6819470524787903,
      "learning_rate": 1.2756231197396214e-05,
      "loss": 1.146,
      "step": 1969
    },
    {
      "epoch": 0.5867570133472328,
      "grad_norm": 0.7169177532196045,
      "learning_rate": 1.2743416256037126e-05,
      "loss": 1.1605,
      "step": 1970
    },
    {
      "epoch": 0.5870548595469015,
      "grad_norm": 0.7096365690231323,
      "learning_rate": 1.2730596440827312e-05,
      "loss": 1.1413,
      "step": 1971
    },
    {
      "epoch": 0.5873527057465702,
      "grad_norm": 0.6927030682563782,
      "learning_rate": 1.2717771774541973e-05,
      "loss": 1.1548,
      "step": 1972
    },
    {
      "epoch": 0.5876505519462387,
      "grad_norm": 0.6682929396629333,
      "learning_rate": 1.2704942279964936e-05,
      "loss": 1.1461,
      "step": 1973
    },
    {
      "epoch": 0.5879483981459074,
      "grad_norm": 0.6722991466522217,
      "learning_rate": 1.2692107979888599e-05,
      "loss": 1.1464,
      "step": 1974
    },
    {
      "epoch": 0.588246244345576,
      "grad_norm": 0.7016704082489014,
      "learning_rate": 1.26792688971139e-05,
      "loss": 1.1398,
      "step": 1975
    },
    {
      "epoch": 0.5885440905452447,
      "grad_norm": 0.6672192215919495,
      "learning_rate": 1.2666425054450276e-05,
      "loss": 1.129,
      "step": 1976
    },
    {
      "epoch": 0.5888419367449134,
      "grad_norm": 0.6987119913101196,
      "learning_rate": 1.265357647471561e-05,
      "loss": 1.1381,
      "step": 1977
    },
    {
      "epoch": 0.589139782944582,
      "grad_norm": 0.6947020292282104,
      "learning_rate": 1.2640723180736216e-05,
      "loss": 1.1359,
      "step": 1978
    },
    {
      "epoch": 0.5894376291442507,
      "grad_norm": 0.7157933712005615,
      "learning_rate": 1.2627865195346771e-05,
      "loss": 1.1496,
      "step": 1979
    },
    {
      "epoch": 0.5897354753439192,
      "grad_norm": 0.6700463891029358,
      "learning_rate": 1.2615002541390293e-05,
      "loss": 1.1428,
      "step": 1980
    },
    {
      "epoch": 0.5900333215435879,
      "grad_norm": 0.6767447590827942,
      "learning_rate": 1.2602135241718094e-05,
      "loss": 1.1427,
      "step": 1981
    },
    {
      "epoch": 0.5903311677432566,
      "grad_norm": 0.7008209824562073,
      "learning_rate": 1.2589263319189731e-05,
      "loss": 1.1386,
      "step": 1982
    },
    {
      "epoch": 0.5906290139429252,
      "grad_norm": 0.7000613808631897,
      "learning_rate": 1.2576386796672985e-05,
      "loss": 1.1385,
      "step": 1983
    },
    {
      "epoch": 0.5909268601425939,
      "grad_norm": 0.647712767124176,
      "learning_rate": 1.2563505697043805e-05,
      "loss": 1.1293,
      "step": 1984
    },
    {
      "epoch": 0.5912247063422625,
      "grad_norm": 0.680243968963623,
      "learning_rate": 1.255062004318627e-05,
      "loss": 1.1275,
      "step": 1985
    },
    {
      "epoch": 0.5915225525419312,
      "grad_norm": 0.6873776912689209,
      "learning_rate": 1.2537729857992549e-05,
      "loss": 1.1359,
      "step": 1986
    },
    {
      "epoch": 0.5918203987415998,
      "grad_norm": 0.7014140486717224,
      "learning_rate": 1.2524835164362866e-05,
      "loss": 1.1477,
      "step": 1987
    },
    {
      "epoch": 0.5921182449412684,
      "grad_norm": 0.6844916343688965,
      "learning_rate": 1.2511935985205446e-05,
      "loss": 1.1427,
      "step": 1988
    },
    {
      "epoch": 0.5924160911409371,
      "grad_norm": 0.7088662385940552,
      "learning_rate": 1.2499032343436492e-05,
      "loss": 1.1271,
      "step": 1989
    },
    {
      "epoch": 0.5927139373406057,
      "grad_norm": 0.6848406195640564,
      "learning_rate": 1.2486124261980134e-05,
      "loss": 1.1558,
      "step": 1990
    },
    {
      "epoch": 0.5930117835402744,
      "grad_norm": 0.7162519097328186,
      "learning_rate": 1.247321176376838e-05,
      "loss": 1.1365,
      "step": 1991
    },
    {
      "epoch": 0.5933096297399431,
      "grad_norm": 0.683224081993103,
      "learning_rate": 1.2460294871741102e-05,
      "loss": 1.1578,
      "step": 1992
    },
    {
      "epoch": 0.5936074759396117,
      "grad_norm": 0.6849249601364136,
      "learning_rate": 1.2447373608845956e-05,
      "loss": 1.1393,
      "step": 1993
    },
    {
      "epoch": 0.5939053221392804,
      "grad_norm": 0.707406222820282,
      "learning_rate": 1.2434447998038379e-05,
      "loss": 1.1604,
      "step": 1994
    },
    {
      "epoch": 0.5942031683389489,
      "grad_norm": 0.6589220762252808,
      "learning_rate": 1.2421518062281532e-05,
      "loss": 1.143,
      "step": 1995
    },
    {
      "epoch": 0.5945010145386176,
      "grad_norm": 0.6841745972633362,
      "learning_rate": 1.2408583824546247e-05,
      "loss": 1.1427,
      "step": 1996
    },
    {
      "epoch": 0.5947988607382863,
      "grad_norm": 0.6861356496810913,
      "learning_rate": 1.2395645307811014e-05,
      "loss": 1.1349,
      "step": 1997
    },
    {
      "epoch": 0.5950967069379549,
      "grad_norm": 0.7200164198875427,
      "learning_rate": 1.2382702535061917e-05,
      "loss": 1.1632,
      "step": 1998
    },
    {
      "epoch": 0.5953945531376236,
      "grad_norm": 0.6684669256210327,
      "learning_rate": 1.2369755529292598e-05,
      "loss": 1.156,
      "step": 1999
    },
    {
      "epoch": 0.5956923993372922,
      "grad_norm": 0.6981649994850159,
      "learning_rate": 1.2356804313504227e-05,
      "loss": 1.158,
      "step": 2000
    },
    {
      "epoch": 0.5956923993372922,
      "eval_loss": 1.2598075866699219,
      "eval_runtime": 15.0206,
      "eval_samples_per_second": 115.442,
      "eval_steps_per_second": 14.447,
      "step": 2000
    },
    {
      "epoch": 0.5959902455369609,
      "grad_norm": 0.6953814029693604,
      "learning_rate": 1.2343848910705449e-05,
      "loss": 1.1342,
      "step": 2001
    },
    {
      "epoch": 0.5962880917366294,
      "grad_norm": 0.6801334023475647,
      "learning_rate": 1.2330889343912347e-05,
      "loss": 1.1403,
      "step": 2002
    },
    {
      "epoch": 0.5965859379362981,
      "grad_norm": 0.6968345642089844,
      "learning_rate": 1.2317925636148404e-05,
      "loss": 1.1533,
      "step": 2003
    },
    {
      "epoch": 0.5968837841359668,
      "grad_norm": 0.7123230695724487,
      "learning_rate": 1.2304957810444455e-05,
      "loss": 1.1193,
      "step": 2004
    },
    {
      "epoch": 0.5971816303356354,
      "grad_norm": 0.6797386407852173,
      "learning_rate": 1.229198588983866e-05,
      "loss": 1.122,
      "step": 2005
    },
    {
      "epoch": 0.5974794765353041,
      "grad_norm": 0.6835977435112,
      "learning_rate": 1.2279009897376444e-05,
      "loss": 1.1418,
      "step": 2006
    },
    {
      "epoch": 0.5977773227349727,
      "grad_norm": 0.6877365708351135,
      "learning_rate": 1.226602985611047e-05,
      "loss": 1.1463,
      "step": 2007
    },
    {
      "epoch": 0.5980751689346414,
      "grad_norm": 0.6915541887283325,
      "learning_rate": 1.2253045789100597e-05,
      "loss": 1.1283,
      "step": 2008
    },
    {
      "epoch": 0.5983730151343101,
      "grad_norm": 0.7037631273269653,
      "learning_rate": 1.224005771941383e-05,
      "loss": 1.1731,
      "step": 2009
    },
    {
      "epoch": 0.5986708613339786,
      "grad_norm": 0.6700471639633179,
      "learning_rate": 1.2227065670124288e-05,
      "loss": 1.129,
      "step": 2010
    },
    {
      "epoch": 0.5989687075336473,
      "grad_norm": 0.6914956569671631,
      "learning_rate": 1.2214069664313159e-05,
      "loss": 1.1436,
      "step": 2011
    },
    {
      "epoch": 0.5992665537333159,
      "grad_norm": 0.7339617013931274,
      "learning_rate": 1.2201069725068666e-05,
      "loss": 1.1442,
      "step": 2012
    },
    {
      "epoch": 0.5995643999329846,
      "grad_norm": 0.6981349587440491,
      "learning_rate": 1.218806587548601e-05,
      "loss": 1.1531,
      "step": 2013
    },
    {
      "epoch": 0.5998622461326533,
      "grad_norm": 0.6497629880905151,
      "learning_rate": 1.2175058138667343e-05,
      "loss": 1.1289,
      "step": 2014
    },
    {
      "epoch": 0.6001600923323219,
      "grad_norm": 0.6974544525146484,
      "learning_rate": 1.2162046537721728e-05,
      "loss": 1.1299,
      "step": 2015
    },
    {
      "epoch": 0.6004579385319906,
      "grad_norm": 0.686618983745575,
      "learning_rate": 1.2149031095765087e-05,
      "loss": 1.1418,
      "step": 2016
    },
    {
      "epoch": 0.6007557847316591,
      "grad_norm": 0.665843665599823,
      "learning_rate": 1.2136011835920167e-05,
      "loss": 1.1393,
      "step": 2017
    },
    {
      "epoch": 0.6010536309313278,
      "grad_norm": 0.6776585578918457,
      "learning_rate": 1.2122988781316499e-05,
      "loss": 1.1381,
      "step": 2018
    },
    {
      "epoch": 0.6013514771309965,
      "grad_norm": 0.6570384502410889,
      "learning_rate": 1.2109961955090354e-05,
      "loss": 1.1495,
      "step": 2019
    },
    {
      "epoch": 0.6016493233306651,
      "grad_norm": 0.729286253452301,
      "learning_rate": 1.2096931380384705e-05,
      "loss": 1.1399,
      "step": 2020
    },
    {
      "epoch": 0.6019471695303338,
      "grad_norm": 0.6583660840988159,
      "learning_rate": 1.2083897080349181e-05,
      "loss": 1.1323,
      "step": 2021
    },
    {
      "epoch": 0.6022450157300024,
      "grad_norm": 0.6847886443138123,
      "learning_rate": 1.2070859078140034e-05,
      "loss": 1.1582,
      "step": 2022
    },
    {
      "epoch": 0.6025428619296711,
      "grad_norm": 0.7107734680175781,
      "learning_rate": 1.2057817396920091e-05,
      "loss": 1.1387,
      "step": 2023
    },
    {
      "epoch": 0.6028407081293398,
      "grad_norm": 0.6747090816497803,
      "learning_rate": 1.2044772059858718e-05,
      "loss": 1.1437,
      "step": 2024
    },
    {
      "epoch": 0.6031385543290083,
      "grad_norm": 0.6978010535240173,
      "learning_rate": 1.2031723090131771e-05,
      "loss": 1.1368,
      "step": 2025
    },
    {
      "epoch": 0.603436400528677,
      "grad_norm": 0.6720296144485474,
      "learning_rate": 1.2018670510921558e-05,
      "loss": 1.1442,
      "step": 2026
    },
    {
      "epoch": 0.6037342467283456,
      "grad_norm": 0.677786648273468,
      "learning_rate": 1.2005614345416808e-05,
      "loss": 1.1459,
      "step": 2027
    },
    {
      "epoch": 0.6040320929280143,
      "grad_norm": 0.6514233946800232,
      "learning_rate": 1.1992554616812613e-05,
      "loss": 1.1566,
      "step": 2028
    },
    {
      "epoch": 0.604329939127683,
      "grad_norm": 0.7058207988739014,
      "learning_rate": 1.1979491348310402e-05,
      "loss": 1.1481,
      "step": 2029
    },
    {
      "epoch": 0.6046277853273516,
      "grad_norm": 0.7299528121948242,
      "learning_rate": 1.1966424563117886e-05,
      "loss": 1.1507,
      "step": 2030
    },
    {
      "epoch": 0.6049256315270203,
      "grad_norm": 0.677352786064148,
      "learning_rate": 1.1953354284449028e-05,
      "loss": 1.1496,
      "step": 2031
    },
    {
      "epoch": 0.6052234777266888,
      "grad_norm": 0.6946365237236023,
      "learning_rate": 1.1940280535523998e-05,
      "loss": 1.1223,
      "step": 2032
    },
    {
      "epoch": 0.6055213239263575,
      "grad_norm": 0.7094717025756836,
      "learning_rate": 1.1927203339569128e-05,
      "loss": 1.1469,
      "step": 2033
    },
    {
      "epoch": 0.6058191701260262,
      "grad_norm": 0.6808870434761047,
      "learning_rate": 1.1914122719816874e-05,
      "loss": 1.1406,
      "step": 2034
    },
    {
      "epoch": 0.6061170163256948,
      "grad_norm": 0.6985769867897034,
      "learning_rate": 1.1901038699505779e-05,
      "loss": 1.1391,
      "step": 2035
    },
    {
      "epoch": 0.6064148625253635,
      "grad_norm": 0.6645321846008301,
      "learning_rate": 1.1887951301880421e-05,
      "loss": 1.1477,
      "step": 2036
    },
    {
      "epoch": 0.6067127087250321,
      "grad_norm": 0.6965930461883545,
      "learning_rate": 1.1874860550191385e-05,
      "loss": 1.1421,
      "step": 2037
    },
    {
      "epoch": 0.6070105549247008,
      "grad_norm": 0.6906652450561523,
      "learning_rate": 1.1861766467695204e-05,
      "loss": 1.1391,
      "step": 2038
    },
    {
      "epoch": 0.6073084011243695,
      "grad_norm": 0.6769377589225769,
      "learning_rate": 1.1848669077654342e-05,
      "loss": 1.1576,
      "step": 2039
    },
    {
      "epoch": 0.607606247324038,
      "grad_norm": 0.7366375923156738,
      "learning_rate": 1.1835568403337131e-05,
      "loss": 1.1433,
      "step": 2040
    },
    {
      "epoch": 0.6079040935237067,
      "grad_norm": 0.6799471378326416,
      "learning_rate": 1.1822464468017735e-05,
      "loss": 1.1444,
      "step": 2041
    },
    {
      "epoch": 0.6082019397233753,
      "grad_norm": 0.6852772831916809,
      "learning_rate": 1.1809357294976122e-05,
      "loss": 1.148,
      "step": 2042
    },
    {
      "epoch": 0.608499785923044,
      "grad_norm": 0.7014915943145752,
      "learning_rate": 1.1796246907498001e-05,
      "loss": 1.1647,
      "step": 2043
    },
    {
      "epoch": 0.6087976321227127,
      "grad_norm": 0.6880201697349548,
      "learning_rate": 1.1783133328874797e-05,
      "loss": 1.1338,
      "step": 2044
    },
    {
      "epoch": 0.6090954783223813,
      "grad_norm": 0.664293110370636,
      "learning_rate": 1.1770016582403605e-05,
      "loss": 1.1414,
      "step": 2045
    },
    {
      "epoch": 0.60939332452205,
      "grad_norm": 0.6811484098434448,
      "learning_rate": 1.1756896691387141e-05,
      "loss": 1.1408,
      "step": 2046
    },
    {
      "epoch": 0.6096911707217185,
      "grad_norm": 0.6927862167358398,
      "learning_rate": 1.1743773679133721e-05,
      "loss": 1.1462,
      "step": 2047
    },
    {
      "epoch": 0.6099890169213872,
      "grad_norm": 0.6863269805908203,
      "learning_rate": 1.173064756895719e-05,
      "loss": 1.1527,
      "step": 2048
    },
    {
      "epoch": 0.6102868631210558,
      "grad_norm": 0.6798273324966431,
      "learning_rate": 1.171751838417691e-05,
      "loss": 1.1363,
      "step": 2049
    },
    {
      "epoch": 0.6105847093207245,
      "grad_norm": 0.6919499039649963,
      "learning_rate": 1.17043861481177e-05,
      "loss": 1.1402,
      "step": 2050
    },
    {
      "epoch": 0.6108825555203932,
      "grad_norm": 0.682786762714386,
      "learning_rate": 1.1691250884109791e-05,
      "loss": 1.1492,
      "step": 2051
    },
    {
      "epoch": 0.6111804017200618,
      "grad_norm": 0.7061185836791992,
      "learning_rate": 1.167811261548881e-05,
      "loss": 1.1459,
      "step": 2052
    },
    {
      "epoch": 0.6114782479197305,
      "grad_norm": 0.7178184986114502,
      "learning_rate": 1.1664971365595714e-05,
      "loss": 1.1432,
      "step": 2053
    },
    {
      "epoch": 0.6117760941193991,
      "grad_norm": 0.7019155621528625,
      "learning_rate": 1.1651827157776754e-05,
      "loss": 1.1404,
      "step": 2054
    },
    {
      "epoch": 0.6120739403190677,
      "grad_norm": 0.661493718624115,
      "learning_rate": 1.1638680015383434e-05,
      "loss": 1.1528,
      "step": 2055
    },
    {
      "epoch": 0.6123717865187364,
      "grad_norm": 0.6963799595832825,
      "learning_rate": 1.1625529961772481e-05,
      "loss": 1.1513,
      "step": 2056
    },
    {
      "epoch": 0.612669632718405,
      "grad_norm": 0.654822051525116,
      "learning_rate": 1.161237702030579e-05,
      "loss": 1.137,
      "step": 2057
    },
    {
      "epoch": 0.6129674789180737,
      "grad_norm": 0.647721529006958,
      "learning_rate": 1.1599221214350376e-05,
      "loss": 1.1442,
      "step": 2058
    },
    {
      "epoch": 0.6132653251177423,
      "grad_norm": 0.6791771054267883,
      "learning_rate": 1.1586062567278361e-05,
      "loss": 1.1363,
      "step": 2059
    },
    {
      "epoch": 0.613563171317411,
      "grad_norm": 0.6780409216880798,
      "learning_rate": 1.1572901102466902e-05,
      "loss": 1.1325,
      "step": 2060
    },
    {
      "epoch": 0.6138610175170797,
      "grad_norm": 0.6959864497184753,
      "learning_rate": 1.1559736843298164e-05,
      "loss": 1.1542,
      "step": 2061
    },
    {
      "epoch": 0.6141588637167482,
      "grad_norm": 0.6783259510993958,
      "learning_rate": 1.154656981315928e-05,
      "loss": 1.1444,
      "step": 2062
    },
    {
      "epoch": 0.6144567099164169,
      "grad_norm": 0.6543260812759399,
      "learning_rate": 1.15334000354423e-05,
      "loss": 1.122,
      "step": 2063
    },
    {
      "epoch": 0.6147545561160855,
      "grad_norm": 0.6644838452339172,
      "learning_rate": 1.1520227533544164e-05,
      "loss": 1.1311,
      "step": 2064
    },
    {
      "epoch": 0.6150524023157542,
      "grad_norm": 0.6796572208404541,
      "learning_rate": 1.150705233086664e-05,
      "loss": 1.1446,
      "step": 2065
    },
    {
      "epoch": 0.6153502485154229,
      "grad_norm": 0.6283748149871826,
      "learning_rate": 1.1493874450816302e-05,
      "loss": 1.1303,
      "step": 2066
    },
    {
      "epoch": 0.6156480947150915,
      "grad_norm": 0.7160385847091675,
      "learning_rate": 1.1480693916804484e-05,
      "loss": 1.1519,
      "step": 2067
    },
    {
      "epoch": 0.6159459409147602,
      "grad_norm": 0.6848066449165344,
      "learning_rate": 1.1467510752247222e-05,
      "loss": 1.1424,
      "step": 2068
    },
    {
      "epoch": 0.6162437871144287,
      "grad_norm": 0.6833692789077759,
      "learning_rate": 1.1454324980565239e-05,
      "loss": 1.1506,
      "step": 2069
    },
    {
      "epoch": 0.6165416333140974,
      "grad_norm": 0.669808030128479,
      "learning_rate": 1.1441136625183885e-05,
      "loss": 1.1278,
      "step": 2070
    },
    {
      "epoch": 0.616839479513766,
      "grad_norm": 0.6874905228614807,
      "learning_rate": 1.1427945709533092e-05,
      "loss": 1.1208,
      "step": 2071
    },
    {
      "epoch": 0.6171373257134347,
      "grad_norm": 0.6975827813148499,
      "learning_rate": 1.1414752257047352e-05,
      "loss": 1.1594,
      "step": 2072
    },
    {
      "epoch": 0.6174351719131034,
      "grad_norm": 0.6715501546859741,
      "learning_rate": 1.1401556291165659e-05,
      "loss": 1.1332,
      "step": 2073
    },
    {
      "epoch": 0.617733018112772,
      "grad_norm": 0.6932191848754883,
      "learning_rate": 1.1388357835331472e-05,
      "loss": 1.1381,
      "step": 2074
    },
    {
      "epoch": 0.6180308643124407,
      "grad_norm": 0.7032844424247742,
      "learning_rate": 1.1375156912992674e-05,
      "loss": 1.1404,
      "step": 2075
    },
    {
      "epoch": 0.6183287105121094,
      "grad_norm": 0.6798725128173828,
      "learning_rate": 1.1361953547601532e-05,
      "loss": 1.1354,
      "step": 2076
    },
    {
      "epoch": 0.6186265567117779,
      "grad_norm": 0.6799058318138123,
      "learning_rate": 1.1348747762614647e-05,
      "loss": 1.1308,
      "step": 2077
    },
    {
      "epoch": 0.6189244029114466,
      "grad_norm": 0.6758434772491455,
      "learning_rate": 1.1335539581492925e-05,
      "loss": 1.1316,
      "step": 2078
    },
    {
      "epoch": 0.6192222491111152,
      "grad_norm": 0.7036418318748474,
      "learning_rate": 1.1322329027701529e-05,
      "loss": 1.1371,
      "step": 2079
    },
    {
      "epoch": 0.6195200953107839,
      "grad_norm": 0.6844927072525024,
      "learning_rate": 1.1309116124709832e-05,
      "loss": 1.1347,
      "step": 2080
    },
    {
      "epoch": 0.6198179415104526,
      "grad_norm": 0.6902563571929932,
      "learning_rate": 1.1295900895991384e-05,
      "loss": 1.1177,
      "step": 2081
    },
    {
      "epoch": 0.6201157877101212,
      "grad_norm": 0.6700466871261597,
      "learning_rate": 1.1282683365023866e-05,
      "loss": 1.1231,
      "step": 2082
    },
    {
      "epoch": 0.6204136339097899,
      "grad_norm": 0.6827241778373718,
      "learning_rate": 1.1269463555289049e-05,
      "loss": 1.1318,
      "step": 2083
    },
    {
      "epoch": 0.6207114801094584,
      "grad_norm": 0.7119588851928711,
      "learning_rate": 1.1256241490272753e-05,
      "loss": 1.1209,
      "step": 2084
    },
    {
      "epoch": 0.6210093263091271,
      "grad_norm": 0.6701129078865051,
      "learning_rate": 1.1243017193464802e-05,
      "loss": 1.1491,
      "step": 2085
    },
    {
      "epoch": 0.6213071725087957,
      "grad_norm": 0.6561959385871887,
      "learning_rate": 1.1229790688358995e-05,
      "loss": 1.13,
      "step": 2086
    },
    {
      "epoch": 0.6216050187084644,
      "grad_norm": 0.7233445644378662,
      "learning_rate": 1.1216561998453038e-05,
      "loss": 1.1499,
      "step": 2087
    },
    {
      "epoch": 0.6219028649081331,
      "grad_norm": 0.6652736663818359,
      "learning_rate": 1.120333114724853e-05,
      "loss": 1.1386,
      "step": 2088
    },
    {
      "epoch": 0.6222007111078017,
      "grad_norm": 0.6734875440597534,
      "learning_rate": 1.1190098158250904e-05,
      "loss": 1.1575,
      "step": 2089
    },
    {
      "epoch": 0.6224985573074704,
      "grad_norm": 0.672557532787323,
      "learning_rate": 1.1176863054969396e-05,
      "loss": 1.1314,
      "step": 2090
    },
    {
      "epoch": 0.622796403507139,
      "grad_norm": 0.6741400957107544,
      "learning_rate": 1.1163625860916996e-05,
      "loss": 1.1351,
      "step": 2091
    },
    {
      "epoch": 0.6230942497068076,
      "grad_norm": 0.6708105206489563,
      "learning_rate": 1.1150386599610406e-05,
      "loss": 1.1552,
      "step": 2092
    },
    {
      "epoch": 0.6233920959064763,
      "grad_norm": 0.6946595907211304,
      "learning_rate": 1.1137145294570002e-05,
      "loss": 1.1564,
      "step": 2093
    },
    {
      "epoch": 0.6236899421061449,
      "grad_norm": 0.7070640325546265,
      "learning_rate": 1.1123901969319795e-05,
      "loss": 1.1365,
      "step": 2094
    },
    {
      "epoch": 0.6239877883058136,
      "grad_norm": 0.6755852103233337,
      "learning_rate": 1.1110656647387376e-05,
      "loss": 1.1283,
      "step": 2095
    },
    {
      "epoch": 0.6242856345054822,
      "grad_norm": 0.670581579208374,
      "learning_rate": 1.1097409352303896e-05,
      "loss": 1.1319,
      "step": 2096
    },
    {
      "epoch": 0.6245834807051509,
      "grad_norm": 0.6783638000488281,
      "learning_rate": 1.1084160107603998e-05,
      "loss": 1.1333,
      "step": 2097
    },
    {
      "epoch": 0.6248813269048196,
      "grad_norm": 0.6916413307189941,
      "learning_rate": 1.1070908936825795e-05,
      "loss": 1.1392,
      "step": 2098
    },
    {
      "epoch": 0.6251791731044881,
      "grad_norm": 0.7078578472137451,
      "learning_rate": 1.1057655863510826e-05,
      "loss": 1.1413,
      "step": 2099
    },
    {
      "epoch": 0.6254770193041568,
      "grad_norm": 0.6493625044822693,
      "learning_rate": 1.1044400911203998e-05,
      "loss": 1.1417,
      "step": 2100
    },
    {
      "epoch": 0.6257748655038254,
      "grad_norm": 0.6634359955787659,
      "learning_rate": 1.1031144103453572e-05,
      "loss": 1.1363,
      "step": 2101
    },
    {
      "epoch": 0.6260727117034941,
      "grad_norm": 0.672347903251648,
      "learning_rate": 1.1017885463811094e-05,
      "loss": 1.1303,
      "step": 2102
    },
    {
      "epoch": 0.6263705579031628,
      "grad_norm": 0.6813924312591553,
      "learning_rate": 1.1004625015831362e-05,
      "loss": 1.141,
      "step": 2103
    },
    {
      "epoch": 0.6266684041028314,
      "grad_norm": 0.6820005178451538,
      "learning_rate": 1.0991362783072399e-05,
      "loss": 1.136,
      "step": 2104
    },
    {
      "epoch": 0.6269662503025001,
      "grad_norm": 0.6779842972755432,
      "learning_rate": 1.0978098789095386e-05,
      "loss": 1.1296,
      "step": 2105
    },
    {
      "epoch": 0.6272640965021687,
      "grad_norm": 0.6828364729881287,
      "learning_rate": 1.0964833057464645e-05,
      "loss": 1.1373,
      "step": 2106
    },
    {
      "epoch": 0.6275619427018373,
      "grad_norm": 0.6571195125579834,
      "learning_rate": 1.0951565611747572e-05,
      "loss": 1.1429,
      "step": 2107
    },
    {
      "epoch": 0.627859788901506,
      "grad_norm": 0.6917222738265991,
      "learning_rate": 1.093829647551462e-05,
      "loss": 1.1264,
      "step": 2108
    },
    {
      "epoch": 0.6281576351011746,
      "grad_norm": 0.6963003277778625,
      "learning_rate": 1.092502567233924e-05,
      "loss": 1.1379,
      "step": 2109
    },
    {
      "epoch": 0.6284554813008433,
      "grad_norm": 0.6526333093643188,
      "learning_rate": 1.0911753225797841e-05,
      "loss": 1.1272,
      "step": 2110
    },
    {
      "epoch": 0.6287533275005119,
      "grad_norm": 0.6915467381477356,
      "learning_rate": 1.0898479159469756e-05,
      "loss": 1.1511,
      "step": 2111
    },
    {
      "epoch": 0.6290511737001806,
      "grad_norm": 0.7004953026771545,
      "learning_rate": 1.0885203496937198e-05,
      "loss": 1.1401,
      "step": 2112
    },
    {
      "epoch": 0.6293490198998493,
      "grad_norm": 0.6833409667015076,
      "learning_rate": 1.087192626178521e-05,
      "loss": 1.12,
      "step": 2113
    },
    {
      "epoch": 0.6296468660995178,
      "grad_norm": 0.6882947087287903,
      "learning_rate": 1.0858647477601633e-05,
      "loss": 1.1286,
      "step": 2114
    },
    {
      "epoch": 0.6299447122991865,
      "grad_norm": 0.7226567268371582,
      "learning_rate": 1.0845367167977056e-05,
      "loss": 1.152,
      "step": 2115
    },
    {
      "epoch": 0.6302425584988551,
      "grad_norm": 0.6765889525413513,
      "learning_rate": 1.0832085356504787e-05,
      "loss": 1.1324,
      "step": 2116
    },
    {
      "epoch": 0.6305404046985238,
      "grad_norm": 0.7241336703300476,
      "learning_rate": 1.0818802066780785e-05,
      "loss": 1.135,
      "step": 2117
    },
    {
      "epoch": 0.6308382508981925,
      "grad_norm": 0.6980672478675842,
      "learning_rate": 1.0805517322403654e-05,
      "loss": 1.1359,
      "step": 2118
    },
    {
      "epoch": 0.6311360970978611,
      "grad_norm": 0.6826876997947693,
      "learning_rate": 1.079223114697457e-05,
      "loss": 1.1377,
      "step": 2119
    },
    {
      "epoch": 0.6314339432975298,
      "grad_norm": 0.6888819932937622,
      "learning_rate": 1.077894356409726e-05,
      "loss": 1.1324,
      "step": 2120
    },
    {
      "epoch": 0.6317317894971984,
      "grad_norm": 0.7051321268081665,
      "learning_rate": 1.0765654597377941e-05,
      "loss": 1.1213,
      "step": 2121
    },
    {
      "epoch": 0.632029635696867,
      "grad_norm": 0.6855942606925964,
      "learning_rate": 1.0752364270425294e-05,
      "loss": 1.1167,
      "step": 2122
    },
    {
      "epoch": 0.6323274818965356,
      "grad_norm": 0.654977560043335,
      "learning_rate": 1.0739072606850425e-05,
      "loss": 1.143,
      "step": 2123
    },
    {
      "epoch": 0.6326253280962043,
      "grad_norm": 0.6699578166007996,
      "learning_rate": 1.07257796302668e-05,
      "loss": 1.1428,
      "step": 2124
    },
    {
      "epoch": 0.632923174295873,
      "grad_norm": 0.6766052842140198,
      "learning_rate": 1.0712485364290224e-05,
      "loss": 1.132,
      "step": 2125
    },
    {
      "epoch": 0.6332210204955416,
      "grad_norm": 0.6708699464797974,
      "learning_rate": 1.0699189832538795e-05,
      "loss": 1.1438,
      "step": 2126
    },
    {
      "epoch": 0.6335188666952103,
      "grad_norm": 0.686987578868866,
      "learning_rate": 1.0685893058632855e-05,
      "loss": 1.1424,
      "step": 2127
    },
    {
      "epoch": 0.633816712894879,
      "grad_norm": 0.691982626914978,
      "learning_rate": 1.0672595066194959e-05,
      "loss": 1.1338,
      "step": 2128
    },
    {
      "epoch": 0.6341145590945475,
      "grad_norm": 0.6472573280334473,
      "learning_rate": 1.0659295878849822e-05,
      "loss": 1.1205,
      "step": 2129
    },
    {
      "epoch": 0.6344124052942162,
      "grad_norm": 0.6601813435554504,
      "learning_rate": 1.0645995520224279e-05,
      "loss": 1.1397,
      "step": 2130
    },
    {
      "epoch": 0.6347102514938848,
      "grad_norm": 0.6598876118659973,
      "learning_rate": 1.0632694013947254e-05,
      "loss": 1.1335,
      "step": 2131
    },
    {
      "epoch": 0.6350080976935535,
      "grad_norm": 0.6868003606796265,
      "learning_rate": 1.0619391383649701e-05,
      "loss": 1.1409,
      "step": 2132
    },
    {
      "epoch": 0.6353059438932221,
      "grad_norm": 0.6610628962516785,
      "learning_rate": 1.0606087652964585e-05,
      "loss": 1.1435,
      "step": 2133
    },
    {
      "epoch": 0.6356037900928908,
      "grad_norm": 0.6688719987869263,
      "learning_rate": 1.0592782845526808e-05,
      "loss": 1.1451,
      "step": 2134
    },
    {
      "epoch": 0.6359016362925595,
      "grad_norm": 0.6963590383529663,
      "learning_rate": 1.0579476984973191e-05,
      "loss": 1.1372,
      "step": 2135
    },
    {
      "epoch": 0.6361994824922281,
      "grad_norm": 0.6594535708427429,
      "learning_rate": 1.056617009494244e-05,
      "loss": 1.1404,
      "step": 2136
    },
    {
      "epoch": 0.6364973286918967,
      "grad_norm": 0.6717032790184021,
      "learning_rate": 1.0552862199075067e-05,
      "loss": 1.1446,
      "step": 2137
    },
    {
      "epoch": 0.6367951748915653,
      "grad_norm": 0.6859241127967834,
      "learning_rate": 1.0539553321013388e-05,
      "loss": 1.1508,
      "step": 2138
    },
    {
      "epoch": 0.637093021091234,
      "grad_norm": 0.6779268980026245,
      "learning_rate": 1.0526243484401453e-05,
      "loss": 1.1278,
      "step": 2139
    },
    {
      "epoch": 0.6373908672909027,
      "grad_norm": 0.6476926207542419,
      "learning_rate": 1.0512932712885027e-05,
      "loss": 1.1105,
      "step": 2140
    },
    {
      "epoch": 0.6376887134905713,
      "grad_norm": 0.7045871019363403,
      "learning_rate": 1.0499621030111523e-05,
      "loss": 1.1318,
      "step": 2141
    },
    {
      "epoch": 0.63798655969024,
      "grad_norm": 0.6880457401275635,
      "learning_rate": 1.048630845972998e-05,
      "loss": 1.1342,
      "step": 2142
    },
    {
      "epoch": 0.6382844058899086,
      "grad_norm": 0.6653203964233398,
      "learning_rate": 1.0472995025391015e-05,
      "loss": 1.1309,
      "step": 2143
    },
    {
      "epoch": 0.6385822520895772,
      "grad_norm": 0.6541640162467957,
      "learning_rate": 1.0459680750746775e-05,
      "loss": 1.1354,
      "step": 2144
    },
    {
      "epoch": 0.6388800982892459,
      "grad_norm": 0.6553900837898254,
      "learning_rate": 1.0446365659450906e-05,
      "loss": 1.135,
      "step": 2145
    },
    {
      "epoch": 0.6391779444889145,
      "grad_norm": 0.6877648830413818,
      "learning_rate": 1.0433049775158497e-05,
      "loss": 1.1417,
      "step": 2146
    },
    {
      "epoch": 0.6394757906885832,
      "grad_norm": 0.7078217267990112,
      "learning_rate": 1.041973312152605e-05,
      "loss": 1.1342,
      "step": 2147
    },
    {
      "epoch": 0.6397736368882518,
      "grad_norm": 0.6624390482902527,
      "learning_rate": 1.040641572221144e-05,
      "loss": 1.119,
      "step": 2148
    },
    {
      "epoch": 0.6400714830879205,
      "grad_norm": 0.6610829830169678,
      "learning_rate": 1.0393097600873854e-05,
      "loss": 1.1298,
      "step": 2149
    },
    {
      "epoch": 0.6403693292875892,
      "grad_norm": 0.6842729449272156,
      "learning_rate": 1.037977878117377e-05,
      "loss": 1.132,
      "step": 2150
    },
    {
      "epoch": 0.6406671754872577,
      "grad_norm": 0.6737504005432129,
      "learning_rate": 1.0366459286772907e-05,
      "loss": 1.1406,
      "step": 2151
    },
    {
      "epoch": 0.6409650216869264,
      "grad_norm": 0.6738559603691101,
      "learning_rate": 1.0353139141334179e-05,
      "loss": 1.1388,
      "step": 2152
    },
    {
      "epoch": 0.641262867886595,
      "grad_norm": 0.6509170532226562,
      "learning_rate": 1.033981836852166e-05,
      "loss": 1.1224,
      "step": 2153
    },
    {
      "epoch": 0.6415607140862637,
      "grad_norm": 0.6704468727111816,
      "learning_rate": 1.0326496992000535e-05,
      "loss": 1.1462,
      "step": 2154
    },
    {
      "epoch": 0.6418585602859324,
      "grad_norm": 0.6882107257843018,
      "learning_rate": 1.0313175035437065e-05,
      "loss": 1.1526,
      "step": 2155
    },
    {
      "epoch": 0.642156406485601,
      "grad_norm": 0.6650741696357727,
      "learning_rate": 1.0299852522498535e-05,
      "loss": 1.1451,
      "step": 2156
    },
    {
      "epoch": 0.6424542526852697,
      "grad_norm": 0.6646947264671326,
      "learning_rate": 1.0286529476853228e-05,
      "loss": 1.1435,
      "step": 2157
    },
    {
      "epoch": 0.6427520988849383,
      "grad_norm": 0.6840659379959106,
      "learning_rate": 1.0273205922170369e-05,
      "loss": 1.1604,
      "step": 2158
    },
    {
      "epoch": 0.6430499450846069,
      "grad_norm": 0.6488549113273621,
      "learning_rate": 1.0259881882120082e-05,
      "loss": 1.1323,
      "step": 2159
    },
    {
      "epoch": 0.6433477912842755,
      "grad_norm": 0.6758552193641663,
      "learning_rate": 1.0246557380373366e-05,
      "loss": 1.1439,
      "step": 2160
    },
    {
      "epoch": 0.6436456374839442,
      "grad_norm": 0.6631988286972046,
      "learning_rate": 1.0233232440602027e-05,
      "loss": 1.143,
      "step": 2161
    },
    {
      "epoch": 0.6439434836836129,
      "grad_norm": 0.6931985020637512,
      "learning_rate": 1.0219907086478655e-05,
      "loss": 1.1387,
      "step": 2162
    },
    {
      "epoch": 0.6442413298832815,
      "grad_norm": 0.6758232712745667,
      "learning_rate": 1.020658134167658e-05,
      "loss": 1.1471,
      "step": 2163
    },
    {
      "epoch": 0.6445391760829502,
      "grad_norm": 0.6741811037063599,
      "learning_rate": 1.0193255229869813e-05,
      "loss": 1.1486,
      "step": 2164
    },
    {
      "epoch": 0.6448370222826189,
      "grad_norm": 0.6731691956520081,
      "learning_rate": 1.017992877473304e-05,
      "loss": 1.1542,
      "step": 2165
    },
    {
      "epoch": 0.6451348684822874,
      "grad_norm": 0.678414523601532,
      "learning_rate": 1.0166601999941528e-05,
      "loss": 1.1368,
      "step": 2166
    },
    {
      "epoch": 0.6454327146819561,
      "grad_norm": 0.6740637421607971,
      "learning_rate": 1.0153274929171134e-05,
      "loss": 1.132,
      "step": 2167
    },
    {
      "epoch": 0.6457305608816247,
      "grad_norm": 0.722102701663971,
      "learning_rate": 1.0139947586098235e-05,
      "loss": 1.1326,
      "step": 2168
    },
    {
      "epoch": 0.6460284070812934,
      "grad_norm": 0.6707723736763,
      "learning_rate": 1.0126619994399684e-05,
      "loss": 1.1299,
      "step": 2169
    },
    {
      "epoch": 0.646326253280962,
      "grad_norm": 0.6616982817649841,
      "learning_rate": 1.0113292177752791e-05,
      "loss": 1.1226,
      "step": 2170
    },
    {
      "epoch": 0.6466240994806307,
      "grad_norm": 0.680734395980835,
      "learning_rate": 1.0099964159835249e-05,
      "loss": 1.1402,
      "step": 2171
    },
    {
      "epoch": 0.6469219456802994,
      "grad_norm": 0.7017213106155396,
      "learning_rate": 1.0086635964325121e-05,
      "loss": 1.1272,
      "step": 2172
    },
    {
      "epoch": 0.647219791879968,
      "grad_norm": 0.6704266667366028,
      "learning_rate": 1.0073307614900778e-05,
      "loss": 1.1277,
      "step": 2173
    },
    {
      "epoch": 0.6475176380796366,
      "grad_norm": 0.7329978346824646,
      "learning_rate": 1.0059979135240868e-05,
      "loss": 1.1559,
      "step": 2174
    },
    {
      "epoch": 0.6478154842793052,
      "grad_norm": 0.6982532143592834,
      "learning_rate": 1.0046650549024267e-05,
      "loss": 1.1158,
      "step": 2175
    },
    {
      "epoch": 0.6481133304789739,
      "grad_norm": 0.65510094165802,
      "learning_rate": 1.0033321879930045e-05,
      "loss": 1.1408,
      "step": 2176
    },
    {
      "epoch": 0.6484111766786426,
      "grad_norm": 0.6939747333526611,
      "learning_rate": 1.0019993151637419e-05,
      "loss": 1.1317,
      "step": 2177
    },
    {
      "epoch": 0.6487090228783112,
      "grad_norm": 0.6660928130149841,
      "learning_rate": 1.0006664387825709e-05,
      "loss": 1.1295,
      "step": 2178
    },
    {
      "epoch": 0.6490068690779799,
      "grad_norm": 0.6851565837860107,
      "learning_rate": 9.993335612174295e-06,
      "loss": 1.142,
      "step": 2179
    },
    {
      "epoch": 0.6493047152776485,
      "grad_norm": 0.6870267987251282,
      "learning_rate": 9.980006848362583e-06,
      "loss": 1.118,
      "step": 2180
    },
    {
      "epoch": 0.6496025614773171,
      "grad_norm": 0.6830543875694275,
      "learning_rate": 9.966678120069957e-06,
      "loss": 1.157,
      "step": 2181
    },
    {
      "epoch": 0.6499004076769858,
      "grad_norm": 0.6850574612617493,
      "learning_rate": 9.953349450975736e-06,
      "loss": 1.1442,
      "step": 2182
    },
    {
      "epoch": 0.6501982538766544,
      "grad_norm": 0.6731210350990295,
      "learning_rate": 9.940020864759137e-06,
      "loss": 1.1424,
      "step": 2183
    },
    {
      "epoch": 0.6504961000763231,
      "grad_norm": 0.6542707681655884,
      "learning_rate": 9.926692385099224e-06,
      "loss": 1.1502,
      "step": 2184
    },
    {
      "epoch": 0.6507939462759917,
      "grad_norm": 0.6714897155761719,
      "learning_rate": 9.913364035674882e-06,
      "loss": 1.1329,
      "step": 2185
    },
    {
      "epoch": 0.6510917924756604,
      "grad_norm": 0.6601583957672119,
      "learning_rate": 9.900035840164753e-06,
      "loss": 1.1494,
      "step": 2186
    },
    {
      "epoch": 0.6513896386753291,
      "grad_norm": 0.6607711911201477,
      "learning_rate": 9.88670782224721e-06,
      "loss": 1.1337,
      "step": 2187
    },
    {
      "epoch": 0.6516874848749977,
      "grad_norm": 0.6423702836036682,
      "learning_rate": 9.873380005600316e-06,
      "loss": 1.1307,
      "step": 2188
    },
    {
      "epoch": 0.6519853310746663,
      "grad_norm": 0.6527242660522461,
      "learning_rate": 9.860052413901766e-06,
      "loss": 1.1159,
      "step": 2189
    },
    {
      "epoch": 0.6522831772743349,
      "grad_norm": 0.6683933734893799,
      "learning_rate": 9.846725070828866e-06,
      "loss": 1.1317,
      "step": 2190
    },
    {
      "epoch": 0.6525810234740036,
      "grad_norm": 0.6529133915901184,
      "learning_rate": 9.833398000058473e-06,
      "loss": 1.1351,
      "step": 2191
    },
    {
      "epoch": 0.6528788696736723,
      "grad_norm": 0.6819683313369751,
      "learning_rate": 9.820071225266962e-06,
      "loss": 1.1413,
      "step": 2192
    },
    {
      "epoch": 0.6531767158733409,
      "grad_norm": 0.6981674432754517,
      "learning_rate": 9.806744770130185e-06,
      "loss": 1.1179,
      "step": 2193
    },
    {
      "epoch": 0.6534745620730096,
      "grad_norm": 0.6672588586807251,
      "learning_rate": 9.793418658323422e-06,
      "loss": 1.1218,
      "step": 2194
    },
    {
      "epoch": 0.6537724082726782,
      "grad_norm": 0.6528816223144531,
      "learning_rate": 9.780092913521345e-06,
      "loss": 1.1382,
      "step": 2195
    },
    {
      "epoch": 0.6540702544723468,
      "grad_norm": 0.6421229839324951,
      "learning_rate": 9.766767559397978e-06,
      "loss": 1.1354,
      "step": 2196
    },
    {
      "epoch": 0.6543681006720155,
      "grad_norm": 0.6811882853507996,
      "learning_rate": 9.753442619626638e-06,
      "loss": 1.1385,
      "step": 2197
    },
    {
      "epoch": 0.6546659468716841,
      "grad_norm": 0.6837962865829468,
      "learning_rate": 9.74011811787992e-06,
      "loss": 1.1442,
      "step": 2198
    },
    {
      "epoch": 0.6549637930713528,
      "grad_norm": 0.6546584963798523,
      "learning_rate": 9.726794077829636e-06,
      "loss": 1.1351,
      "step": 2199
    },
    {
      "epoch": 0.6552616392710214,
      "grad_norm": 0.6966769099235535,
      "learning_rate": 9.713470523146777e-06,
      "loss": 1.1441,
      "step": 2200
    },
    {
      "epoch": 0.6555594854706901,
      "grad_norm": 0.6399734020233154,
      "learning_rate": 9.700147477501469e-06,
      "loss": 1.1362,
      "step": 2201
    },
    {
      "epoch": 0.6558573316703588,
      "grad_norm": 0.6557201743125916,
      "learning_rate": 9.686824964562942e-06,
      "loss": 1.1278,
      "step": 2202
    },
    {
      "epoch": 0.6561551778700274,
      "grad_norm": 0.6978292465209961,
      "learning_rate": 9.673503007999469e-06,
      "loss": 1.1415,
      "step": 2203
    },
    {
      "epoch": 0.656453024069696,
      "grad_norm": 0.636643648147583,
      "learning_rate": 9.660181631478343e-06,
      "loss": 1.1339,
      "step": 2204
    },
    {
      "epoch": 0.6567508702693646,
      "grad_norm": 0.6766774654388428,
      "learning_rate": 9.646860858665825e-06,
      "loss": 1.131,
      "step": 2205
    },
    {
      "epoch": 0.6570487164690333,
      "grad_norm": 0.6620157957077026,
      "learning_rate": 9.633540713227095e-06,
      "loss": 1.1424,
      "step": 2206
    },
    {
      "epoch": 0.657346562668702,
      "grad_norm": 0.6546555161476135,
      "learning_rate": 9.620221218826233e-06,
      "loss": 1.1525,
      "step": 2207
    },
    {
      "epoch": 0.6576444088683706,
      "grad_norm": 0.644629180431366,
      "learning_rate": 9.606902399126148e-06,
      "loss": 1.1358,
      "step": 2208
    },
    {
      "epoch": 0.6579422550680393,
      "grad_norm": 0.6972518563270569,
      "learning_rate": 9.593584277788563e-06,
      "loss": 1.1308,
      "step": 2209
    },
    {
      "epoch": 0.6582401012677079,
      "grad_norm": 0.6693728566169739,
      "learning_rate": 9.580266878473952e-06,
      "loss": 1.1327,
      "step": 2210
    },
    {
      "epoch": 0.6585379474673765,
      "grad_norm": 0.691297173500061,
      "learning_rate": 9.566950224841506e-06,
      "loss": 1.1339,
      "step": 2211
    },
    {
      "epoch": 0.6588357936670451,
      "grad_norm": 0.7126587629318237,
      "learning_rate": 9.553634340549098e-06,
      "loss": 1.1566,
      "step": 2212
    },
    {
      "epoch": 0.6591336398667138,
      "grad_norm": 0.6529019474983215,
      "learning_rate": 9.540319249253229e-06,
      "loss": 1.1228,
      "step": 2213
    },
    {
      "epoch": 0.6594314860663825,
      "grad_norm": 0.6743404269218445,
      "learning_rate": 9.527004974608987e-06,
      "loss": 1.1424,
      "step": 2214
    },
    {
      "epoch": 0.6597293322660511,
      "grad_norm": 0.6484033465385437,
      "learning_rate": 9.513691540270023e-06,
      "loss": 1.1401,
      "step": 2215
    },
    {
      "epoch": 0.6600271784657198,
      "grad_norm": 0.6615245342254639,
      "learning_rate": 9.500378969888479e-06,
      "loss": 1.1387,
      "step": 2216
    },
    {
      "epoch": 0.6603250246653884,
      "grad_norm": 0.6319909691810608,
      "learning_rate": 9.487067287114978e-06,
      "loss": 1.1248,
      "step": 2217
    },
    {
      "epoch": 0.6606228708650571,
      "grad_norm": 0.6804768443107605,
      "learning_rate": 9.47375651559855e-06,
      "loss": 1.1398,
      "step": 2218
    },
    {
      "epoch": 0.6609207170647257,
      "grad_norm": 0.6620609164237976,
      "learning_rate": 9.460446678986616e-06,
      "loss": 1.1277,
      "step": 2219
    },
    {
      "epoch": 0.6612185632643943,
      "grad_norm": 0.6449120044708252,
      "learning_rate": 9.447137800924937e-06,
      "loss": 1.1288,
      "step": 2220
    },
    {
      "epoch": 0.661516409464063,
      "grad_norm": 0.6873780488967896,
      "learning_rate": 9.433829905057562e-06,
      "loss": 1.1278,
      "step": 2221
    },
    {
      "epoch": 0.6618142556637316,
      "grad_norm": 0.6534948945045471,
      "learning_rate": 9.420523015026809e-06,
      "loss": 1.1296,
      "step": 2222
    },
    {
      "epoch": 0.6621121018634003,
      "grad_norm": 0.642666757106781,
      "learning_rate": 9.407217154473195e-06,
      "loss": 1.132,
      "step": 2223
    },
    {
      "epoch": 0.662409948063069,
      "grad_norm": 0.6640914678573608,
      "learning_rate": 9.393912347035417e-06,
      "loss": 1.1279,
      "step": 2224
    },
    {
      "epoch": 0.6627077942627376,
      "grad_norm": 0.6830190420150757,
      "learning_rate": 9.380608616350297e-06,
      "loss": 1.1473,
      "step": 2225
    },
    {
      "epoch": 0.6630056404624062,
      "grad_norm": 0.6736395359039307,
      "learning_rate": 9.367305986052748e-06,
      "loss": 1.1445,
      "step": 2226
    },
    {
      "epoch": 0.6633034866620748,
      "grad_norm": 0.6638702154159546,
      "learning_rate": 9.354004479775722e-06,
      "loss": 1.1213,
      "step": 2227
    },
    {
      "epoch": 0.6636013328617435,
      "grad_norm": 0.6619669198989868,
      "learning_rate": 9.340704121150182e-06,
      "loss": 1.1306,
      "step": 2228
    },
    {
      "epoch": 0.6638991790614122,
      "grad_norm": 0.6406480669975281,
      "learning_rate": 9.327404933805041e-06,
      "loss": 1.1426,
      "step": 2229
    },
    {
      "epoch": 0.6641970252610808,
      "grad_norm": 0.6560547947883606,
      "learning_rate": 9.314106941367148e-06,
      "loss": 1.1335,
      "step": 2230
    },
    {
      "epoch": 0.6644948714607495,
      "grad_norm": 0.6443886160850525,
      "learning_rate": 9.300810167461209e-06,
      "loss": 1.1436,
      "step": 2231
    },
    {
      "epoch": 0.6647927176604181,
      "grad_norm": 0.6449397206306458,
      "learning_rate": 9.28751463570978e-06,
      "loss": 1.1408,
      "step": 2232
    },
    {
      "epoch": 0.6650905638600868,
      "grad_norm": 0.6805150508880615,
      "learning_rate": 9.274220369733205e-06,
      "loss": 1.1372,
      "step": 2233
    },
    {
      "epoch": 0.6653884100597554,
      "grad_norm": 0.6383156180381775,
      "learning_rate": 9.260927393149579e-06,
      "loss": 1.1332,
      "step": 2234
    },
    {
      "epoch": 0.665686256259424,
      "grad_norm": 0.6433404088020325,
      "learning_rate": 9.247635729574707e-06,
      "loss": 1.1138,
      "step": 2235
    },
    {
      "epoch": 0.6659841024590927,
      "grad_norm": 0.6738380193710327,
      "learning_rate": 9.234345402622065e-06,
      "loss": 1.1274,
      "step": 2236
    },
    {
      "epoch": 0.6662819486587613,
      "grad_norm": 0.6623657941818237,
      "learning_rate": 9.221056435902747e-06,
      "loss": 1.118,
      "step": 2237
    },
    {
      "epoch": 0.66657979485843,
      "grad_norm": 0.6688159704208374,
      "learning_rate": 9.207768853025432e-06,
      "loss": 1.1415,
      "step": 2238
    },
    {
      "epoch": 0.6668776410580987,
      "grad_norm": 0.6684173345565796,
      "learning_rate": 9.19448267759635e-06,
      "loss": 1.1348,
      "step": 2239
    },
    {
      "epoch": 0.6671754872577673,
      "grad_norm": 0.6793352961540222,
      "learning_rate": 9.181197933219218e-06,
      "loss": 1.1259,
      "step": 2240
    },
    {
      "epoch": 0.6674733334574359,
      "grad_norm": 0.6629734039306641,
      "learning_rate": 9.167914643495217e-06,
      "loss": 1.1491,
      "step": 2241
    },
    {
      "epoch": 0.6677711796571045,
      "grad_norm": 0.6473674178123474,
      "learning_rate": 9.154632832022945e-06,
      "loss": 1.1342,
      "step": 2242
    },
    {
      "epoch": 0.6680690258567732,
      "grad_norm": 0.6652291417121887,
      "learning_rate": 9.141352522398369e-06,
      "loss": 1.1369,
      "step": 2243
    },
    {
      "epoch": 0.6683668720564419,
      "grad_norm": 0.6642878651618958,
      "learning_rate": 9.128073738214792e-06,
      "loss": 1.1241,
      "step": 2244
    },
    {
      "epoch": 0.6686647182561105,
      "grad_norm": 0.6756296753883362,
      "learning_rate": 9.114796503062805e-06,
      "loss": 1.1277,
      "step": 2245
    },
    {
      "epoch": 0.6689625644557792,
      "grad_norm": 0.6632552146911621,
      "learning_rate": 9.101520840530245e-06,
      "loss": 1.1201,
      "step": 2246
    },
    {
      "epoch": 0.6692604106554478,
      "grad_norm": 0.7091302871704102,
      "learning_rate": 9.088246774202162e-06,
      "loss": 1.1489,
      "step": 2247
    },
    {
      "epoch": 0.6695582568551164,
      "grad_norm": 0.6719722747802734,
      "learning_rate": 9.074974327660763e-06,
      "loss": 1.1046,
      "step": 2248
    },
    {
      "epoch": 0.669856103054785,
      "grad_norm": 0.6545542478561401,
      "learning_rate": 9.061703524485383e-06,
      "loss": 1.1661,
      "step": 2249
    },
    {
      "epoch": 0.6701539492544537,
      "grad_norm": 0.6651651263237,
      "learning_rate": 9.048434388252431e-06,
      "loss": 1.1301,
      "step": 2250
    },
    {
      "epoch": 0.6704517954541224,
      "grad_norm": 0.6643115282058716,
      "learning_rate": 9.035166942535358e-06,
      "loss": 1.1341,
      "step": 2251
    },
    {
      "epoch": 0.670749641653791,
      "grad_norm": 0.6458839774131775,
      "learning_rate": 9.021901210904616e-06,
      "loss": 1.1472,
      "step": 2252
    },
    {
      "epoch": 0.6710474878534597,
      "grad_norm": 0.6839348673820496,
      "learning_rate": 9.008637216927604e-06,
      "loss": 1.1482,
      "step": 2253
    },
    {
      "epoch": 0.6713453340531284,
      "grad_norm": 0.6548159122467041,
      "learning_rate": 8.99537498416864e-06,
      "loss": 1.1258,
      "step": 2254
    },
    {
      "epoch": 0.671643180252797,
      "grad_norm": 0.6492699384689331,
      "learning_rate": 8.982114536188911e-06,
      "loss": 1.13,
      "step": 2255
    },
    {
      "epoch": 0.6719410264524656,
      "grad_norm": 0.6461219787597656,
      "learning_rate": 8.96885589654643e-06,
      "loss": 1.1217,
      "step": 2256
    },
    {
      "epoch": 0.6722388726521342,
      "grad_norm": 0.6555005311965942,
      "learning_rate": 8.955599088796002e-06,
      "loss": 1.1374,
      "step": 2257
    },
    {
      "epoch": 0.6725367188518029,
      "grad_norm": 0.693929135799408,
      "learning_rate": 8.942344136489174e-06,
      "loss": 1.15,
      "step": 2258
    },
    {
      "epoch": 0.6728345650514715,
      "grad_norm": 0.6770069003105164,
      "learning_rate": 8.929091063174205e-06,
      "loss": 1.1427,
      "step": 2259
    },
    {
      "epoch": 0.6731324112511402,
      "grad_norm": 0.6837203502655029,
      "learning_rate": 8.915839892396004e-06,
      "loss": 1.1349,
      "step": 2260
    },
    {
      "epoch": 0.6734302574508089,
      "grad_norm": 0.6547746062278748,
      "learning_rate": 8.902590647696104e-06,
      "loss": 1.14,
      "step": 2261
    },
    {
      "epoch": 0.6737281036504775,
      "grad_norm": 0.6486870646476746,
      "learning_rate": 8.889343352612624e-06,
      "loss": 1.1258,
      "step": 2262
    },
    {
      "epoch": 0.6740259498501461,
      "grad_norm": 0.640608549118042,
      "learning_rate": 8.87609803068021e-06,
      "loss": 1.1265,
      "step": 2263
    },
    {
      "epoch": 0.6743237960498147,
      "grad_norm": 0.6568483114242554,
      "learning_rate": 8.862854705430003e-06,
      "loss": 1.123,
      "step": 2264
    },
    {
      "epoch": 0.6746216422494834,
      "grad_norm": 0.6563929915428162,
      "learning_rate": 8.849613400389599e-06,
      "loss": 1.1329,
      "step": 2265
    },
    {
      "epoch": 0.6749194884491521,
      "grad_norm": 0.6328772902488708,
      "learning_rate": 8.836374139083011e-06,
      "loss": 1.1211,
      "step": 2266
    },
    {
      "epoch": 0.6752173346488207,
      "grad_norm": 0.6664602756500244,
      "learning_rate": 8.823136945030607e-06,
      "loss": 1.1269,
      "step": 2267
    },
    {
      "epoch": 0.6755151808484894,
      "grad_norm": 0.6447811722755432,
      "learning_rate": 8.809901841749101e-06,
      "loss": 1.1361,
      "step": 2268
    },
    {
      "epoch": 0.675813027048158,
      "grad_norm": 0.662611722946167,
      "learning_rate": 8.796668852751476e-06,
      "loss": 1.1337,
      "step": 2269
    },
    {
      "epoch": 0.6761108732478267,
      "grad_norm": 0.6567407250404358,
      "learning_rate": 8.783438001546967e-06,
      "loss": 1.1512,
      "step": 2270
    },
    {
      "epoch": 0.6764087194474953,
      "grad_norm": 0.6595656275749207,
      "learning_rate": 8.77020931164101e-06,
      "loss": 1.1099,
      "step": 2271
    },
    {
      "epoch": 0.6767065656471639,
      "grad_norm": 0.6460320949554443,
      "learning_rate": 8.756982806535201e-06,
      "loss": 1.1258,
      "step": 2272
    },
    {
      "epoch": 0.6770044118468326,
      "grad_norm": 0.6761646866798401,
      "learning_rate": 8.74375850972725e-06,
      "loss": 1.1527,
      "step": 2273
    },
    {
      "epoch": 0.6773022580465012,
      "grad_norm": 0.6467899680137634,
      "learning_rate": 8.730536444710955e-06,
      "loss": 1.121,
      "step": 2274
    },
    {
      "epoch": 0.6776001042461699,
      "grad_norm": 0.6596039533615112,
      "learning_rate": 8.717316634976137e-06,
      "loss": 1.144,
      "step": 2275
    },
    {
      "epoch": 0.6778979504458386,
      "grad_norm": 0.6496226191520691,
      "learning_rate": 8.70409910400862e-06,
      "loss": 1.1325,
      "step": 2276
    },
    {
      "epoch": 0.6781957966455072,
      "grad_norm": 0.6658417582511902,
      "learning_rate": 8.690883875290173e-06,
      "loss": 1.1404,
      "step": 2277
    },
    {
      "epoch": 0.6784936428451758,
      "grad_norm": 0.6632696986198425,
      "learning_rate": 8.677670972298472e-06,
      "loss": 1.1126,
      "step": 2278
    },
    {
      "epoch": 0.6787914890448444,
      "grad_norm": 0.6605827212333679,
      "learning_rate": 8.664460418507077e-06,
      "loss": 1.1335,
      "step": 2279
    },
    {
      "epoch": 0.6790893352445131,
      "grad_norm": 0.6379512548446655,
      "learning_rate": 8.651252237385354e-06,
      "loss": 1.1257,
      "step": 2280
    },
    {
      "epoch": 0.6793871814441818,
      "grad_norm": 0.6435519456863403,
      "learning_rate": 8.638046452398473e-06,
      "loss": 1.1272,
      "step": 2281
    },
    {
      "epoch": 0.6796850276438504,
      "grad_norm": 0.676419734954834,
      "learning_rate": 8.62484308700733e-06,
      "loss": 1.1313,
      "step": 2282
    },
    {
      "epoch": 0.6799828738435191,
      "grad_norm": 0.6505060791969299,
      "learning_rate": 8.61164216466853e-06,
      "loss": 1.1262,
      "step": 2283
    },
    {
      "epoch": 0.6802807200431877,
      "grad_norm": 0.6730798482894897,
      "learning_rate": 8.598443708834345e-06,
      "loss": 1.1155,
      "step": 2284
    },
    {
      "epoch": 0.6805785662428564,
      "grad_norm": 0.6733856201171875,
      "learning_rate": 8.58524774295265e-06,
      "loss": 1.1435,
      "step": 2285
    },
    {
      "epoch": 0.680876412442525,
      "grad_norm": 0.6233336329460144,
      "learning_rate": 8.572054290466911e-06,
      "loss": 1.1193,
      "step": 2286
    },
    {
      "epoch": 0.6811742586421936,
      "grad_norm": 0.6938530206680298,
      "learning_rate": 8.55886337481612e-06,
      "loss": 1.1221,
      "step": 2287
    },
    {
      "epoch": 0.6814721048418623,
      "grad_norm": 0.6775168180465698,
      "learning_rate": 8.545675019434761e-06,
      "loss": 1.1317,
      "step": 2288
    },
    {
      "epoch": 0.6817699510415309,
      "grad_norm": 0.6954297423362732,
      "learning_rate": 8.532489247752778e-06,
      "loss": 1.1447,
      "step": 2289
    },
    {
      "epoch": 0.6820677972411996,
      "grad_norm": 0.6656424403190613,
      "learning_rate": 8.519306083195518e-06,
      "loss": 1.1297,
      "step": 2290
    },
    {
      "epoch": 0.6823656434408683,
      "grad_norm": 0.6949697136878967,
      "learning_rate": 8.506125549183698e-06,
      "loss": 1.1361,
      "step": 2291
    },
    {
      "epoch": 0.6826634896405369,
      "grad_norm": 0.6478724479675293,
      "learning_rate": 8.492947669133362e-06,
      "loss": 1.1365,
      "step": 2292
    },
    {
      "epoch": 0.6829613358402055,
      "grad_norm": 0.6534660458564758,
      "learning_rate": 8.479772466455838e-06,
      "loss": 1.1364,
      "step": 2293
    },
    {
      "epoch": 0.6832591820398741,
      "grad_norm": 0.6817371845245361,
      "learning_rate": 8.4665999645577e-06,
      "loss": 1.1176,
      "step": 2294
    },
    {
      "epoch": 0.6835570282395428,
      "grad_norm": 0.6439172029495239,
      "learning_rate": 8.453430186840721e-06,
      "loss": 1.1199,
      "step": 2295
    },
    {
      "epoch": 0.6838548744392114,
      "grad_norm": 0.6628604531288147,
      "learning_rate": 8.440263156701835e-06,
      "loss": 1.1412,
      "step": 2296
    },
    {
      "epoch": 0.6841527206388801,
      "grad_norm": 0.6778392195701599,
      "learning_rate": 8.427098897533101e-06,
      "loss": 1.1455,
      "step": 2297
    },
    {
      "epoch": 0.6844505668385488,
      "grad_norm": 0.7025892734527588,
      "learning_rate": 8.413937432721644e-06,
      "loss": 1.1281,
      "step": 2298
    },
    {
      "epoch": 0.6847484130382174,
      "grad_norm": 0.653150737285614,
      "learning_rate": 8.400778785649626e-06,
      "loss": 1.1314,
      "step": 2299
    },
    {
      "epoch": 0.6850462592378861,
      "grad_norm": 0.6294824481010437,
      "learning_rate": 8.387622979694215e-06,
      "loss": 1.1224,
      "step": 2300
    },
    {
      "epoch": 0.6853441054375546,
      "grad_norm": 0.6526135802268982,
      "learning_rate": 8.374470038227522e-06,
      "loss": 1.1445,
      "step": 2301
    },
    {
      "epoch": 0.6856419516372233,
      "grad_norm": 0.6501734256744385,
      "learning_rate": 8.361319984616568e-06,
      "loss": 1.1291,
      "step": 2302
    },
    {
      "epoch": 0.685939797836892,
      "grad_norm": 0.6555228233337402,
      "learning_rate": 8.348172842223251e-06,
      "loss": 1.1335,
      "step": 2303
    },
    {
      "epoch": 0.6862376440365606,
      "grad_norm": 0.6802067756652832,
      "learning_rate": 8.335028634404292e-06,
      "loss": 1.1314,
      "step": 2304
    },
    {
      "epoch": 0.6865354902362293,
      "grad_norm": 0.6444189548492432,
      "learning_rate": 8.321887384511191e-06,
      "loss": 1.1214,
      "step": 2305
    },
    {
      "epoch": 0.686833336435898,
      "grad_norm": 0.6357744336128235,
      "learning_rate": 8.308749115890212e-06,
      "loss": 1.1177,
      "step": 2306
    },
    {
      "epoch": 0.6871311826355666,
      "grad_norm": 0.655558705329895,
      "learning_rate": 8.295613851882305e-06,
      "loss": 1.1322,
      "step": 2307
    },
    {
      "epoch": 0.6874290288352352,
      "grad_norm": 0.6493598222732544,
      "learning_rate": 8.282481615823092e-06,
      "loss": 1.1397,
      "step": 2308
    },
    {
      "epoch": 0.6877268750349038,
      "grad_norm": 0.6626792550086975,
      "learning_rate": 8.269352431042813e-06,
      "loss": 1.1241,
      "step": 2309
    },
    {
      "epoch": 0.6880247212345725,
      "grad_norm": 0.6442959904670715,
      "learning_rate": 8.256226320866282e-06,
      "loss": 1.1329,
      "step": 2310
    },
    {
      "epoch": 0.6883225674342411,
      "grad_norm": 0.6626075506210327,
      "learning_rate": 8.24310330861286e-06,
      "loss": 1.1384,
      "step": 2311
    },
    {
      "epoch": 0.6886204136339098,
      "grad_norm": 0.6521046161651611,
      "learning_rate": 8.229983417596399e-06,
      "loss": 1.1117,
      "step": 2312
    },
    {
      "epoch": 0.6889182598335785,
      "grad_norm": 0.6688541769981384,
      "learning_rate": 8.216866671125205e-06,
      "loss": 1.1327,
      "step": 2313
    },
    {
      "epoch": 0.6892161060332471,
      "grad_norm": 0.6318952441215515,
      "learning_rate": 8.203753092502002e-06,
      "loss": 1.1273,
      "step": 2314
    },
    {
      "epoch": 0.6895139522329158,
      "grad_norm": 0.6557425260543823,
      "learning_rate": 8.19064270502388e-06,
      "loss": 1.1262,
      "step": 2315
    },
    {
      "epoch": 0.6898117984325843,
      "grad_norm": 0.6658375859260559,
      "learning_rate": 8.177535531982266e-06,
      "loss": 1.1262,
      "step": 2316
    },
    {
      "epoch": 0.690109644632253,
      "grad_norm": 0.7000569105148315,
      "learning_rate": 8.164431596662872e-06,
      "loss": 1.1401,
      "step": 2317
    },
    {
      "epoch": 0.6904074908319217,
      "grad_norm": 0.634206235408783,
      "learning_rate": 8.15133092234566e-06,
      "loss": 1.1201,
      "step": 2318
    },
    {
      "epoch": 0.6907053370315903,
      "grad_norm": 0.6456895470619202,
      "learning_rate": 8.1382335323048e-06,
      "loss": 1.139,
      "step": 2319
    },
    {
      "epoch": 0.691003183231259,
      "grad_norm": 0.6743967533111572,
      "learning_rate": 8.125139449808618e-06,
      "loss": 1.1445,
      "step": 2320
    },
    {
      "epoch": 0.6913010294309276,
      "grad_norm": 0.6877643465995789,
      "learning_rate": 8.11204869811958e-06,
      "loss": 1.1319,
      "step": 2321
    },
    {
      "epoch": 0.6915988756305963,
      "grad_norm": 0.658657431602478,
      "learning_rate": 8.098961300494221e-06,
      "loss": 1.1439,
      "step": 2322
    },
    {
      "epoch": 0.6918967218302648,
      "grad_norm": 0.6721373200416565,
      "learning_rate": 8.085877280183126e-06,
      "loss": 1.135,
      "step": 2323
    },
    {
      "epoch": 0.6921945680299335,
      "grad_norm": 0.6626288890838623,
      "learning_rate": 8.072796660430873e-06,
      "loss": 1.1281,
      "step": 2324
    },
    {
      "epoch": 0.6924924142296022,
      "grad_norm": 0.6611378192901611,
      "learning_rate": 8.059719464476e-06,
      "loss": 1.1355,
      "step": 2325
    },
    {
      "epoch": 0.6927902604292708,
      "grad_norm": 0.6538137197494507,
      "learning_rate": 8.046645715550972e-06,
      "loss": 1.1217,
      "step": 2326
    },
    {
      "epoch": 0.6930881066289395,
      "grad_norm": 0.6530148983001709,
      "learning_rate": 8.033575436882115e-06,
      "loss": 1.1108,
      "step": 2327
    },
    {
      "epoch": 0.6933859528286082,
      "grad_norm": 0.6648549437522888,
      "learning_rate": 8.0205086516896e-06,
      "loss": 1.1339,
      "step": 2328
    },
    {
      "epoch": 0.6936837990282768,
      "grad_norm": 0.6547403931617737,
      "learning_rate": 8.007445383187387e-06,
      "loss": 1.1257,
      "step": 2329
    },
    {
      "epoch": 0.6939816452279454,
      "grad_norm": 0.6955691576004028,
      "learning_rate": 7.994385654583192e-06,
      "loss": 1.1492,
      "step": 2330
    },
    {
      "epoch": 0.694279491427614,
      "grad_norm": 0.6498315334320068,
      "learning_rate": 7.981329489078447e-06,
      "loss": 1.1219,
      "step": 2331
    },
    {
      "epoch": 0.6945773376272827,
      "grad_norm": 0.642760157585144,
      "learning_rate": 7.968276909868234e-06,
      "loss": 1.1297,
      "step": 2332
    },
    {
      "epoch": 0.6948751838269513,
      "grad_norm": 0.6677197217941284,
      "learning_rate": 7.955227940141287e-06,
      "loss": 1.1416,
      "step": 2333
    },
    {
      "epoch": 0.69517303002662,
      "grad_norm": 0.6572502255439758,
      "learning_rate": 7.94218260307991e-06,
      "loss": 1.1303,
      "step": 2334
    },
    {
      "epoch": 0.6954708762262887,
      "grad_norm": 0.6804179549217224,
      "learning_rate": 7.92914092185997e-06,
      "loss": 1.1468,
      "step": 2335
    },
    {
      "epoch": 0.6957687224259573,
      "grad_norm": 0.659386396408081,
      "learning_rate": 7.916102919650826e-06,
      "loss": 1.1284,
      "step": 2336
    },
    {
      "epoch": 0.696066568625626,
      "grad_norm": 0.6514713764190674,
      "learning_rate": 7.9030686196153e-06,
      "loss": 1.1385,
      "step": 2337
    },
    {
      "epoch": 0.6963644148252945,
      "grad_norm": 0.6449576616287231,
      "learning_rate": 7.890038044909651e-06,
      "loss": 1.1282,
      "step": 2338
    },
    {
      "epoch": 0.6966622610249632,
      "grad_norm": 0.6538965702056885,
      "learning_rate": 7.877011218683503e-06,
      "loss": 1.1114,
      "step": 2339
    },
    {
      "epoch": 0.6969601072246319,
      "grad_norm": 0.661262035369873,
      "learning_rate": 7.863988164079837e-06,
      "loss": 1.1419,
      "step": 2340
    },
    {
      "epoch": 0.6972579534243005,
      "grad_norm": 0.6642847657203674,
      "learning_rate": 7.850968904234918e-06,
      "loss": 1.1269,
      "step": 2341
    },
    {
      "epoch": 0.6975557996239692,
      "grad_norm": 0.6540550589561462,
      "learning_rate": 7.837953462278273e-06,
      "loss": 1.1283,
      "step": 2342
    },
    {
      "epoch": 0.6978536458236378,
      "grad_norm": 0.6738083362579346,
      "learning_rate": 7.82494186133266e-06,
      "loss": 1.1382,
      "step": 2343
    },
    {
      "epoch": 0.6981514920233065,
      "grad_norm": 0.6360319256782532,
      "learning_rate": 7.811934124513994e-06,
      "loss": 1.1176,
      "step": 2344
    },
    {
      "epoch": 0.6984493382229751,
      "grad_norm": 0.6675420999526978,
      "learning_rate": 7.798930274931338e-06,
      "loss": 1.1442,
      "step": 2345
    },
    {
      "epoch": 0.6987471844226437,
      "grad_norm": 0.6733787059783936,
      "learning_rate": 7.785930335686845e-06,
      "loss": 1.1464,
      "step": 2346
    },
    {
      "epoch": 0.6990450306223124,
      "grad_norm": 0.6493228673934937,
      "learning_rate": 7.772934329875714e-06,
      "loss": 1.1188,
      "step": 2347
    },
    {
      "epoch": 0.699342876821981,
      "grad_norm": 0.6385716199874878,
      "learning_rate": 7.759942280586174e-06,
      "loss": 1.1249,
      "step": 2348
    },
    {
      "epoch": 0.6996407230216497,
      "grad_norm": 0.6851052045822144,
      "learning_rate": 7.746954210899404e-06,
      "loss": 1.1167,
      "step": 2349
    },
    {
      "epoch": 0.6999385692213184,
      "grad_norm": 0.6785746216773987,
      "learning_rate": 7.733970143889531e-06,
      "loss": 1.1132,
      "step": 2350
    },
    {
      "epoch": 0.700236415420987,
      "grad_norm": 0.6543266177177429,
      "learning_rate": 7.72099010262356e-06,
      "loss": 1.1431,
      "step": 2351
    },
    {
      "epoch": 0.7005342616206557,
      "grad_norm": 0.6770211458206177,
      "learning_rate": 7.708014110161342e-06,
      "loss": 1.1373,
      "step": 2352
    },
    {
      "epoch": 0.7008321078203242,
      "grad_norm": 0.696246862411499,
      "learning_rate": 7.695042189555547e-06,
      "loss": 1.1261,
      "step": 2353
    },
    {
      "epoch": 0.7011299540199929,
      "grad_norm": 0.6616124510765076,
      "learning_rate": 7.682074363851598e-06,
      "loss": 1.1286,
      "step": 2354
    },
    {
      "epoch": 0.7014278002196616,
      "grad_norm": 0.6715471148490906,
      "learning_rate": 7.669110656087655e-06,
      "loss": 1.1258,
      "step": 2355
    },
    {
      "epoch": 0.7017256464193302,
      "grad_norm": 0.641290009021759,
      "learning_rate": 7.656151089294553e-06,
      "loss": 1.1381,
      "step": 2356
    },
    {
      "epoch": 0.7020234926189989,
      "grad_norm": 0.6536280512809753,
      "learning_rate": 7.643195686495773e-06,
      "loss": 1.1198,
      "step": 2357
    },
    {
      "epoch": 0.7023213388186675,
      "grad_norm": 0.6527756452560425,
      "learning_rate": 7.630244470707404e-06,
      "loss": 1.1348,
      "step": 2358
    },
    {
      "epoch": 0.7026191850183362,
      "grad_norm": 0.660698652267456,
      "learning_rate": 7.617297464938086e-06,
      "loss": 1.1125,
      "step": 2359
    },
    {
      "epoch": 0.7029170312180048,
      "grad_norm": 0.6370183825492859,
      "learning_rate": 7.604354692188986e-06,
      "loss": 1.1118,
      "step": 2360
    },
    {
      "epoch": 0.7032148774176734,
      "grad_norm": 0.6420510411262512,
      "learning_rate": 7.591416175453753e-06,
      "loss": 1.1237,
      "step": 2361
    },
    {
      "epoch": 0.7035127236173421,
      "grad_norm": 0.6494168043136597,
      "learning_rate": 7.578481937718469e-06,
      "loss": 1.1244,
      "step": 2362
    },
    {
      "epoch": 0.7038105698170107,
      "grad_norm": 0.6912646293640137,
      "learning_rate": 7.56555200196162e-06,
      "loss": 1.1437,
      "step": 2363
    },
    {
      "epoch": 0.7041084160166794,
      "grad_norm": 0.6695464849472046,
      "learning_rate": 7.552626391154047e-06,
      "loss": 1.1188,
      "step": 2364
    },
    {
      "epoch": 0.7044062622163481,
      "grad_norm": 0.6330601572990417,
      "learning_rate": 7.539705128258904e-06,
      "loss": 1.1082,
      "step": 2365
    },
    {
      "epoch": 0.7047041084160167,
      "grad_norm": 0.6537820100784302,
      "learning_rate": 7.526788236231622e-06,
      "loss": 1.1004,
      "step": 2366
    },
    {
      "epoch": 0.7050019546156854,
      "grad_norm": 0.6711965799331665,
      "learning_rate": 7.5138757380198714e-06,
      "loss": 1.135,
      "step": 2367
    },
    {
      "epoch": 0.7052998008153539,
      "grad_norm": 0.6476768255233765,
      "learning_rate": 7.500967656563513e-06,
      "loss": 1.1453,
      "step": 2368
    },
    {
      "epoch": 0.7055976470150226,
      "grad_norm": 0.6849035620689392,
      "learning_rate": 7.488064014794558e-06,
      "loss": 1.1268,
      "step": 2369
    },
    {
      "epoch": 0.7058954932146912,
      "grad_norm": 0.6721174716949463,
      "learning_rate": 7.475164835637141e-06,
      "loss": 1.146,
      "step": 2370
    },
    {
      "epoch": 0.7061933394143599,
      "grad_norm": 0.6894459128379822,
      "learning_rate": 7.462270142007455e-06,
      "loss": 1.1351,
      "step": 2371
    },
    {
      "epoch": 0.7064911856140286,
      "grad_norm": 0.6625561714172363,
      "learning_rate": 7.4493799568137335e-06,
      "loss": 1.1123,
      "step": 2372
    },
    {
      "epoch": 0.7067890318136972,
      "grad_norm": 0.6407567858695984,
      "learning_rate": 7.436494302956198e-06,
      "loss": 1.1176,
      "step": 2373
    },
    {
      "epoch": 0.7070868780133659,
      "grad_norm": 0.7022687196731567,
      "learning_rate": 7.423613203327016e-06,
      "loss": 1.1254,
      "step": 2374
    },
    {
      "epoch": 0.7073847242130344,
      "grad_norm": 0.6738529801368713,
      "learning_rate": 7.410736680810272e-06,
      "loss": 1.1141,
      "step": 2375
    },
    {
      "epoch": 0.7076825704127031,
      "grad_norm": 0.6907637715339661,
      "learning_rate": 7.39786475828191e-06,
      "loss": 1.1381,
      "step": 2376
    },
    {
      "epoch": 0.7079804166123718,
      "grad_norm": 0.6544174551963806,
      "learning_rate": 7.384997458609708e-06,
      "loss": 1.1364,
      "step": 2377
    },
    {
      "epoch": 0.7082782628120404,
      "grad_norm": 0.7052673697471619,
      "learning_rate": 7.372134804653232e-06,
      "loss": 1.1268,
      "step": 2378
    },
    {
      "epoch": 0.7085761090117091,
      "grad_norm": 0.6845062375068665,
      "learning_rate": 7.359276819263786e-06,
      "loss": 1.1191,
      "step": 2379
    },
    {
      "epoch": 0.7088739552113777,
      "grad_norm": 0.652862012386322,
      "learning_rate": 7.346423525284393e-06,
      "loss": 1.127,
      "step": 2380
    },
    {
      "epoch": 0.7091718014110464,
      "grad_norm": 0.6795114278793335,
      "learning_rate": 7.333574945549727e-06,
      "loss": 1.1205,
      "step": 2381
    },
    {
      "epoch": 0.7094696476107151,
      "grad_norm": 0.6889280080795288,
      "learning_rate": 7.320731102886102e-06,
      "loss": 1.1298,
      "step": 2382
    },
    {
      "epoch": 0.7097674938103836,
      "grad_norm": 0.6738985776901245,
      "learning_rate": 7.3078920201114045e-06,
      "loss": 1.1287,
      "step": 2383
    },
    {
      "epoch": 0.7100653400100523,
      "grad_norm": 0.6460881233215332,
      "learning_rate": 7.295057720035066e-06,
      "loss": 1.1255,
      "step": 2384
    },
    {
      "epoch": 0.7103631862097209,
      "grad_norm": 0.6541775465011597,
      "learning_rate": 7.28222822545803e-06,
      "loss": 1.1374,
      "step": 2385
    },
    {
      "epoch": 0.7106610324093896,
      "grad_norm": 0.6951769590377808,
      "learning_rate": 7.269403559172691e-06,
      "loss": 1.1364,
      "step": 2386
    },
    {
      "epoch": 0.7109588786090583,
      "grad_norm": 0.6582971215248108,
      "learning_rate": 7.2565837439628775e-06,
      "loss": 1.1336,
      "step": 2387
    },
    {
      "epoch": 0.7112567248087269,
      "grad_norm": 0.6246393918991089,
      "learning_rate": 7.24376880260379e-06,
      "loss": 1.1171,
      "step": 2388
    },
    {
      "epoch": 0.7115545710083956,
      "grad_norm": 0.6703428030014038,
      "learning_rate": 7.230958757861972e-06,
      "loss": 1.1411,
      "step": 2389
    },
    {
      "epoch": 0.7118524172080641,
      "grad_norm": 0.6569355130195618,
      "learning_rate": 7.218153632495277e-06,
      "loss": 1.1323,
      "step": 2390
    },
    {
      "epoch": 0.7121502634077328,
      "grad_norm": 0.6425116658210754,
      "learning_rate": 7.205353449252807e-06,
      "loss": 1.1208,
      "step": 2391
    },
    {
      "epoch": 0.7124481096074015,
      "grad_norm": 0.6701973676681519,
      "learning_rate": 7.192558230874887e-06,
      "loss": 1.1245,
      "step": 2392
    },
    {
      "epoch": 0.7127459558070701,
      "grad_norm": 0.66794753074646,
      "learning_rate": 7.179768000093027e-06,
      "loss": 1.1208,
      "step": 2393
    },
    {
      "epoch": 0.7130438020067388,
      "grad_norm": 0.6626831889152527,
      "learning_rate": 7.166982779629871e-06,
      "loss": 1.1231,
      "step": 2394
    },
    {
      "epoch": 0.7133416482064074,
      "grad_norm": 0.6558647155761719,
      "learning_rate": 7.154202592199166e-06,
      "loss": 1.1103,
      "step": 2395
    },
    {
      "epoch": 0.7136394944060761,
      "grad_norm": 0.6701839566230774,
      "learning_rate": 7.141427460505712e-06,
      "loss": 1.096,
      "step": 2396
    },
    {
      "epoch": 0.7139373406057448,
      "grad_norm": 0.6452605128288269,
      "learning_rate": 7.1286574072453315e-06,
      "loss": 1.118,
      "step": 2397
    },
    {
      "epoch": 0.7142351868054133,
      "grad_norm": 0.6568548083305359,
      "learning_rate": 7.115892455104827e-06,
      "loss": 1.133,
      "step": 2398
    },
    {
      "epoch": 0.714533033005082,
      "grad_norm": 0.676603376865387,
      "learning_rate": 7.103132626761929e-06,
      "loss": 1.1211,
      "step": 2399
    },
    {
      "epoch": 0.7148308792047506,
      "grad_norm": 0.6407927870750427,
      "learning_rate": 7.090377944885275e-06,
      "loss": 1.1234,
      "step": 2400
    },
    {
      "epoch": 0.7151287254044193,
      "grad_norm": 0.6747771501541138,
      "learning_rate": 7.077628432134352e-06,
      "loss": 1.1267,
      "step": 2401
    },
    {
      "epoch": 0.715426571604088,
      "grad_norm": 0.6743713021278381,
      "learning_rate": 7.064884111159474e-06,
      "loss": 1.1349,
      "step": 2402
    },
    {
      "epoch": 0.7157244178037566,
      "grad_norm": 0.6519574522972107,
      "learning_rate": 7.052145004601717e-06,
      "loss": 1.119,
      "step": 2403
    },
    {
      "epoch": 0.7160222640034253,
      "grad_norm": 0.691630482673645,
      "learning_rate": 7.0394111350929095e-06,
      "loss": 1.1222,
      "step": 2404
    },
    {
      "epoch": 0.7163201102030938,
      "grad_norm": 0.6507120728492737,
      "learning_rate": 7.026682525255563e-06,
      "loss": 1.1295,
      "step": 2405
    },
    {
      "epoch": 0.7166179564027625,
      "grad_norm": 0.670674204826355,
      "learning_rate": 7.013959197702851e-06,
      "loss": 1.1097,
      "step": 2406
    },
    {
      "epoch": 0.7169158026024312,
      "grad_norm": 0.6345235705375671,
      "learning_rate": 7.001241175038562e-06,
      "loss": 1.1324,
      "step": 2407
    },
    {
      "epoch": 0.7172136488020998,
      "grad_norm": 0.6213170289993286,
      "learning_rate": 6.988528479857057e-06,
      "loss": 1.1108,
      "step": 2408
    },
    {
      "epoch": 0.7175114950017685,
      "grad_norm": 0.6558605432510376,
      "learning_rate": 6.975821134743238e-06,
      "loss": 1.1181,
      "step": 2409
    },
    {
      "epoch": 0.7178093412014371,
      "grad_norm": 0.6505529284477234,
      "learning_rate": 6.963119162272498e-06,
      "loss": 1.1174,
      "step": 2410
    },
    {
      "epoch": 0.7181071874011058,
      "grad_norm": 0.6497592329978943,
      "learning_rate": 6.950422585010686e-06,
      "loss": 1.1257,
      "step": 2411
    },
    {
      "epoch": 0.7184050336007743,
      "grad_norm": 0.6436631083488464,
      "learning_rate": 6.937731425514066e-06,
      "loss": 1.1177,
      "step": 2412
    },
    {
      "epoch": 0.718702879800443,
      "grad_norm": 0.6626896858215332,
      "learning_rate": 6.925045706329277e-06,
      "loss": 1.1344,
      "step": 2413
    },
    {
      "epoch": 0.7190007260001117,
      "grad_norm": 0.6666689515113831,
      "learning_rate": 6.912365449993296e-06,
      "loss": 1.1102,
      "step": 2414
    },
    {
      "epoch": 0.7192985721997803,
      "grad_norm": 0.6456999182701111,
      "learning_rate": 6.899690679033389e-06,
      "loss": 1.1214,
      "step": 2415
    },
    {
      "epoch": 0.719596418399449,
      "grad_norm": 0.6803661584854126,
      "learning_rate": 6.887021415967081e-06,
      "loss": 1.1487,
      "step": 2416
    },
    {
      "epoch": 0.7198942645991177,
      "grad_norm": 0.6491440534591675,
      "learning_rate": 6.8743576833021135e-06,
      "loss": 1.1173,
      "step": 2417
    },
    {
      "epoch": 0.7201921107987863,
      "grad_norm": 0.6510524749755859,
      "learning_rate": 6.861699503536397e-06,
      "loss": 1.1203,
      "step": 2418
    },
    {
      "epoch": 0.720489956998455,
      "grad_norm": 0.6692467331886292,
      "learning_rate": 6.849046899157981e-06,
      "loss": 1.1363,
      "step": 2419
    },
    {
      "epoch": 0.7207878031981235,
      "grad_norm": 0.6331855058670044,
      "learning_rate": 6.83639989264501e-06,
      "loss": 1.121,
      "step": 2420
    },
    {
      "epoch": 0.7210856493977922,
      "grad_norm": 0.6723260879516602,
      "learning_rate": 6.823758506465682e-06,
      "loss": 1.1083,
      "step": 2421
    },
    {
      "epoch": 0.7213834955974608,
      "grad_norm": 0.6470404863357544,
      "learning_rate": 6.811122763078213e-06,
      "loss": 1.115,
      "step": 2422
    },
    {
      "epoch": 0.7216813417971295,
      "grad_norm": 0.658702552318573,
      "learning_rate": 6.7984926849307905e-06,
      "loss": 1.1171,
      "step": 2423
    },
    {
      "epoch": 0.7219791879967982,
      "grad_norm": 0.6641106009483337,
      "learning_rate": 6.785868294461538e-06,
      "loss": 1.124,
      "step": 2424
    },
    {
      "epoch": 0.7222770341964668,
      "grad_norm": 0.6548857688903809,
      "learning_rate": 6.773249614098479e-06,
      "loss": 1.128,
      "step": 2425
    },
    {
      "epoch": 0.7225748803961355,
      "grad_norm": 0.6588870882987976,
      "learning_rate": 6.760636666259485e-06,
      "loss": 1.1108,
      "step": 2426
    },
    {
      "epoch": 0.722872726595804,
      "grad_norm": 0.6510763764381409,
      "learning_rate": 6.7480294733522515e-06,
      "loss": 1.1048,
      "step": 2427
    },
    {
      "epoch": 0.7231705727954727,
      "grad_norm": 0.6822078227996826,
      "learning_rate": 6.735428057774243e-06,
      "loss": 1.1374,
      "step": 2428
    },
    {
      "epoch": 0.7234684189951414,
      "grad_norm": 0.6451256275177002,
      "learning_rate": 6.7228324419126616e-06,
      "loss": 1.1194,
      "step": 2429
    },
    {
      "epoch": 0.72376626519481,
      "grad_norm": 0.6505199074745178,
      "learning_rate": 6.710242648144413e-06,
      "loss": 1.1159,
      "step": 2430
    },
    {
      "epoch": 0.7240641113944787,
      "grad_norm": 0.661376416683197,
      "learning_rate": 6.697658698836047e-06,
      "loss": 1.1065,
      "step": 2431
    },
    {
      "epoch": 0.7243619575941473,
      "grad_norm": 0.6661626100540161,
      "learning_rate": 6.6850806163437486e-06,
      "loss": 1.1372,
      "step": 2432
    },
    {
      "epoch": 0.724659803793816,
      "grad_norm": 0.6728044152259827,
      "learning_rate": 6.672508423013255e-06,
      "loss": 1.141,
      "step": 2433
    },
    {
      "epoch": 0.7249576499934847,
      "grad_norm": 0.6580676436424255,
      "learning_rate": 6.659942141179865e-06,
      "loss": 1.1213,
      "step": 2434
    },
    {
      "epoch": 0.7252554961931532,
      "grad_norm": 0.6308066248893738,
      "learning_rate": 6.647381793168357e-06,
      "loss": 1.1303,
      "step": 2435
    },
    {
      "epoch": 0.7255533423928219,
      "grad_norm": 0.6446589827537537,
      "learning_rate": 6.634827401292981e-06,
      "loss": 1.1367,
      "step": 2436
    },
    {
      "epoch": 0.7258511885924905,
      "grad_norm": 0.6492798328399658,
      "learning_rate": 6.6222789878574e-06,
      "loss": 1.129,
      "step": 2437
    },
    {
      "epoch": 0.7261490347921592,
      "grad_norm": 0.652402400970459,
      "learning_rate": 6.609736575154653e-06,
      "loss": 1.1214,
      "step": 2438
    },
    {
      "epoch": 0.7264468809918279,
      "grad_norm": 0.6489751935005188,
      "learning_rate": 6.5972001854671245e-06,
      "loss": 1.1109,
      "step": 2439
    },
    {
      "epoch": 0.7267447271914965,
      "grad_norm": 0.6233725547790527,
      "learning_rate": 6.584669841066491e-06,
      "loss": 1.1276,
      "step": 2440
    },
    {
      "epoch": 0.7270425733911652,
      "grad_norm": 0.6752001047134399,
      "learning_rate": 6.572145564213701e-06,
      "loss": 1.1212,
      "step": 2441
    },
    {
      "epoch": 0.7273404195908337,
      "grad_norm": 0.658059298992157,
      "learning_rate": 6.559627377158911e-06,
      "loss": 1.1365,
      "step": 2442
    },
    {
      "epoch": 0.7276382657905024,
      "grad_norm": 0.63454270362854,
      "learning_rate": 6.547115302141465e-06,
      "loss": 1.1135,
      "step": 2443
    },
    {
      "epoch": 0.727936111990171,
      "grad_norm": 0.6496579051017761,
      "learning_rate": 6.534609361389847e-06,
      "loss": 1.1194,
      "step": 2444
    },
    {
      "epoch": 0.7282339581898397,
      "grad_norm": 0.6694057583808899,
      "learning_rate": 6.522109577121646e-06,
      "loss": 1.1281,
      "step": 2445
    },
    {
      "epoch": 0.7285318043895084,
      "grad_norm": 0.6600738167762756,
      "learning_rate": 6.5096159715435105e-06,
      "loss": 1.1298,
      "step": 2446
    },
    {
      "epoch": 0.728829650589177,
      "grad_norm": 0.6369644999504089,
      "learning_rate": 6.497128566851112e-06,
      "loss": 1.1354,
      "step": 2447
    },
    {
      "epoch": 0.7291274967888457,
      "grad_norm": 0.6564714312553406,
      "learning_rate": 6.4846473852291054e-06,
      "loss": 1.1153,
      "step": 2448
    },
    {
      "epoch": 0.7294253429885144,
      "grad_norm": 0.6412684321403503,
      "learning_rate": 6.472172448851095e-06,
      "loss": 1.1265,
      "step": 2449
    },
    {
      "epoch": 0.7297231891881829,
      "grad_norm": 0.6584504842758179,
      "learning_rate": 6.459703779879585e-06,
      "loss": 1.1249,
      "step": 2450
    },
    {
      "epoch": 0.7300210353878516,
      "grad_norm": 0.6557819843292236,
      "learning_rate": 6.447241400465942e-06,
      "loss": 1.1428,
      "step": 2451
    },
    {
      "epoch": 0.7303188815875202,
      "grad_norm": 0.6439695954322815,
      "learning_rate": 6.434785332750368e-06,
      "loss": 1.1195,
      "step": 2452
    },
    {
      "epoch": 0.7306167277871889,
      "grad_norm": 0.6753891110420227,
      "learning_rate": 6.422335598861846e-06,
      "loss": 1.1277,
      "step": 2453
    },
    {
      "epoch": 0.7309145739868576,
      "grad_norm": 0.6375412344932556,
      "learning_rate": 6.409892220918108e-06,
      "loss": 1.1196,
      "step": 2454
    },
    {
      "epoch": 0.7312124201865262,
      "grad_norm": 0.6691757440567017,
      "learning_rate": 6.397455221025595e-06,
      "loss": 1.1238,
      "step": 2455
    },
    {
      "epoch": 0.7315102663861949,
      "grad_norm": 0.6578329205513,
      "learning_rate": 6.385024621279411e-06,
      "loss": 1.117,
      "step": 2456
    },
    {
      "epoch": 0.7318081125858634,
      "grad_norm": 0.6711255311965942,
      "learning_rate": 6.3726004437633e-06,
      "loss": 1.1278,
      "step": 2457
    },
    {
      "epoch": 0.7321059587855321,
      "grad_norm": 0.672351062297821,
      "learning_rate": 6.360182710549587e-06,
      "loss": 1.1387,
      "step": 2458
    },
    {
      "epoch": 0.7324038049852007,
      "grad_norm": 0.6791733503341675,
      "learning_rate": 6.347771443699157e-06,
      "loss": 1.1192,
      "step": 2459
    },
    {
      "epoch": 0.7327016511848694,
      "grad_norm": 0.6581363081932068,
      "learning_rate": 6.335366665261401e-06,
      "loss": 1.1073,
      "step": 2460
    },
    {
      "epoch": 0.7329994973845381,
      "grad_norm": 0.637516975402832,
      "learning_rate": 6.322968397274181e-06,
      "loss": 1.1201,
      "step": 2461
    },
    {
      "epoch": 0.7332973435842067,
      "grad_norm": 0.6580083966255188,
      "learning_rate": 6.310576661763801e-06,
      "loss": 1.1403,
      "step": 2462
    },
    {
      "epoch": 0.7335951897838754,
      "grad_norm": 0.6366851925849915,
      "learning_rate": 6.298191480744951e-06,
      "loss": 1.1136,
      "step": 2463
    },
    {
      "epoch": 0.733893035983544,
      "grad_norm": 0.6497539281845093,
      "learning_rate": 6.285812876220682e-06,
      "loss": 1.1168,
      "step": 2464
    },
    {
      "epoch": 0.7341908821832126,
      "grad_norm": 0.674433171749115,
      "learning_rate": 6.273440870182365e-06,
      "loss": 1.1272,
      "step": 2465
    },
    {
      "epoch": 0.7344887283828813,
      "grad_norm": 0.6604278683662415,
      "learning_rate": 6.261075484609635e-06,
      "loss": 1.1253,
      "step": 2466
    },
    {
      "epoch": 0.7347865745825499,
      "grad_norm": 0.649469256401062,
      "learning_rate": 6.248716741470376e-06,
      "loss": 1.116,
      "step": 2467
    },
    {
      "epoch": 0.7350844207822186,
      "grad_norm": 0.6281045079231262,
      "learning_rate": 6.236364662720668e-06,
      "loss": 1.1179,
      "step": 2468
    },
    {
      "epoch": 0.7353822669818872,
      "grad_norm": 0.6268283128738403,
      "learning_rate": 6.224019270304757e-06,
      "loss": 1.1274,
      "step": 2469
    },
    {
      "epoch": 0.7356801131815559,
      "grad_norm": 0.6421892642974854,
      "learning_rate": 6.211680586155003e-06,
      "loss": 1.1212,
      "step": 2470
    },
    {
      "epoch": 0.7359779593812246,
      "grad_norm": 0.6579568386077881,
      "learning_rate": 6.199348632191848e-06,
      "loss": 1.1267,
      "step": 2471
    },
    {
      "epoch": 0.7362758055808931,
      "grad_norm": 0.6650598645210266,
      "learning_rate": 6.187023430323782e-06,
      "loss": 1.1333,
      "step": 2472
    },
    {
      "epoch": 0.7365736517805618,
      "grad_norm": 0.6522694826126099,
      "learning_rate": 6.174705002447301e-06,
      "loss": 1.1284,
      "step": 2473
    },
    {
      "epoch": 0.7368714979802304,
      "grad_norm": 0.6557116508483887,
      "learning_rate": 6.162393370446861e-06,
      "loss": 1.1274,
      "step": 2474
    },
    {
      "epoch": 0.7371693441798991,
      "grad_norm": 0.6358445286750793,
      "learning_rate": 6.150088556194843e-06,
      "loss": 1.1083,
      "step": 2475
    },
    {
      "epoch": 0.7374671903795678,
      "grad_norm": 0.623590350151062,
      "learning_rate": 6.1377905815515255e-06,
      "loss": 1.1172,
      "step": 2476
    },
    {
      "epoch": 0.7377650365792364,
      "grad_norm": 0.6461567282676697,
      "learning_rate": 6.125499468365028e-06,
      "loss": 1.1303,
      "step": 2477
    },
    {
      "epoch": 0.7380628827789051,
      "grad_norm": 0.6435955166816711,
      "learning_rate": 6.113215238471284e-06,
      "loss": 1.115,
      "step": 2478
    },
    {
      "epoch": 0.7383607289785737,
      "grad_norm": 0.6394869089126587,
      "learning_rate": 6.100937913693996e-06,
      "loss": 1.1389,
      "step": 2479
    },
    {
      "epoch": 0.7386585751782423,
      "grad_norm": 0.631363034248352,
      "learning_rate": 6.0886675158445976e-06,
      "loss": 1.1313,
      "step": 2480
    },
    {
      "epoch": 0.738956421377911,
      "grad_norm": 0.6259300708770752,
      "learning_rate": 6.076404066722224e-06,
      "loss": 1.0893,
      "step": 2481
    },
    {
      "epoch": 0.7392542675775796,
      "grad_norm": 0.6553505063056946,
      "learning_rate": 6.064147588113656e-06,
      "loss": 1.1323,
      "step": 2482
    },
    {
      "epoch": 0.7395521137772483,
      "grad_norm": 0.6537043452262878,
      "learning_rate": 6.051898101793294e-06,
      "loss": 1.113,
      "step": 2483
    },
    {
      "epoch": 0.7398499599769169,
      "grad_norm": 0.6394085884094238,
      "learning_rate": 6.039655629523122e-06,
      "loss": 1.1212,
      "step": 2484
    },
    {
      "epoch": 0.7401478061765856,
      "grad_norm": 0.6330388188362122,
      "learning_rate": 6.027420193052656e-06,
      "loss": 1.1103,
      "step": 2485
    },
    {
      "epoch": 0.7404456523762543,
      "grad_norm": 0.6646959781646729,
      "learning_rate": 6.0151918141189156e-06,
      "loss": 1.1271,
      "step": 2486
    },
    {
      "epoch": 0.7407434985759228,
      "grad_norm": 0.6774660348892212,
      "learning_rate": 6.002970514446382e-06,
      "loss": 1.1382,
      "step": 2487
    },
    {
      "epoch": 0.7410413447755915,
      "grad_norm": 0.6656184196472168,
      "learning_rate": 5.990756315746955e-06,
      "loss": 1.1305,
      "step": 2488
    },
    {
      "epoch": 0.7413391909752601,
      "grad_norm": 0.6606112122535706,
      "learning_rate": 5.978549239719929e-06,
      "loss": 1.1302,
      "step": 2489
    },
    {
      "epoch": 0.7416370371749288,
      "grad_norm": 0.6574504971504211,
      "learning_rate": 5.966349308051933e-06,
      "loss": 1.1426,
      "step": 2490
    },
    {
      "epoch": 0.7419348833745975,
      "grad_norm": 0.6344863176345825,
      "learning_rate": 5.954156542416915e-06,
      "loss": 1.1079,
      "step": 2491
    },
    {
      "epoch": 0.7422327295742661,
      "grad_norm": 0.6461744904518127,
      "learning_rate": 5.941970964476085e-06,
      "loss": 1.1264,
      "step": 2492
    },
    {
      "epoch": 0.7425305757739348,
      "grad_norm": 0.6462838053703308,
      "learning_rate": 5.929792595877881e-06,
      "loss": 1.1273,
      "step": 2493
    },
    {
      "epoch": 0.7428284219736033,
      "grad_norm": 0.6438243389129639,
      "learning_rate": 5.917621458257944e-06,
      "loss": 1.1142,
      "step": 2494
    },
    {
      "epoch": 0.743126268173272,
      "grad_norm": 0.63351970911026,
      "learning_rate": 5.905457573239055e-06,
      "loss": 1.1248,
      "step": 2495
    },
    {
      "epoch": 0.7434241143729406,
      "grad_norm": 0.645048201084137,
      "learning_rate": 5.8933009624311235e-06,
      "loss": 1.1081,
      "step": 2496
    },
    {
      "epoch": 0.7437219605726093,
      "grad_norm": 0.6271378993988037,
      "learning_rate": 5.881151647431125e-06,
      "loss": 1.1315,
      "step": 2497
    },
    {
      "epoch": 0.744019806772278,
      "grad_norm": 0.6755720973014832,
      "learning_rate": 5.869009649823079e-06,
      "loss": 1.1388,
      "step": 2498
    },
    {
      "epoch": 0.7443176529719466,
      "grad_norm": 0.6588972806930542,
      "learning_rate": 5.856874991178008e-06,
      "loss": 1.1228,
      "step": 2499
    },
    {
      "epoch": 0.7446154991716153,
      "grad_norm": 0.6581892967224121,
      "learning_rate": 5.844747693053889e-06,
      "loss": 1.1234,
      "step": 2500
    },
    {
      "epoch": 0.744913345371284,
      "grad_norm": 0.675911009311676,
      "learning_rate": 5.832627776995625e-06,
      "loss": 1.1195,
      "step": 2501
    },
    {
      "epoch": 0.7452111915709525,
      "grad_norm": 0.663588285446167,
      "learning_rate": 5.820515264535011e-06,
      "loss": 1.1117,
      "step": 2502
    },
    {
      "epoch": 0.7455090377706212,
      "grad_norm": 0.6592284440994263,
      "learning_rate": 5.808410177190677e-06,
      "loss": 1.1023,
      "step": 2503
    },
    {
      "epoch": 0.7458068839702898,
      "grad_norm": 0.6150014996528625,
      "learning_rate": 5.796312536468074e-06,
      "loss": 1.0933,
      "step": 2504
    },
    {
      "epoch": 0.7461047301699585,
      "grad_norm": 0.6301830410957336,
      "learning_rate": 5.784222363859418e-06,
      "loss": 1.1151,
      "step": 2505
    },
    {
      "epoch": 0.7464025763696271,
      "grad_norm": 0.6780192852020264,
      "learning_rate": 5.772139680843652e-06,
      "loss": 1.1281,
      "step": 2506
    },
    {
      "epoch": 0.7467004225692958,
      "grad_norm": 0.6618151664733887,
      "learning_rate": 5.760064508886423e-06,
      "loss": 1.1321,
      "step": 2507
    },
    {
      "epoch": 0.7469982687689645,
      "grad_norm": 0.6129297614097595,
      "learning_rate": 5.747996869440029e-06,
      "loss": 1.1174,
      "step": 2508
    },
    {
      "epoch": 0.747296114968633,
      "grad_norm": 0.6798166632652283,
      "learning_rate": 5.735936783943392e-06,
      "loss": 1.1171,
      "step": 2509
    },
    {
      "epoch": 0.7475939611683017,
      "grad_norm": 0.6655552387237549,
      "learning_rate": 5.723884273822e-06,
      "loss": 1.1107,
      "step": 2510
    },
    {
      "epoch": 0.7478918073679703,
      "grad_norm": 0.6476085782051086,
      "learning_rate": 5.711839360487897e-06,
      "loss": 1.1299,
      "step": 2511
    },
    {
      "epoch": 0.748189653567639,
      "grad_norm": 0.6329128742218018,
      "learning_rate": 5.699802065339626e-06,
      "loss": 1.1167,
      "step": 2512
    },
    {
      "epoch": 0.7484874997673077,
      "grad_norm": 0.6603522300720215,
      "learning_rate": 5.687772409762193e-06,
      "loss": 1.1262,
      "step": 2513
    },
    {
      "epoch": 0.7487853459669763,
      "grad_norm": 0.6658760905265808,
      "learning_rate": 5.675750415127032e-06,
      "loss": 1.1207,
      "step": 2514
    },
    {
      "epoch": 0.749083192166645,
      "grad_norm": 0.6558886766433716,
      "learning_rate": 5.663736102791979e-06,
      "loss": 1.1167,
      "step": 2515
    },
    {
      "epoch": 0.7493810383663136,
      "grad_norm": 0.659781277179718,
      "learning_rate": 5.651729494101201e-06,
      "loss": 1.1414,
      "step": 2516
    },
    {
      "epoch": 0.7496788845659822,
      "grad_norm": 0.6265398859977722,
      "learning_rate": 5.6397306103851944e-06,
      "loss": 1.1167,
      "step": 2517
    },
    {
      "epoch": 0.7499767307656509,
      "grad_norm": 0.6758785843849182,
      "learning_rate": 5.627739472960727e-06,
      "loss": 1.1082,
      "step": 2518
    },
    {
      "epoch": 0.7502745769653195,
      "grad_norm": 0.6608585715293884,
      "learning_rate": 5.615756103130809e-06,
      "loss": 1.1286,
      "step": 2519
    },
    {
      "epoch": 0.7505724231649882,
      "grad_norm": 0.6259558796882629,
      "learning_rate": 5.6037805221846395e-06,
      "loss": 1.1079,
      "step": 2520
    },
    {
      "epoch": 0.7508702693646568,
      "grad_norm": 0.6467676162719727,
      "learning_rate": 5.591812751397592e-06,
      "loss": 1.1307,
      "step": 2521
    },
    {
      "epoch": 0.7511681155643255,
      "grad_norm": 0.6549900770187378,
      "learning_rate": 5.579852812031165e-06,
      "loss": 1.1196,
      "step": 2522
    },
    {
      "epoch": 0.7514659617639942,
      "grad_norm": 0.6461799144744873,
      "learning_rate": 5.56790072533293e-06,
      "loss": 1.1072,
      "step": 2523
    },
    {
      "epoch": 0.7517638079636627,
      "grad_norm": 0.6516606211662292,
      "learning_rate": 5.555956512536522e-06,
      "loss": 1.1227,
      "step": 2524
    },
    {
      "epoch": 0.7520616541633314,
      "grad_norm": 0.6406022310256958,
      "learning_rate": 5.544020194861587e-06,
      "loss": 1.1176,
      "step": 2525
    },
    {
      "epoch": 0.752359500363,
      "grad_norm": 0.6313644647598267,
      "learning_rate": 5.5320917935137315e-06,
      "loss": 1.114,
      "step": 2526
    },
    {
      "epoch": 0.7526573465626687,
      "grad_norm": 0.7081498503684998,
      "learning_rate": 5.520171329684514e-06,
      "loss": 1.1313,
      "step": 2527
    },
    {
      "epoch": 0.7529551927623374,
      "grad_norm": 0.6540417075157166,
      "learning_rate": 5.508258824551387e-06,
      "loss": 1.1135,
      "step": 2528
    },
    {
      "epoch": 0.753253038962006,
      "grad_norm": 0.6520769596099854,
      "learning_rate": 5.496354299277652e-06,
      "loss": 1.1216,
      "step": 2529
    },
    {
      "epoch": 0.7535508851616747,
      "grad_norm": 0.6553327441215515,
      "learning_rate": 5.484457775012451e-06,
      "loss": 1.1296,
      "step": 2530
    },
    {
      "epoch": 0.7538487313613433,
      "grad_norm": 0.6505500674247742,
      "learning_rate": 5.4725692728906994e-06,
      "loss": 1.1166,
      "step": 2531
    },
    {
      "epoch": 0.7541465775610119,
      "grad_norm": 0.6597558856010437,
      "learning_rate": 5.460688814033075e-06,
      "loss": 1.1101,
      "step": 2532
    },
    {
      "epoch": 0.7544444237606805,
      "grad_norm": 0.6336129903793335,
      "learning_rate": 5.448816419545948e-06,
      "loss": 1.1182,
      "step": 2533
    },
    {
      "epoch": 0.7547422699603492,
      "grad_norm": 0.6562846302986145,
      "learning_rate": 5.43695211052137e-06,
      "loss": 1.0862,
      "step": 2534
    },
    {
      "epoch": 0.7550401161600179,
      "grad_norm": 0.6787288784980774,
      "learning_rate": 5.4250959080370305e-06,
      "loss": 1.1191,
      "step": 2535
    },
    {
      "epoch": 0.7553379623596865,
      "grad_norm": 0.6476746797561646,
      "learning_rate": 5.413247833156219e-06,
      "loss": 1.1319,
      "step": 2536
    },
    {
      "epoch": 0.7556358085593552,
      "grad_norm": 0.639701247215271,
      "learning_rate": 5.401407906927776e-06,
      "loss": 1.1097,
      "step": 2537
    },
    {
      "epoch": 0.7559336547590239,
      "grad_norm": 0.6820856332778931,
      "learning_rate": 5.389576150386075e-06,
      "loss": 1.1432,
      "step": 2538
    },
    {
      "epoch": 0.7562315009586924,
      "grad_norm": 0.6809478402137756,
      "learning_rate": 5.377752584550977e-06,
      "loss": 1.1404,
      "step": 2539
    },
    {
      "epoch": 0.7565293471583611,
      "grad_norm": 0.6640024781227112,
      "learning_rate": 5.365937230427778e-06,
      "loss": 1.1304,
      "step": 2540
    },
    {
      "epoch": 0.7568271933580297,
      "grad_norm": 0.6428071856498718,
      "learning_rate": 5.354130109007198e-06,
      "loss": 1.1109,
      "step": 2541
    },
    {
      "epoch": 0.7571250395576984,
      "grad_norm": 0.6499603986740112,
      "learning_rate": 5.342331241265332e-06,
      "loss": 1.1409,
      "step": 2542
    },
    {
      "epoch": 0.757422885757367,
      "grad_norm": 0.6639237999916077,
      "learning_rate": 5.330540648163598e-06,
      "loss": 1.1258,
      "step": 2543
    },
    {
      "epoch": 0.7577207319570357,
      "grad_norm": 0.6549901366233826,
      "learning_rate": 5.3187583506487295e-06,
      "loss": 1.1281,
      "step": 2544
    },
    {
      "epoch": 0.7580185781567044,
      "grad_norm": 0.66259765625,
      "learning_rate": 5.306984369652712e-06,
      "loss": 1.1165,
      "step": 2545
    },
    {
      "epoch": 0.758316424356373,
      "grad_norm": 0.654653012752533,
      "learning_rate": 5.295218726092768e-06,
      "loss": 1.1377,
      "step": 2546
    },
    {
      "epoch": 0.7586142705560416,
      "grad_norm": 0.6512067914009094,
      "learning_rate": 5.283461440871288e-06,
      "loss": 1.1101,
      "step": 2547
    },
    {
      "epoch": 0.7589121167557102,
      "grad_norm": 0.6499033570289612,
      "learning_rate": 5.271712534875831e-06,
      "loss": 1.1344,
      "step": 2548
    },
    {
      "epoch": 0.7592099629553789,
      "grad_norm": 0.6413191556930542,
      "learning_rate": 5.259972028979068e-06,
      "loss": 1.128,
      "step": 2549
    },
    {
      "epoch": 0.7595078091550476,
      "grad_norm": 0.6522189974784851,
      "learning_rate": 5.248239944038736e-06,
      "loss": 1.1323,
      "step": 2550
    },
    {
      "epoch": 0.7598056553547162,
      "grad_norm": 0.6561805009841919,
      "learning_rate": 5.2365163008976195e-06,
      "loss": 1.125,
      "step": 2551
    },
    {
      "epoch": 0.7601035015543849,
      "grad_norm": 0.6276010870933533,
      "learning_rate": 5.224801120383509e-06,
      "loss": 1.1394,
      "step": 2552
    },
    {
      "epoch": 0.7604013477540535,
      "grad_norm": 0.6578109264373779,
      "learning_rate": 5.213094423309151e-06,
      "loss": 1.116,
      "step": 2553
    },
    {
      "epoch": 0.7606991939537221,
      "grad_norm": 0.6285254955291748,
      "learning_rate": 5.201396230472227e-06,
      "loss": 1.1198,
      "step": 2554
    },
    {
      "epoch": 0.7609970401533908,
      "grad_norm": 0.6422551870346069,
      "learning_rate": 5.189706562655314e-06,
      "loss": 1.1198,
      "step": 2555
    },
    {
      "epoch": 0.7612948863530594,
      "grad_norm": 0.6566483974456787,
      "learning_rate": 5.1780254406258305e-06,
      "loss": 1.1308,
      "step": 2556
    },
    {
      "epoch": 0.7615927325527281,
      "grad_norm": 0.6529741883277893,
      "learning_rate": 5.1663528851360245e-06,
      "loss": 1.1302,
      "step": 2557
    },
    {
      "epoch": 0.7618905787523967,
      "grad_norm": 0.6382930278778076,
      "learning_rate": 5.154688916922922e-06,
      "loss": 1.1136,
      "step": 2558
    },
    {
      "epoch": 0.7621884249520654,
      "grad_norm": 0.6377841830253601,
      "learning_rate": 5.1430335567082965e-06,
      "loss": 1.1156,
      "step": 2559
    },
    {
      "epoch": 0.7624862711517341,
      "grad_norm": 0.6446092128753662,
      "learning_rate": 5.131386825198619e-06,
      "loss": 1.1236,
      "step": 2560
    },
    {
      "epoch": 0.7627841173514027,
      "grad_norm": 0.6068905591964722,
      "learning_rate": 5.119748743085038e-06,
      "loss": 1.108,
      "step": 2561
    },
    {
      "epoch": 0.7630819635510713,
      "grad_norm": 0.6301131844520569,
      "learning_rate": 5.1081193310433395e-06,
      "loss": 1.1228,
      "step": 2562
    },
    {
      "epoch": 0.7633798097507399,
      "grad_norm": 0.6675803661346436,
      "learning_rate": 5.096498609733898e-06,
      "loss": 1.1204,
      "step": 2563
    },
    {
      "epoch": 0.7636776559504086,
      "grad_norm": 0.644855797290802,
      "learning_rate": 5.08488659980165e-06,
      "loss": 1.1225,
      "step": 2564
    },
    {
      "epoch": 0.7639755021500773,
      "grad_norm": 0.6491459608078003,
      "learning_rate": 5.073283321876068e-06,
      "loss": 1.1322,
      "step": 2565
    },
    {
      "epoch": 0.7642733483497459,
      "grad_norm": 0.6436959505081177,
      "learning_rate": 5.061688796571095e-06,
      "loss": 1.1346,
      "step": 2566
    },
    {
      "epoch": 0.7645711945494146,
      "grad_norm": 0.6464740037918091,
      "learning_rate": 5.050103044485126e-06,
      "loss": 1.1074,
      "step": 2567
    },
    {
      "epoch": 0.7648690407490832,
      "grad_norm": 0.6588334441184998,
      "learning_rate": 5.038526086200981e-06,
      "loss": 1.1051,
      "step": 2568
    },
    {
      "epoch": 0.7651668869487518,
      "grad_norm": 0.6370975971221924,
      "learning_rate": 5.0269579422858546e-06,
      "loss": 1.1202,
      "step": 2569
    },
    {
      "epoch": 0.7654647331484205,
      "grad_norm": 0.6383762955665588,
      "learning_rate": 5.01539863329127e-06,
      "loss": 1.1212,
      "step": 2570
    },
    {
      "epoch": 0.7657625793480891,
      "grad_norm": 0.6825690269470215,
      "learning_rate": 5.003848179753068e-06,
      "loss": 1.1252,
      "step": 2571
    },
    {
      "epoch": 0.7660604255477578,
      "grad_norm": 0.6495948433876038,
      "learning_rate": 4.992306602191351e-06,
      "loss": 1.1151,
      "step": 2572
    },
    {
      "epoch": 0.7663582717474264,
      "grad_norm": 0.6384968161582947,
      "learning_rate": 4.98077392111046e-06,
      "loss": 1.1104,
      "step": 2573
    },
    {
      "epoch": 0.7666561179470951,
      "grad_norm": 0.6416631937026978,
      "learning_rate": 4.9692501569989145e-06,
      "loss": 1.1085,
      "step": 2574
    },
    {
      "epoch": 0.7669539641467638,
      "grad_norm": 0.64227294921875,
      "learning_rate": 4.95773533032941e-06,
      "loss": 1.1207,
      "step": 2575
    },
    {
      "epoch": 0.7672518103464323,
      "grad_norm": 0.6304708123207092,
      "learning_rate": 4.946229461558759e-06,
      "loss": 1.1238,
      "step": 2576
    },
    {
      "epoch": 0.767549656546101,
      "grad_norm": 0.6152022480964661,
      "learning_rate": 4.93473257112785e-06,
      "loss": 1.1103,
      "step": 2577
    },
    {
      "epoch": 0.7678475027457696,
      "grad_norm": 0.6523952484130859,
      "learning_rate": 4.923244679461634e-06,
      "loss": 1.1251,
      "step": 2578
    },
    {
      "epoch": 0.7681453489454383,
      "grad_norm": 0.6700926423072815,
      "learning_rate": 4.911765806969071e-06,
      "loss": 1.125,
      "step": 2579
    },
    {
      "epoch": 0.768443195145107,
      "grad_norm": 0.6404693126678467,
      "learning_rate": 4.90029597404309e-06,
      "loss": 1.1251,
      "step": 2580
    },
    {
      "epoch": 0.7687410413447756,
      "grad_norm": 0.6581672430038452,
      "learning_rate": 4.888835201060571e-06,
      "loss": 1.118,
      "step": 2581
    },
    {
      "epoch": 0.7690388875444443,
      "grad_norm": 0.6518725752830505,
      "learning_rate": 4.8773835083823e-06,
      "loss": 1.1055,
      "step": 2582
    },
    {
      "epoch": 0.7693367337441129,
      "grad_norm": 0.6362202763557434,
      "learning_rate": 4.8659409163529155e-06,
      "loss": 1.1205,
      "step": 2583
    },
    {
      "epoch": 0.7696345799437815,
      "grad_norm": 0.6340486407279968,
      "learning_rate": 4.854507445300905e-06,
      "loss": 1.1125,
      "step": 2584
    },
    {
      "epoch": 0.7699324261434501,
      "grad_norm": 0.650761604309082,
      "learning_rate": 4.843083115538544e-06,
      "loss": 1.1237,
      "step": 2585
    },
    {
      "epoch": 0.7702302723431188,
      "grad_norm": 0.6580442786216736,
      "learning_rate": 4.831667947361874e-06,
      "loss": 1.1229,
      "step": 2586
    },
    {
      "epoch": 0.7705281185427875,
      "grad_norm": 0.6613978743553162,
      "learning_rate": 4.820261961050648e-06,
      "loss": 1.1112,
      "step": 2587
    },
    {
      "epoch": 0.7708259647424561,
      "grad_norm": 0.6609376668930054,
      "learning_rate": 4.80886517686832e-06,
      "loss": 1.1206,
      "step": 2588
    },
    {
      "epoch": 0.7711238109421248,
      "grad_norm": 0.6205573678016663,
      "learning_rate": 4.797477615061992e-06,
      "loss": 1.1034,
      "step": 2589
    },
    {
      "epoch": 0.7714216571417934,
      "grad_norm": 0.6581565141677856,
      "learning_rate": 4.786099295862374e-06,
      "loss": 1.1381,
      "step": 2590
    },
    {
      "epoch": 0.771719503341462,
      "grad_norm": 0.6357449293136597,
      "learning_rate": 4.774730239483767e-06,
      "loss": 1.1061,
      "step": 2591
    },
    {
      "epoch": 0.7720173495411307,
      "grad_norm": 0.6518653035163879,
      "learning_rate": 4.7633704661240135e-06,
      "loss": 1.1104,
      "step": 2592
    },
    {
      "epoch": 0.7723151957407993,
      "grad_norm": 0.6604530811309814,
      "learning_rate": 4.752019995964456e-06,
      "loss": 1.1471,
      "step": 2593
    },
    {
      "epoch": 0.772613041940468,
      "grad_norm": 0.6559156775474548,
      "learning_rate": 4.740678849169917e-06,
      "loss": 1.1122,
      "step": 2594
    },
    {
      "epoch": 0.7729108881401366,
      "grad_norm": 0.6490145325660706,
      "learning_rate": 4.729347045888657e-06,
      "loss": 1.1124,
      "step": 2595
    },
    {
      "epoch": 0.7732087343398053,
      "grad_norm": 0.6466209292411804,
      "learning_rate": 4.718024606252335e-06,
      "loss": 1.1167,
      "step": 2596
    },
    {
      "epoch": 0.773506580539474,
      "grad_norm": 0.6479471325874329,
      "learning_rate": 4.7067115503759665e-06,
      "loss": 1.099,
      "step": 2597
    },
    {
      "epoch": 0.7738044267391426,
      "grad_norm": 0.6590579152107239,
      "learning_rate": 4.695407898357909e-06,
      "loss": 1.1086,
      "step": 2598
    },
    {
      "epoch": 0.7741022729388112,
      "grad_norm": 0.6334704160690308,
      "learning_rate": 4.684113670279811e-06,
      "loss": 1.1035,
      "step": 2599
    },
    {
      "epoch": 0.7744001191384798,
      "grad_norm": 0.6378445625305176,
      "learning_rate": 4.6728288862065705e-06,
      "loss": 1.1157,
      "step": 2600
    },
    {
      "epoch": 0.7746979653381485,
      "grad_norm": 0.6320429444313049,
      "learning_rate": 4.661553566186311e-06,
      "loss": 1.1116,
      "step": 2601
    },
    {
      "epoch": 0.7749958115378172,
      "grad_norm": 0.660942554473877,
      "learning_rate": 4.650287730250347e-06,
      "loss": 1.121,
      "step": 2602
    },
    {
      "epoch": 0.7752936577374858,
      "grad_norm": 0.629319965839386,
      "learning_rate": 4.639031398413146e-06,
      "loss": 1.1323,
      "step": 2603
    },
    {
      "epoch": 0.7755915039371545,
      "grad_norm": 0.665974497795105,
      "learning_rate": 4.627784590672278e-06,
      "loss": 1.1423,
      "step": 2604
    },
    {
      "epoch": 0.7758893501368231,
      "grad_norm": 0.6397601962089539,
      "learning_rate": 4.616547327008405e-06,
      "loss": 1.1225,
      "step": 2605
    },
    {
      "epoch": 0.7761871963364917,
      "grad_norm": 0.6469012498855591,
      "learning_rate": 4.605319627385233e-06,
      "loss": 1.1039,
      "step": 2606
    },
    {
      "epoch": 0.7764850425361604,
      "grad_norm": 0.6670868396759033,
      "learning_rate": 4.594101511749466e-06,
      "loss": 1.1282,
      "step": 2607
    },
    {
      "epoch": 0.776782888735829,
      "grad_norm": 0.643539309501648,
      "learning_rate": 4.582893000030793e-06,
      "loss": 1.0886,
      "step": 2608
    },
    {
      "epoch": 0.7770807349354977,
      "grad_norm": 0.6341370940208435,
      "learning_rate": 4.571694112141835e-06,
      "loss": 1.1279,
      "step": 2609
    },
    {
      "epoch": 0.7773785811351663,
      "grad_norm": 0.6564409136772156,
      "learning_rate": 4.560504867978124e-06,
      "loss": 1.1152,
      "step": 2610
    },
    {
      "epoch": 0.777676427334835,
      "grad_norm": 0.6988517045974731,
      "learning_rate": 4.549325287418042e-06,
      "loss": 1.1282,
      "step": 2611
    },
    {
      "epoch": 0.7779742735345037,
      "grad_norm": 0.650214433670044,
      "learning_rate": 4.538155390322819e-06,
      "loss": 1.0892,
      "step": 2612
    },
    {
      "epoch": 0.7782721197341723,
      "grad_norm": 0.6532634496688843,
      "learning_rate": 4.526995196536481e-06,
      "loss": 1.1101,
      "step": 2613
    },
    {
      "epoch": 0.7785699659338409,
      "grad_norm": 0.6516757011413574,
      "learning_rate": 4.515844725885803e-06,
      "loss": 1.1138,
      "step": 2614
    },
    {
      "epoch": 0.7788678121335095,
      "grad_norm": 0.677932620048523,
      "learning_rate": 4.5047039981802975e-06,
      "loss": 1.1238,
      "step": 2615
    },
    {
      "epoch": 0.7791656583331782,
      "grad_norm": 0.6460375785827637,
      "learning_rate": 4.493573033212169e-06,
      "loss": 1.1213,
      "step": 2616
    },
    {
      "epoch": 0.7794635045328469,
      "grad_norm": 0.6288262605667114,
      "learning_rate": 4.482451850756264e-06,
      "loss": 1.1276,
      "step": 2617
    },
    {
      "epoch": 0.7797613507325155,
      "grad_norm": 0.6383118629455566,
      "learning_rate": 4.471340470570067e-06,
      "loss": 1.1016,
      "step": 2618
    },
    {
      "epoch": 0.7800591969321842,
      "grad_norm": 0.6307514905929565,
      "learning_rate": 4.4602389123936406e-06,
      "loss": 1.1242,
      "step": 2619
    },
    {
      "epoch": 0.7803570431318528,
      "grad_norm": 0.6557974219322205,
      "learning_rate": 4.449147195949592e-06,
      "loss": 1.1137,
      "step": 2620
    },
    {
      "epoch": 0.7806548893315214,
      "grad_norm": 0.6584435701370239,
      "learning_rate": 4.438065340943054e-06,
      "loss": 1.1297,
      "step": 2621
    },
    {
      "epoch": 0.78095273553119,
      "grad_norm": 0.652807891368866,
      "learning_rate": 4.426993367061635e-06,
      "loss": 1.111,
      "step": 2622
    },
    {
      "epoch": 0.7812505817308587,
      "grad_norm": 0.6330335736274719,
      "learning_rate": 4.415931293975394e-06,
      "loss": 1.1164,
      "step": 2623
    },
    {
      "epoch": 0.7815484279305274,
      "grad_norm": 0.6485841274261475,
      "learning_rate": 4.404879141336788e-06,
      "loss": 1.1265,
      "step": 2624
    },
    {
      "epoch": 0.781846274130196,
      "grad_norm": 0.6486736536026001,
      "learning_rate": 4.393836928780664e-06,
      "loss": 1.1008,
      "step": 2625
    },
    {
      "epoch": 0.7821441203298647,
      "grad_norm": 0.6311295628547668,
      "learning_rate": 4.3828046759242035e-06,
      "loss": 1.114,
      "step": 2626
    },
    {
      "epoch": 0.7824419665295334,
      "grad_norm": 0.6363070011138916,
      "learning_rate": 4.371782402366892e-06,
      "loss": 1.1251,
      "step": 2627
    },
    {
      "epoch": 0.782739812729202,
      "grad_norm": 0.6627604961395264,
      "learning_rate": 4.360770127690488e-06,
      "loss": 1.1278,
      "step": 2628
    },
    {
      "epoch": 0.7830376589288706,
      "grad_norm": 0.6399460434913635,
      "learning_rate": 4.349767871458993e-06,
      "loss": 1.122,
      "step": 2629
    },
    {
      "epoch": 0.7833355051285392,
      "grad_norm": 0.6483947038650513,
      "learning_rate": 4.338775653218595e-06,
      "loss": 1.1346,
      "step": 2630
    },
    {
      "epoch": 0.7836333513282079,
      "grad_norm": 0.6400676965713501,
      "learning_rate": 4.327793492497664e-06,
      "loss": 1.1152,
      "step": 2631
    },
    {
      "epoch": 0.7839311975278765,
      "grad_norm": 0.663577139377594,
      "learning_rate": 4.316821408806694e-06,
      "loss": 1.1141,
      "step": 2632
    },
    {
      "epoch": 0.7842290437275452,
      "grad_norm": 0.6395473480224609,
      "learning_rate": 4.305859421638282e-06,
      "loss": 1.1143,
      "step": 2633
    },
    {
      "epoch": 0.7845268899272139,
      "grad_norm": 0.64128577709198,
      "learning_rate": 4.294907550467083e-06,
      "loss": 1.1103,
      "step": 2634
    },
    {
      "epoch": 0.7848247361268825,
      "grad_norm": 0.6428948640823364,
      "learning_rate": 4.283965814749774e-06,
      "loss": 1.1176,
      "step": 2635
    },
    {
      "epoch": 0.7851225823265511,
      "grad_norm": 0.6589243412017822,
      "learning_rate": 4.273034233925041e-06,
      "loss": 1.1237,
      "step": 2636
    },
    {
      "epoch": 0.7854204285262197,
      "grad_norm": 0.646443784236908,
      "learning_rate": 4.262112827413525e-06,
      "loss": 1.1103,
      "step": 2637
    },
    {
      "epoch": 0.7857182747258884,
      "grad_norm": 0.6383097171783447,
      "learning_rate": 4.2512016146177794e-06,
      "loss": 1.1042,
      "step": 2638
    },
    {
      "epoch": 0.7860161209255571,
      "grad_norm": 0.6303218007087708,
      "learning_rate": 4.24030061492226e-06,
      "loss": 1.1055,
      "step": 2639
    },
    {
      "epoch": 0.7863139671252257,
      "grad_norm": 0.6458702087402344,
      "learning_rate": 4.2294098476932824e-06,
      "loss": 1.1125,
      "step": 2640
    },
    {
      "epoch": 0.7866118133248944,
      "grad_norm": 0.6410305500030518,
      "learning_rate": 4.218529332278966e-06,
      "loss": 1.119,
      "step": 2641
    },
    {
      "epoch": 0.786909659524563,
      "grad_norm": 0.624588131904602,
      "learning_rate": 4.20765908800923e-06,
      "loss": 1.1213,
      "step": 2642
    },
    {
      "epoch": 0.7872075057242317,
      "grad_norm": 0.6483225226402283,
      "learning_rate": 4.1967991341957525e-06,
      "loss": 1.1187,
      "step": 2643
    },
    {
      "epoch": 0.7875053519239003,
      "grad_norm": 0.6634080410003662,
      "learning_rate": 4.18594949013191e-06,
      "loss": 1.1415,
      "step": 2644
    },
    {
      "epoch": 0.7878031981235689,
      "grad_norm": 0.6645361185073853,
      "learning_rate": 4.175110175092778e-06,
      "loss": 1.1019,
      "step": 2645
    },
    {
      "epoch": 0.7881010443232376,
      "grad_norm": 0.645251452922821,
      "learning_rate": 4.164281208335083e-06,
      "loss": 1.1148,
      "step": 2646
    },
    {
      "epoch": 0.7883988905229062,
      "grad_norm": 0.6608096361160278,
      "learning_rate": 4.1534626090971576e-06,
      "loss": 1.1401,
      "step": 2647
    },
    {
      "epoch": 0.7886967367225749,
      "grad_norm": 0.6601690053939819,
      "learning_rate": 4.1426543965989195e-06,
      "loss": 1.123,
      "step": 2648
    },
    {
      "epoch": 0.7889945829222436,
      "grad_norm": 0.6269056797027588,
      "learning_rate": 4.131856590041837e-06,
      "loss": 1.1099,
      "step": 2649
    },
    {
      "epoch": 0.7892924291219122,
      "grad_norm": 0.6598504185676575,
      "learning_rate": 4.121069208608894e-06,
      "loss": 1.1203,
      "step": 2650
    },
    {
      "epoch": 0.7895902753215808,
      "grad_norm": 0.6615498065948486,
      "learning_rate": 4.110292271464541e-06,
      "loss": 1.0992,
      "step": 2651
    },
    {
      "epoch": 0.7898881215212494,
      "grad_norm": 0.6454700231552124,
      "learning_rate": 4.099525797754686e-06,
      "loss": 1.1255,
      "step": 2652
    },
    {
      "epoch": 0.7901859677209181,
      "grad_norm": 0.6393318176269531,
      "learning_rate": 4.088769806606644e-06,
      "loss": 1.1203,
      "step": 2653
    },
    {
      "epoch": 0.7904838139205868,
      "grad_norm": 0.6227574944496155,
      "learning_rate": 4.078024317129102e-06,
      "loss": 1.1154,
      "step": 2654
    },
    {
      "epoch": 0.7907816601202554,
      "grad_norm": 0.6515089273452759,
      "learning_rate": 4.067289348412099e-06,
      "loss": 1.1156,
      "step": 2655
    },
    {
      "epoch": 0.7910795063199241,
      "grad_norm": 0.6524357795715332,
      "learning_rate": 4.0565649195269804e-06,
      "loss": 1.1122,
      "step": 2656
    },
    {
      "epoch": 0.7913773525195927,
      "grad_norm": 0.6529024243354797,
      "learning_rate": 4.04585104952636e-06,
      "loss": 1.1199,
      "step": 2657
    },
    {
      "epoch": 0.7916751987192613,
      "grad_norm": 0.6678121089935303,
      "learning_rate": 4.035147757444102e-06,
      "loss": 1.1209,
      "step": 2658
    },
    {
      "epoch": 0.79197304491893,
      "grad_norm": 0.6546996235847473,
      "learning_rate": 4.024455062295274e-06,
      "loss": 1.1198,
      "step": 2659
    },
    {
      "epoch": 0.7922708911185986,
      "grad_norm": 0.6392356753349304,
      "learning_rate": 4.013772983076123e-06,
      "loss": 1.1015,
      "step": 2660
    },
    {
      "epoch": 0.7925687373182673,
      "grad_norm": 0.6423717141151428,
      "learning_rate": 4.003101538764023e-06,
      "loss": 1.0993,
      "step": 2661
    },
    {
      "epoch": 0.7928665835179359,
      "grad_norm": 0.6462405920028687,
      "learning_rate": 3.992440748317466e-06,
      "loss": 1.1129,
      "step": 2662
    },
    {
      "epoch": 0.7931644297176046,
      "grad_norm": 0.6170511245727539,
      "learning_rate": 3.98179063067602e-06,
      "loss": 1.1068,
      "step": 2663
    },
    {
      "epoch": 0.7934622759172733,
      "grad_norm": 0.686282217502594,
      "learning_rate": 3.971151204760277e-06,
      "loss": 1.1267,
      "step": 2664
    },
    {
      "epoch": 0.7937601221169419,
      "grad_norm": 0.6582475304603577,
      "learning_rate": 3.960522489471847e-06,
      "loss": 1.1093,
      "step": 2665
    },
    {
      "epoch": 0.7940579683166105,
      "grad_norm": 0.6401075124740601,
      "learning_rate": 3.94990450369331e-06,
      "loss": 1.1264,
      "step": 2666
    },
    {
      "epoch": 0.7943558145162791,
      "grad_norm": 0.6683831810951233,
      "learning_rate": 3.9392972662881815e-06,
      "loss": 1.1099,
      "step": 2667
    },
    {
      "epoch": 0.7946536607159478,
      "grad_norm": 0.6492847204208374,
      "learning_rate": 3.9287007961008775e-06,
      "loss": 1.1144,
      "step": 2668
    },
    {
      "epoch": 0.7949515069156164,
      "grad_norm": 0.6592220664024353,
      "learning_rate": 3.918115111956695e-06,
      "loss": 1.1088,
      "step": 2669
    },
    {
      "epoch": 0.7952493531152851,
      "grad_norm": 0.6303004622459412,
      "learning_rate": 3.907540232661767e-06,
      "loss": 1.1117,
      "step": 2670
    },
    {
      "epoch": 0.7955471993149538,
      "grad_norm": 0.6449137330055237,
      "learning_rate": 3.8969761770030235e-06,
      "loss": 1.1289,
      "step": 2671
    },
    {
      "epoch": 0.7958450455146224,
      "grad_norm": 0.6571239829063416,
      "learning_rate": 3.886422963748173e-06,
      "loss": 1.1049,
      "step": 2672
    },
    {
      "epoch": 0.796142891714291,
      "grad_norm": 0.6562924981117249,
      "learning_rate": 3.875880611645662e-06,
      "loss": 1.1047,
      "step": 2673
    },
    {
      "epoch": 0.7964407379139596,
      "grad_norm": 0.6350340247154236,
      "learning_rate": 3.865349139424634e-06,
      "loss": 1.1313,
      "step": 2674
    },
    {
      "epoch": 0.7967385841136283,
      "grad_norm": 0.6584562063217163,
      "learning_rate": 3.8548285657949126e-06,
      "loss": 1.1232,
      "step": 2675
    },
    {
      "epoch": 0.797036430313297,
      "grad_norm": 0.656667172908783,
      "learning_rate": 3.8443189094469535e-06,
      "loss": 1.1209,
      "step": 2676
    },
    {
      "epoch": 0.7973342765129656,
      "grad_norm": 0.6487510800361633,
      "learning_rate": 3.833820189051822e-06,
      "loss": 1.1129,
      "step": 2677
    },
    {
      "epoch": 0.7976321227126343,
      "grad_norm": 0.6394895911216736,
      "learning_rate": 3.823332423261148e-06,
      "loss": 1.1119,
      "step": 2678
    },
    {
      "epoch": 0.797929968912303,
      "grad_norm": 0.6427561044692993,
      "learning_rate": 3.8128556307071062e-06,
      "loss": 1.1294,
      "step": 2679
    },
    {
      "epoch": 0.7982278151119716,
      "grad_norm": 0.6471041440963745,
      "learning_rate": 3.8023898300023776e-06,
      "loss": 1.1083,
      "step": 2680
    },
    {
      "epoch": 0.7985256613116402,
      "grad_norm": 0.6643339395523071,
      "learning_rate": 3.7919350397401054e-06,
      "loss": 1.1207,
      "step": 2681
    },
    {
      "epoch": 0.7988235075113088,
      "grad_norm": 0.6423395276069641,
      "learning_rate": 3.781491278493883e-06,
      "loss": 1.0921,
      "step": 2682
    },
    {
      "epoch": 0.7991213537109775,
      "grad_norm": 0.6330326199531555,
      "learning_rate": 3.7710585648177076e-06,
      "loss": 1.1086,
      "step": 2683
    },
    {
      "epoch": 0.7994191999106461,
      "grad_norm": 0.6721494197845459,
      "learning_rate": 3.7606369172459445e-06,
      "loss": 1.1367,
      "step": 2684
    },
    {
      "epoch": 0.7997170461103148,
      "grad_norm": 0.6488641500473022,
      "learning_rate": 3.750226354293305e-06,
      "loss": 1.1234,
      "step": 2685
    },
    {
      "epoch": 0.8000148923099835,
      "grad_norm": 0.6591084003448486,
      "learning_rate": 3.7398268944548043e-06,
      "loss": 1.1088,
      "step": 2686
    },
    {
      "epoch": 0.8003127385096521,
      "grad_norm": 0.6554153561592102,
      "learning_rate": 3.72943855620574e-06,
      "loss": 1.1141,
      "step": 2687
    },
    {
      "epoch": 0.8006105847093207,
      "grad_norm": 0.6440781950950623,
      "learning_rate": 3.7190613580016376e-06,
      "loss": 1.1237,
      "step": 2688
    },
    {
      "epoch": 0.8009084309089893,
      "grad_norm": 0.6390606164932251,
      "learning_rate": 3.7086953182782413e-06,
      "loss": 1.1283,
      "step": 2689
    },
    {
      "epoch": 0.801206277108658,
      "grad_norm": 0.6526691913604736,
      "learning_rate": 3.6983404554514746e-06,
      "loss": 1.0996,
      "step": 2690
    },
    {
      "epoch": 0.8015041233083267,
      "grad_norm": 0.6426694393157959,
      "learning_rate": 3.687996787917393e-06,
      "loss": 1.1178,
      "step": 2691
    },
    {
      "epoch": 0.8018019695079953,
      "grad_norm": 0.6561403870582581,
      "learning_rate": 3.6776643340521713e-06,
      "loss": 1.1094,
      "step": 2692
    },
    {
      "epoch": 0.802099815707664,
      "grad_norm": 0.6362612247467041,
      "learning_rate": 3.667343112212065e-06,
      "loss": 1.1133,
      "step": 2693
    },
    {
      "epoch": 0.8023976619073326,
      "grad_norm": 0.6687363982200623,
      "learning_rate": 3.6570331407333635e-06,
      "loss": 1.1241,
      "step": 2694
    },
    {
      "epoch": 0.8026955081070013,
      "grad_norm": 0.6407457590103149,
      "learning_rate": 3.646734437932381e-06,
      "loss": 1.1102,
      "step": 2695
    },
    {
      "epoch": 0.8029933543066698,
      "grad_norm": 0.6397781372070312,
      "learning_rate": 3.636447022105407e-06,
      "loss": 1.1141,
      "step": 2696
    },
    {
      "epoch": 0.8032912005063385,
      "grad_norm": 0.6469733715057373,
      "learning_rate": 3.626170911528684e-06,
      "loss": 1.1376,
      "step": 2697
    },
    {
      "epoch": 0.8035890467060072,
      "grad_norm": 0.6357889771461487,
      "learning_rate": 3.615906124458358e-06,
      "loss": 1.1096,
      "step": 2698
    },
    {
      "epoch": 0.8038868929056758,
      "grad_norm": 0.6261385083198547,
      "learning_rate": 3.605652679130469e-06,
      "loss": 1.1039,
      "step": 2699
    },
    {
      "epoch": 0.8041847391053445,
      "grad_norm": 0.656443178653717,
      "learning_rate": 3.5954105937609084e-06,
      "loss": 1.1151,
      "step": 2700
    },
    {
      "epoch": 0.8044825853050132,
      "grad_norm": 0.6494108438491821,
      "learning_rate": 3.585179886545379e-06,
      "loss": 1.1286,
      "step": 2701
    },
    {
      "epoch": 0.8047804315046818,
      "grad_norm": 0.6190235018730164,
      "learning_rate": 3.5749605756593697e-06,
      "loss": 1.1123,
      "step": 2702
    },
    {
      "epoch": 0.8050782777043504,
      "grad_norm": 0.6462294459342957,
      "learning_rate": 3.564752679258128e-06,
      "loss": 1.1065,
      "step": 2703
    },
    {
      "epoch": 0.805376123904019,
      "grad_norm": 0.6349015831947327,
      "learning_rate": 3.5545562154766245e-06,
      "loss": 1.0999,
      "step": 2704
    },
    {
      "epoch": 0.8056739701036877,
      "grad_norm": 0.6420944333076477,
      "learning_rate": 3.5443712024295095e-06,
      "loss": 1.1253,
      "step": 2705
    },
    {
      "epoch": 0.8059718163033563,
      "grad_norm": 0.623308539390564,
      "learning_rate": 3.5341976582111003e-06,
      "loss": 1.0964,
      "step": 2706
    },
    {
      "epoch": 0.806269662503025,
      "grad_norm": 0.6176472902297974,
      "learning_rate": 3.5240356008953367e-06,
      "loss": 1.113,
      "step": 2707
    },
    {
      "epoch": 0.8065675087026937,
      "grad_norm": 0.6491343975067139,
      "learning_rate": 3.5138850485357436e-06,
      "loss": 1.0966,
      "step": 2708
    },
    {
      "epoch": 0.8068653549023623,
      "grad_norm": 0.6221631765365601,
      "learning_rate": 3.503746019165416e-06,
      "loss": 1.1088,
      "step": 2709
    },
    {
      "epoch": 0.807163201102031,
      "grad_norm": 0.6192895174026489,
      "learning_rate": 3.4936185307969796e-06,
      "loss": 1.1136,
      "step": 2710
    },
    {
      "epoch": 0.8074610473016995,
      "grad_norm": 0.6357495188713074,
      "learning_rate": 3.4835026014225413e-06,
      "loss": 1.1277,
      "step": 2711
    },
    {
      "epoch": 0.8077588935013682,
      "grad_norm": 0.6195947527885437,
      "learning_rate": 3.4733982490136884e-06,
      "loss": 1.1182,
      "step": 2712
    },
    {
      "epoch": 0.8080567397010369,
      "grad_norm": 0.6505992412567139,
      "learning_rate": 3.463305491521437e-06,
      "loss": 1.1168,
      "step": 2713
    },
    {
      "epoch": 0.8083545859007055,
      "grad_norm": 0.6459845304489136,
      "learning_rate": 3.453224346876203e-06,
      "loss": 1.1216,
      "step": 2714
    },
    {
      "epoch": 0.8086524321003742,
      "grad_norm": 0.6478413343429565,
      "learning_rate": 3.443154832987765e-06,
      "loss": 1.1129,
      "step": 2715
    },
    {
      "epoch": 0.8089502783000428,
      "grad_norm": 0.6574426889419556,
      "learning_rate": 3.4330969677452496e-06,
      "loss": 1.1152,
      "step": 2716
    },
    {
      "epoch": 0.8092481244997115,
      "grad_norm": 0.636925458908081,
      "learning_rate": 3.4230507690170854e-06,
      "loss": 1.1034,
      "step": 2717
    },
    {
      "epoch": 0.8095459706993801,
      "grad_norm": 0.6324298977851868,
      "learning_rate": 3.4130162546509695e-06,
      "loss": 1.1193,
      "step": 2718
    },
    {
      "epoch": 0.8098438168990487,
      "grad_norm": 0.6640302538871765,
      "learning_rate": 3.4029934424738455e-06,
      "loss": 1.1227,
      "step": 2719
    },
    {
      "epoch": 0.8101416630987174,
      "grad_norm": 0.6425569653511047,
      "learning_rate": 3.3929823502918703e-06,
      "loss": 1.1174,
      "step": 2720
    },
    {
      "epoch": 0.810439509298386,
      "grad_norm": 0.6341091394424438,
      "learning_rate": 3.3829829958903704e-06,
      "loss": 1.1138,
      "step": 2721
    },
    {
      "epoch": 0.8107373554980547,
      "grad_norm": 0.6246615648269653,
      "learning_rate": 3.3729953970338282e-06,
      "loss": 1.1116,
      "step": 2722
    },
    {
      "epoch": 0.8110352016977234,
      "grad_norm": 0.6371176242828369,
      "learning_rate": 3.3630195714658374e-06,
      "loss": 1.1035,
      "step": 2723
    },
    {
      "epoch": 0.811333047897392,
      "grad_norm": 0.644370973110199,
      "learning_rate": 3.3530555369090813e-06,
      "loss": 1.1082,
      "step": 2724
    },
    {
      "epoch": 0.8116308940970607,
      "grad_norm": 0.6362097859382629,
      "learning_rate": 3.3431033110652834e-06,
      "loss": 1.1324,
      "step": 2725
    },
    {
      "epoch": 0.8119287402967292,
      "grad_norm": 0.6245853900909424,
      "learning_rate": 3.3331629116151987e-06,
      "loss": 1.0964,
      "step": 2726
    },
    {
      "epoch": 0.8122265864963979,
      "grad_norm": 0.6418402791023254,
      "learning_rate": 3.3232343562185742e-06,
      "loss": 1.1206,
      "step": 2727
    },
    {
      "epoch": 0.8125244326960666,
      "grad_norm": 0.6407635807991028,
      "learning_rate": 3.3133176625141016e-06,
      "loss": 1.1285,
      "step": 2728
    },
    {
      "epoch": 0.8128222788957352,
      "grad_norm": 0.6410035490989685,
      "learning_rate": 3.3034128481194126e-06,
      "loss": 1.1191,
      "step": 2729
    },
    {
      "epoch": 0.8131201250954039,
      "grad_norm": 0.6521008014678955,
      "learning_rate": 3.2935199306310327e-06,
      "loss": 1.1172,
      "step": 2730
    },
    {
      "epoch": 0.8134179712950725,
      "grad_norm": 0.6405535340309143,
      "learning_rate": 3.2836389276243442e-06,
      "loss": 1.1117,
      "step": 2731
    },
    {
      "epoch": 0.8137158174947412,
      "grad_norm": 0.6362119317054749,
      "learning_rate": 3.273769856653568e-06,
      "loss": 1.1212,
      "step": 2732
    },
    {
      "epoch": 0.8140136636944098,
      "grad_norm": 0.634901225566864,
      "learning_rate": 3.263912735251733e-06,
      "loss": 1.1212,
      "step": 2733
    },
    {
      "epoch": 0.8143115098940784,
      "grad_norm": 0.6256780028343201,
      "learning_rate": 3.2540675809306233e-06,
      "loss": 1.1041,
      "step": 2734
    },
    {
      "epoch": 0.8146093560937471,
      "grad_norm": 0.6250608563423157,
      "learning_rate": 3.244234411180779e-06,
      "loss": 1.0927,
      "step": 2735
    },
    {
      "epoch": 0.8149072022934157,
      "grad_norm": 0.6390246748924255,
      "learning_rate": 3.234413243471436e-06,
      "loss": 1.1367,
      "step": 2736
    },
    {
      "epoch": 0.8152050484930844,
      "grad_norm": 0.6278163194656372,
      "learning_rate": 3.22460409525052e-06,
      "loss": 1.1057,
      "step": 2737
    },
    {
      "epoch": 0.815502894692753,
      "grad_norm": 0.6547091007232666,
      "learning_rate": 3.2148069839445893e-06,
      "loss": 1.133,
      "step": 2738
    },
    {
      "epoch": 0.8158007408924217,
      "grad_norm": 0.6481456160545349,
      "learning_rate": 3.2050219269588302e-06,
      "loss": 1.1251,
      "step": 2739
    },
    {
      "epoch": 0.8160985870920904,
      "grad_norm": 0.6317632794380188,
      "learning_rate": 3.1952489416770083e-06,
      "loss": 1.1073,
      "step": 2740
    },
    {
      "epoch": 0.8163964332917589,
      "grad_norm": 0.6432924866676331,
      "learning_rate": 3.1854880454614466e-06,
      "loss": 1.1089,
      "step": 2741
    },
    {
      "epoch": 0.8166942794914276,
      "grad_norm": 0.6188756227493286,
      "learning_rate": 3.1757392556529833e-06,
      "loss": 1.0959,
      "step": 2742
    },
    {
      "epoch": 0.8169921256910962,
      "grad_norm": 0.6345187425613403,
      "learning_rate": 3.1660025895709547e-06,
      "loss": 1.1318,
      "step": 2743
    },
    {
      "epoch": 0.8172899718907649,
      "grad_norm": 0.6404332518577576,
      "learning_rate": 3.156278064513163e-06,
      "loss": 1.1212,
      "step": 2744
    },
    {
      "epoch": 0.8175878180904336,
      "grad_norm": 0.6379736661911011,
      "learning_rate": 3.146565697755828e-06,
      "loss": 1.1204,
      "step": 2745
    },
    {
      "epoch": 0.8178856642901022,
      "grad_norm": 0.6458266377449036,
      "learning_rate": 3.1368655065535814e-06,
      "loss": 1.1224,
      "step": 2746
    },
    {
      "epoch": 0.8181835104897709,
      "grad_norm": 0.6492531299591064,
      "learning_rate": 3.127177508139423e-06,
      "loss": 1.1067,
      "step": 2747
    },
    {
      "epoch": 0.8184813566894394,
      "grad_norm": 0.6343971490859985,
      "learning_rate": 3.1175017197246814e-06,
      "loss": 1.1112,
      "step": 2748
    },
    {
      "epoch": 0.8187792028891081,
      "grad_norm": 0.625598132610321,
      "learning_rate": 3.1078381584990047e-06,
      "loss": 1.1211,
      "step": 2749
    },
    {
      "epoch": 0.8190770490887768,
      "grad_norm": 0.6241737604141235,
      "learning_rate": 3.098186841630314e-06,
      "loss": 1.1191,
      "step": 2750
    },
    {
      "epoch": 0.8193748952884454,
      "grad_norm": 0.6393356323242188,
      "learning_rate": 3.0885477862647805e-06,
      "loss": 1.1141,
      "step": 2751
    },
    {
      "epoch": 0.8196727414881141,
      "grad_norm": 0.6059439182281494,
      "learning_rate": 3.078921009526785e-06,
      "loss": 1.1003,
      "step": 2752
    },
    {
      "epoch": 0.8199705876877827,
      "grad_norm": 0.6142548322677612,
      "learning_rate": 3.0693065285188995e-06,
      "loss": 1.1153,
      "step": 2753
    },
    {
      "epoch": 0.8202684338874514,
      "grad_norm": 0.6381494402885437,
      "learning_rate": 3.0597043603218567e-06,
      "loss": 1.1078,
      "step": 2754
    },
    {
      "epoch": 0.82056628008712,
      "grad_norm": 0.6559354662895203,
      "learning_rate": 3.050114521994503e-06,
      "loss": 1.0919,
      "step": 2755
    },
    {
      "epoch": 0.8208641262867886,
      "grad_norm": 0.650456428527832,
      "learning_rate": 3.040537030573788e-06,
      "loss": 1.127,
      "step": 2756
    },
    {
      "epoch": 0.8211619724864573,
      "grad_norm": 0.6326876282691956,
      "learning_rate": 3.030971903074731e-06,
      "loss": 1.1033,
      "step": 2757
    },
    {
      "epoch": 0.8214598186861259,
      "grad_norm": 0.6478701233863831,
      "learning_rate": 3.0214191564903694e-06,
      "loss": 1.1045,
      "step": 2758
    },
    {
      "epoch": 0.8217576648857946,
      "grad_norm": 0.6350187659263611,
      "learning_rate": 3.0118788077917626e-06,
      "loss": 1.0971,
      "step": 2759
    },
    {
      "epoch": 0.8220555110854633,
      "grad_norm": 0.6455335021018982,
      "learning_rate": 3.0023508739279384e-06,
      "loss": 1.1182,
      "step": 2760
    },
    {
      "epoch": 0.8223533572851319,
      "grad_norm": 0.6385021209716797,
      "learning_rate": 2.992835371825862e-06,
      "loss": 1.118,
      "step": 2761
    },
    {
      "epoch": 0.8226512034848006,
      "grad_norm": 0.6302372813224792,
      "learning_rate": 2.9833323183904216e-06,
      "loss": 1.1254,
      "step": 2762
    },
    {
      "epoch": 0.8229490496844691,
      "grad_norm": 0.6140434145927429,
      "learning_rate": 2.9738417305043867e-06,
      "loss": 1.1124,
      "step": 2763
    },
    {
      "epoch": 0.8232468958841378,
      "grad_norm": 0.6497140526771545,
      "learning_rate": 2.9643636250283837e-06,
      "loss": 1.1304,
      "step": 2764
    },
    {
      "epoch": 0.8235447420838065,
      "grad_norm": 0.6620893478393555,
      "learning_rate": 2.9548980188008545e-06,
      "loss": 1.1198,
      "step": 2765
    },
    {
      "epoch": 0.8238425882834751,
      "grad_norm": 0.6452887058258057,
      "learning_rate": 2.945444928638044e-06,
      "loss": 1.1133,
      "step": 2766
    },
    {
      "epoch": 0.8241404344831438,
      "grad_norm": 0.6431753039360046,
      "learning_rate": 2.9360043713339616e-06,
      "loss": 1.1135,
      "step": 2767
    },
    {
      "epoch": 0.8244382806828124,
      "grad_norm": 0.617460310459137,
      "learning_rate": 2.926576363660346e-06,
      "loss": 1.1078,
      "step": 2768
    },
    {
      "epoch": 0.8247361268824811,
      "grad_norm": 0.6379134654998779,
      "learning_rate": 2.9171609223666396e-06,
      "loss": 1.1173,
      "step": 2769
    },
    {
      "epoch": 0.8250339730821497,
      "grad_norm": 0.6460497975349426,
      "learning_rate": 2.907758064179965e-06,
      "loss": 1.0954,
      "step": 2770
    },
    {
      "epoch": 0.8253318192818183,
      "grad_norm": 0.6395160555839539,
      "learning_rate": 2.8983678058050934e-06,
      "loss": 1.1254,
      "step": 2771
    },
    {
      "epoch": 0.825629665481487,
      "grad_norm": 0.661530077457428,
      "learning_rate": 2.888990163924401e-06,
      "loss": 1.1141,
      "step": 2772
    },
    {
      "epoch": 0.8259275116811556,
      "grad_norm": 0.6374372243881226,
      "learning_rate": 2.879625155197856e-06,
      "loss": 1.1081,
      "step": 2773
    },
    {
      "epoch": 0.8262253578808243,
      "grad_norm": 0.6358816623687744,
      "learning_rate": 2.8702727962629883e-06,
      "loss": 1.1266,
      "step": 2774
    },
    {
      "epoch": 0.826523204080493,
      "grad_norm": 0.6472240090370178,
      "learning_rate": 2.860933103734842e-06,
      "loss": 1.1218,
      "step": 2775
    },
    {
      "epoch": 0.8268210502801616,
      "grad_norm": 0.6269724369049072,
      "learning_rate": 2.8516060942059677e-06,
      "loss": 1.1078,
      "step": 2776
    },
    {
      "epoch": 0.8271188964798303,
      "grad_norm": 0.6572033166885376,
      "learning_rate": 2.842291784246384e-06,
      "loss": 1.1003,
      "step": 2777
    },
    {
      "epoch": 0.8274167426794988,
      "grad_norm": 0.6654300689697266,
      "learning_rate": 2.832990190403546e-06,
      "loss": 1.1344,
      "step": 2778
    },
    {
      "epoch": 0.8277145888791675,
      "grad_norm": 0.6286570429801941,
      "learning_rate": 2.823701329202313e-06,
      "loss": 1.1061,
      "step": 2779
    },
    {
      "epoch": 0.8280124350788362,
      "grad_norm": 0.6106312870979309,
      "learning_rate": 2.8144252171449316e-06,
      "loss": 1.0872,
      "step": 2780
    },
    {
      "epoch": 0.8283102812785048,
      "grad_norm": 0.6398670077323914,
      "learning_rate": 2.8051618707109984e-06,
      "loss": 1.1194,
      "step": 2781
    },
    {
      "epoch": 0.8286081274781735,
      "grad_norm": 0.63633131980896,
      "learning_rate": 2.7959113063574216e-06,
      "loss": 1.1075,
      "step": 2782
    },
    {
      "epoch": 0.8289059736778421,
      "grad_norm": 0.6191324591636658,
      "learning_rate": 2.7866735405184118e-06,
      "loss": 1.1023,
      "step": 2783
    },
    {
      "epoch": 0.8292038198775108,
      "grad_norm": 0.6473433971405029,
      "learning_rate": 2.7774485896054404e-06,
      "loss": 1.1038,
      "step": 2784
    },
    {
      "epoch": 0.8295016660771793,
      "grad_norm": 0.662238359451294,
      "learning_rate": 2.7682364700072053e-06,
      "loss": 1.1182,
      "step": 2785
    },
    {
      "epoch": 0.829799512276848,
      "grad_norm": 0.6450248956680298,
      "learning_rate": 2.759037198089616e-06,
      "loss": 1.1338,
      "step": 2786
    },
    {
      "epoch": 0.8300973584765167,
      "grad_norm": 0.6293185949325562,
      "learning_rate": 2.7498507901957537e-06,
      "loss": 1.0994,
      "step": 2787
    },
    {
      "epoch": 0.8303952046761853,
      "grad_norm": 0.6551886796951294,
      "learning_rate": 2.7406772626458535e-06,
      "loss": 1.1108,
      "step": 2788
    },
    {
      "epoch": 0.830693050875854,
      "grad_norm": 0.6277971863746643,
      "learning_rate": 2.7315166317372523e-06,
      "loss": 1.1069,
      "step": 2789
    },
    {
      "epoch": 0.8309908970755226,
      "grad_norm": 0.6366534233093262,
      "learning_rate": 2.7223689137443876e-06,
      "loss": 1.1022,
      "step": 2790
    },
    {
      "epoch": 0.8312887432751913,
      "grad_norm": 0.6551545858383179,
      "learning_rate": 2.7132341249187587e-06,
      "loss": 1.1204,
      "step": 2791
    },
    {
      "epoch": 0.83158658947486,
      "grad_norm": 0.6311073303222656,
      "learning_rate": 2.704112281488881e-06,
      "loss": 1.106,
      "step": 2792
    },
    {
      "epoch": 0.8318844356745285,
      "grad_norm": 0.6362341642379761,
      "learning_rate": 2.6950033996602844e-06,
      "loss": 1.1223,
      "step": 2793
    },
    {
      "epoch": 0.8321822818741972,
      "grad_norm": 0.649966835975647,
      "learning_rate": 2.6859074956154717e-06,
      "loss": 1.1225,
      "step": 2794
    },
    {
      "epoch": 0.8324801280738658,
      "grad_norm": 0.6718718409538269,
      "learning_rate": 2.6768245855138784e-06,
      "loss": 1.1302,
      "step": 2795
    },
    {
      "epoch": 0.8327779742735345,
      "grad_norm": 0.6499239206314087,
      "learning_rate": 2.667754685491867e-06,
      "loss": 1.1227,
      "step": 2796
    },
    {
      "epoch": 0.8330758204732032,
      "grad_norm": 0.6348280310630798,
      "learning_rate": 2.6586978116626837e-06,
      "loss": 1.1132,
      "step": 2797
    },
    {
      "epoch": 0.8333736666728718,
      "grad_norm": 0.6653286218643188,
      "learning_rate": 2.649653980116429e-06,
      "loss": 1.1153,
      "step": 2798
    },
    {
      "epoch": 0.8336715128725405,
      "grad_norm": 0.6311964988708496,
      "learning_rate": 2.6406232069200365e-06,
      "loss": 1.0978,
      "step": 2799
    },
    {
      "epoch": 0.833969359072209,
      "grad_norm": 0.6466732621192932,
      "learning_rate": 2.631605508117241e-06,
      "loss": 1.1195,
      "step": 2800
    },
    {
      "epoch": 0.8342672052718777,
      "grad_norm": 0.6353030204772949,
      "learning_rate": 2.6226008997285513e-06,
      "loss": 1.1218,
      "step": 2801
    },
    {
      "epoch": 0.8345650514715464,
      "grad_norm": 0.6905455589294434,
      "learning_rate": 2.6136093977512157e-06,
      "loss": 1.1195,
      "step": 2802
    },
    {
      "epoch": 0.834862897671215,
      "grad_norm": 0.6305209398269653,
      "learning_rate": 2.604631018159197e-06,
      "loss": 1.1165,
      "step": 2803
    },
    {
      "epoch": 0.8351607438708837,
      "grad_norm": 0.6307226419448853,
      "learning_rate": 2.595665776903152e-06,
      "loss": 1.1301,
      "step": 2804
    },
    {
      "epoch": 0.8354585900705523,
      "grad_norm": 0.6256515383720398,
      "learning_rate": 2.586713689910395e-06,
      "loss": 1.1138,
      "step": 2805
    },
    {
      "epoch": 0.835756436270221,
      "grad_norm": 0.6532456278800964,
      "learning_rate": 2.5777747730848646e-06,
      "loss": 1.1004,
      "step": 2806
    },
    {
      "epoch": 0.8360542824698897,
      "grad_norm": 0.6249917149543762,
      "learning_rate": 2.5688490423071076e-06,
      "loss": 1.113,
      "step": 2807
    },
    {
      "epoch": 0.8363521286695582,
      "grad_norm": 0.6597284078598022,
      "learning_rate": 2.559936513434248e-06,
      "loss": 1.0952,
      "step": 2808
    },
    {
      "epoch": 0.8366499748692269,
      "grad_norm": 0.6405438184738159,
      "learning_rate": 2.551037202299944e-06,
      "loss": 1.1164,
      "step": 2809
    },
    {
      "epoch": 0.8369478210688955,
      "grad_norm": 0.6287737488746643,
      "learning_rate": 2.5421511247143826e-06,
      "loss": 1.1191,
      "step": 2810
    },
    {
      "epoch": 0.8372456672685642,
      "grad_norm": 0.643944501876831,
      "learning_rate": 2.533278296464239e-06,
      "loss": 1.1157,
      "step": 2811
    },
    {
      "epoch": 0.8375435134682329,
      "grad_norm": 0.6316015124320984,
      "learning_rate": 2.5244187333126437e-06,
      "loss": 1.1111,
      "step": 2812
    },
    {
      "epoch": 0.8378413596679015,
      "grad_norm": 0.6175442337989807,
      "learning_rate": 2.5155724509991663e-06,
      "loss": 1.0966,
      "step": 2813
    },
    {
      "epoch": 0.8381392058675702,
      "grad_norm": 0.6482809782028198,
      "learning_rate": 2.5067394652397836e-06,
      "loss": 1.1208,
      "step": 2814
    },
    {
      "epoch": 0.8384370520672387,
      "grad_norm": 0.6426069140434265,
      "learning_rate": 2.4979197917268484e-06,
      "loss": 1.0977,
      "step": 2815
    },
    {
      "epoch": 0.8387348982669074,
      "grad_norm": 0.6304159760475159,
      "learning_rate": 2.489113446129059e-06,
      "loss": 1.1031,
      "step": 2816
    },
    {
      "epoch": 0.839032744466576,
      "grad_norm": 0.6366993188858032,
      "learning_rate": 2.4803204440914396e-06,
      "loss": 1.1119,
      "step": 2817
    },
    {
      "epoch": 0.8393305906662447,
      "grad_norm": 0.6106964349746704,
      "learning_rate": 2.471540801235316e-06,
      "loss": 1.0957,
      "step": 2818
    },
    {
      "epoch": 0.8396284368659134,
      "grad_norm": 0.6652907133102417,
      "learning_rate": 2.4627745331582632e-06,
      "loss": 1.1288,
      "step": 2819
    },
    {
      "epoch": 0.839926283065582,
      "grad_norm": 0.6200968623161316,
      "learning_rate": 2.454021655434111e-06,
      "loss": 1.1071,
      "step": 2820
    },
    {
      "epoch": 0.8402241292652507,
      "grad_norm": 0.6320287585258484,
      "learning_rate": 2.445282183612897e-06,
      "loss": 1.1155,
      "step": 2821
    },
    {
      "epoch": 0.8405219754649194,
      "grad_norm": 0.6298106908798218,
      "learning_rate": 2.436556133220833e-06,
      "loss": 1.1101,
      "step": 2822
    },
    {
      "epoch": 0.8408198216645879,
      "grad_norm": 0.6517424583435059,
      "learning_rate": 2.427843519760298e-06,
      "loss": 1.1056,
      "step": 2823
    },
    {
      "epoch": 0.8411176678642566,
      "grad_norm": 0.6429053544998169,
      "learning_rate": 2.4191443587097975e-06,
      "loss": 1.117,
      "step": 2824
    },
    {
      "epoch": 0.8414155140639252,
      "grad_norm": 0.6462130546569824,
      "learning_rate": 2.4104586655239305e-06,
      "loss": 1.1179,
      "step": 2825
    },
    {
      "epoch": 0.8417133602635939,
      "grad_norm": 0.625566303730011,
      "learning_rate": 2.401786455633377e-06,
      "loss": 1.1202,
      "step": 2826
    },
    {
      "epoch": 0.8420112064632626,
      "grad_norm": 0.6242573857307434,
      "learning_rate": 2.393127744444862e-06,
      "loss": 1.1173,
      "step": 2827
    },
    {
      "epoch": 0.8423090526629312,
      "grad_norm": 0.6385311484336853,
      "learning_rate": 2.3844825473411282e-06,
      "loss": 1.1041,
      "step": 2828
    },
    {
      "epoch": 0.8426068988625999,
      "grad_norm": 0.6276160478591919,
      "learning_rate": 2.3758508796809067e-06,
      "loss": 1.1021,
      "step": 2829
    },
    {
      "epoch": 0.8429047450622684,
      "grad_norm": 0.6526584625244141,
      "learning_rate": 2.3672327567988963e-06,
      "loss": 1.1175,
      "step": 2830
    },
    {
      "epoch": 0.8432025912619371,
      "grad_norm": 0.6742414832115173,
      "learning_rate": 2.3586281940057367e-06,
      "loss": 1.1088,
      "step": 2831
    },
    {
      "epoch": 0.8435004374616057,
      "grad_norm": 0.6288987994194031,
      "learning_rate": 2.350037206587964e-06,
      "loss": 1.1139,
      "step": 2832
    },
    {
      "epoch": 0.8437982836612744,
      "grad_norm": 0.6364237070083618,
      "learning_rate": 2.34145980980801e-06,
      "loss": 1.1123,
      "step": 2833
    },
    {
      "epoch": 0.8440961298609431,
      "grad_norm": 0.6482524871826172,
      "learning_rate": 2.3328960189041593e-06,
      "loss": 1.1178,
      "step": 2834
    },
    {
      "epoch": 0.8443939760606117,
      "grad_norm": 0.6284860372543335,
      "learning_rate": 2.324345849090517e-06,
      "loss": 1.1263,
      "step": 2835
    },
    {
      "epoch": 0.8446918222602804,
      "grad_norm": 0.6118963360786438,
      "learning_rate": 2.3158093155570003e-06,
      "loss": 1.1061,
      "step": 2836
    },
    {
      "epoch": 0.8449896684599489,
      "grad_norm": 0.6164010763168335,
      "learning_rate": 2.3072864334692903e-06,
      "loss": 1.1077,
      "step": 2837
    },
    {
      "epoch": 0.8452875146596176,
      "grad_norm": 0.643302857875824,
      "learning_rate": 2.2987772179688263e-06,
      "loss": 1.1191,
      "step": 2838
    },
    {
      "epoch": 0.8455853608592863,
      "grad_norm": 0.6198796629905701,
      "learning_rate": 2.290281684172757e-06,
      "loss": 1.0903,
      "step": 2839
    },
    {
      "epoch": 0.8458832070589549,
      "grad_norm": 0.6291831135749817,
      "learning_rate": 2.281799847173931e-06,
      "loss": 1.116,
      "step": 2840
    },
    {
      "epoch": 0.8461810532586236,
      "grad_norm": 0.663838267326355,
      "learning_rate": 2.2733317220408647e-06,
      "loss": 1.1114,
      "step": 2841
    },
    {
      "epoch": 0.8464788994582922,
      "grad_norm": 0.6386342644691467,
      "learning_rate": 2.2648773238177147e-06,
      "loss": 1.1051,
      "step": 2842
    },
    {
      "epoch": 0.8467767456579609,
      "grad_norm": 0.6362149715423584,
      "learning_rate": 2.2564366675242433e-06,
      "loss": 1.1147,
      "step": 2843
    },
    {
      "epoch": 0.8470745918576296,
      "grad_norm": 0.6314151883125305,
      "learning_rate": 2.2480097681558077e-06,
      "loss": 1.1133,
      "step": 2844
    },
    {
      "epoch": 0.8473724380572981,
      "grad_norm": 0.6264147162437439,
      "learning_rate": 2.2395966406833237e-06,
      "loss": 1.1136,
      "step": 2845
    },
    {
      "epoch": 0.8476702842569668,
      "grad_norm": 0.6644660234451294,
      "learning_rate": 2.231197300053235e-06,
      "loss": 1.1208,
      "step": 2846
    },
    {
      "epoch": 0.8479681304566354,
      "grad_norm": 0.6151517629623413,
      "learning_rate": 2.2228117611874987e-06,
      "loss": 1.1059,
      "step": 2847
    },
    {
      "epoch": 0.8482659766563041,
      "grad_norm": 0.6255171298980713,
      "learning_rate": 2.2144400389835506e-06,
      "loss": 1.1085,
      "step": 2848
    },
    {
      "epoch": 0.8485638228559728,
      "grad_norm": 0.6459829807281494,
      "learning_rate": 2.206082148314276e-06,
      "loss": 1.1024,
      "step": 2849
    },
    {
      "epoch": 0.8488616690556414,
      "grad_norm": 0.6791364550590515,
      "learning_rate": 2.197738104027992e-06,
      "loss": 1.0886,
      "step": 2850
    },
    {
      "epoch": 0.8491595152553101,
      "grad_norm": 0.6318515539169312,
      "learning_rate": 2.18940792094842e-06,
      "loss": 1.1197,
      "step": 2851
    },
    {
      "epoch": 0.8494573614549786,
      "grad_norm": 0.6258553862571716,
      "learning_rate": 2.181091613874645e-06,
      "loss": 1.1218,
      "step": 2852
    },
    {
      "epoch": 0.8497552076546473,
      "grad_norm": 0.6389538645744324,
      "learning_rate": 2.1727891975811098e-06,
      "loss": 1.1058,
      "step": 2853
    },
    {
      "epoch": 0.850053053854316,
      "grad_norm": 0.6531283259391785,
      "learning_rate": 2.1645006868175765e-06,
      "loss": 1.1254,
      "step": 2854
    },
    {
      "epoch": 0.8503509000539846,
      "grad_norm": 0.6217397451400757,
      "learning_rate": 2.1562260963091086e-06,
      "loss": 1.109,
      "step": 2855
    },
    {
      "epoch": 0.8506487462536533,
      "grad_norm": 0.657103419303894,
      "learning_rate": 2.1479654407560256e-06,
      "loss": 1.1153,
      "step": 2856
    },
    {
      "epoch": 0.8509465924533219,
      "grad_norm": 0.6401103734970093,
      "learning_rate": 2.139718734833903e-06,
      "loss": 1.1106,
      "step": 2857
    },
    {
      "epoch": 0.8512444386529906,
      "grad_norm": 0.6249285936355591,
      "learning_rate": 2.131485993193534e-06,
      "loss": 1.1035,
      "step": 2858
    },
    {
      "epoch": 0.8515422848526593,
      "grad_norm": 0.6378918290138245,
      "learning_rate": 2.1232672304608914e-06,
      "loss": 1.1178,
      "step": 2859
    },
    {
      "epoch": 0.8518401310523278,
      "grad_norm": 0.6186591982841492,
      "learning_rate": 2.1150624612371273e-06,
      "loss": 1.1229,
      "step": 2860
    },
    {
      "epoch": 0.8521379772519965,
      "grad_norm": 0.6250067353248596,
      "learning_rate": 2.1068717000985294e-06,
      "loss": 1.1012,
      "step": 2861
    },
    {
      "epoch": 0.8524358234516651,
      "grad_norm": 0.6471474170684814,
      "learning_rate": 2.0986949615964926e-06,
      "loss": 1.1159,
      "step": 2862
    },
    {
      "epoch": 0.8527336696513338,
      "grad_norm": 0.628984808921814,
      "learning_rate": 2.090532260257507e-06,
      "loss": 1.1235,
      "step": 2863
    },
    {
      "epoch": 0.8530315158510025,
      "grad_norm": 0.6324835419654846,
      "learning_rate": 2.082383610583122e-06,
      "loss": 1.1062,
      "step": 2864
    },
    {
      "epoch": 0.8533293620506711,
      "grad_norm": 0.6540479063987732,
      "learning_rate": 2.0742490270499284e-06,
      "loss": 1.1036,
      "step": 2865
    },
    {
      "epoch": 0.8536272082503398,
      "grad_norm": 0.6349487900733948,
      "learning_rate": 2.066128524109515e-06,
      "loss": 1.126,
      "step": 2866
    },
    {
      "epoch": 0.8539250544500083,
      "grad_norm": 0.6401717066764832,
      "learning_rate": 2.0580221161884693e-06,
      "loss": 1.108,
      "step": 2867
    },
    {
      "epoch": 0.854222900649677,
      "grad_norm": 0.6426677107810974,
      "learning_rate": 2.049929817688333e-06,
      "loss": 1.1258,
      "step": 2868
    },
    {
      "epoch": 0.8545207468493456,
      "grad_norm": 0.6508980989456177,
      "learning_rate": 2.0418516429855796e-06,
      "loss": 1.1092,
      "step": 2869
    },
    {
      "epoch": 0.8548185930490143,
      "grad_norm": 0.620776891708374,
      "learning_rate": 2.0337876064315888e-06,
      "loss": 1.1055,
      "step": 2870
    },
    {
      "epoch": 0.855116439248683,
      "grad_norm": 0.6526078581809998,
      "learning_rate": 2.0257377223526285e-06,
      "loss": 1.1345,
      "step": 2871
    },
    {
      "epoch": 0.8554142854483516,
      "grad_norm": 0.6604386568069458,
      "learning_rate": 2.017702005049825e-06,
      "loss": 1.1142,
      "step": 2872
    },
    {
      "epoch": 0.8557121316480203,
      "grad_norm": 0.6667323708534241,
      "learning_rate": 2.009680468799129e-06,
      "loss": 1.1317,
      "step": 2873
    },
    {
      "epoch": 0.856009977847689,
      "grad_norm": 0.647976815700531,
      "learning_rate": 2.0016731278513023e-06,
      "loss": 1.126,
      "step": 2874
    },
    {
      "epoch": 0.8563078240473575,
      "grad_norm": 0.6418489813804626,
      "learning_rate": 1.993679996431892e-06,
      "loss": 1.1171,
      "step": 2875
    },
    {
      "epoch": 0.8566056702470262,
      "grad_norm": 0.6412356495857239,
      "learning_rate": 1.98570108874119e-06,
      "loss": 1.111,
      "step": 2876
    },
    {
      "epoch": 0.8569035164466948,
      "grad_norm": 0.6608038544654846,
      "learning_rate": 1.9777364189542283e-06,
      "loss": 1.1269,
      "step": 2877
    },
    {
      "epoch": 0.8572013626463635,
      "grad_norm": 0.6341904401779175,
      "learning_rate": 1.9697860012207416e-06,
      "loss": 1.1249,
      "step": 2878
    },
    {
      "epoch": 0.8574992088460321,
      "grad_norm": 0.6151966452598572,
      "learning_rate": 1.9618498496651473e-06,
      "loss": 1.1228,
      "step": 2879
    },
    {
      "epoch": 0.8577970550457008,
      "grad_norm": 0.6518832445144653,
      "learning_rate": 1.9539279783865093e-06,
      "loss": 1.1259,
      "step": 2880
    },
    {
      "epoch": 0.8580949012453695,
      "grad_norm": 0.6440809965133667,
      "learning_rate": 1.9460204014585314e-06,
      "loss": 1.1296,
      "step": 2881
    },
    {
      "epoch": 0.858392747445038,
      "grad_norm": 0.6473000645637512,
      "learning_rate": 1.9381271329295215e-06,
      "loss": 1.0971,
      "step": 2882
    },
    {
      "epoch": 0.8586905936447067,
      "grad_norm": 0.6497401595115662,
      "learning_rate": 1.93024818682236e-06,
      "loss": 1.1099,
      "step": 2883
    },
    {
      "epoch": 0.8589884398443753,
      "grad_norm": 0.6472067832946777,
      "learning_rate": 1.922383577134491e-06,
      "loss": 1.1096,
      "step": 2884
    },
    {
      "epoch": 0.859286286044044,
      "grad_norm": 0.6191689372062683,
      "learning_rate": 1.9145333178378876e-06,
      "loss": 1.1212,
      "step": 2885
    },
    {
      "epoch": 0.8595841322437127,
      "grad_norm": 0.6502696871757507,
      "learning_rate": 1.9066974228790235e-06,
      "loss": 1.1134,
      "step": 2886
    },
    {
      "epoch": 0.8598819784433813,
      "grad_norm": 0.6138703227043152,
      "learning_rate": 1.8988759061788565e-06,
      "loss": 1.0941,
      "step": 2887
    },
    {
      "epoch": 0.86017982464305,
      "grad_norm": 0.6289187073707581,
      "learning_rate": 1.8910687816328055e-06,
      "loss": 1.0958,
      "step": 2888
    },
    {
      "epoch": 0.8604776708427186,
      "grad_norm": 0.6473160982131958,
      "learning_rate": 1.8832760631107105e-06,
      "loss": 1.1347,
      "step": 2889
    },
    {
      "epoch": 0.8607755170423872,
      "grad_norm": 0.622821569442749,
      "learning_rate": 1.875497764456825e-06,
      "loss": 1.0988,
      "step": 2890
    },
    {
      "epoch": 0.8610733632420559,
      "grad_norm": 0.6219084858894348,
      "learning_rate": 1.8677338994897853e-06,
      "loss": 1.1043,
      "step": 2891
    },
    {
      "epoch": 0.8613712094417245,
      "grad_norm": 0.6382589340209961,
      "learning_rate": 1.8599844820025847e-06,
      "loss": 1.1184,
      "step": 2892
    },
    {
      "epoch": 0.8616690556413932,
      "grad_norm": 0.6568984389305115,
      "learning_rate": 1.852249525762544e-06,
      "loss": 1.0971,
      "step": 2893
    },
    {
      "epoch": 0.8619669018410618,
      "grad_norm": 0.6421190500259399,
      "learning_rate": 1.8445290445113006e-06,
      "loss": 1.1196,
      "step": 2894
    },
    {
      "epoch": 0.8622647480407305,
      "grad_norm": 0.6508284211158752,
      "learning_rate": 1.8368230519647735e-06,
      "loss": 1.116,
      "step": 2895
    },
    {
      "epoch": 0.8625625942403992,
      "grad_norm": 0.6386772394180298,
      "learning_rate": 1.8291315618131366e-06,
      "loss": 1.1069,
      "step": 2896
    },
    {
      "epoch": 0.8628604404400677,
      "grad_norm": 0.6349539160728455,
      "learning_rate": 1.8214545877208078e-06,
      "loss": 1.1126,
      "step": 2897
    },
    {
      "epoch": 0.8631582866397364,
      "grad_norm": 0.6088827848434448,
      "learning_rate": 1.8137921433264127e-06,
      "loss": 1.1134,
      "step": 2898
    },
    {
      "epoch": 0.863456132839405,
      "grad_norm": 0.6439886689186096,
      "learning_rate": 1.806144242242759e-06,
      "loss": 1.1148,
      "step": 2899
    },
    {
      "epoch": 0.8637539790390737,
      "grad_norm": 0.624866783618927,
      "learning_rate": 1.7985108980568234e-06,
      "loss": 1.1046,
      "step": 2900
    },
    {
      "epoch": 0.8640518252387424,
      "grad_norm": 0.6500748991966248,
      "learning_rate": 1.7908921243297216e-06,
      "loss": 1.0968,
      "step": 2901
    },
    {
      "epoch": 0.864349671438411,
      "grad_norm": 0.6444833278656006,
      "learning_rate": 1.7832879345966824e-06,
      "loss": 1.1168,
      "step": 2902
    },
    {
      "epoch": 0.8646475176380797,
      "grad_norm": 0.6480167508125305,
      "learning_rate": 1.7756983423670248e-06,
      "loss": 1.1042,
      "step": 2903
    },
    {
      "epoch": 0.8649453638377483,
      "grad_norm": 0.6273287534713745,
      "learning_rate": 1.768123361124131e-06,
      "loss": 1.1108,
      "step": 2904
    },
    {
      "epoch": 0.8652432100374169,
      "grad_norm": 0.640640139579773,
      "learning_rate": 1.7605630043254307e-06,
      "loss": 1.1139,
      "step": 2905
    },
    {
      "epoch": 0.8655410562370855,
      "grad_norm": 0.6398541331291199,
      "learning_rate": 1.7530172854023753e-06,
      "loss": 1.1076,
      "step": 2906
    },
    {
      "epoch": 0.8658389024367542,
      "grad_norm": 0.6420118808746338,
      "learning_rate": 1.7454862177604015e-06,
      "loss": 1.1095,
      "step": 2907
    },
    {
      "epoch": 0.8661367486364229,
      "grad_norm": 0.6550958752632141,
      "learning_rate": 1.7379698147789238e-06,
      "loss": 1.1316,
      "step": 2908
    },
    {
      "epoch": 0.8664345948360915,
      "grad_norm": 0.6243671774864197,
      "learning_rate": 1.730468089811307e-06,
      "loss": 1.1138,
      "step": 2909
    },
    {
      "epoch": 0.8667324410357602,
      "grad_norm": 0.6225848197937012,
      "learning_rate": 1.722981056184828e-06,
      "loss": 1.1088,
      "step": 2910
    },
    {
      "epoch": 0.8670302872354289,
      "grad_norm": 0.616462230682373,
      "learning_rate": 1.7155087272006743e-06,
      "loss": 1.1009,
      "step": 2911
    },
    {
      "epoch": 0.8673281334350974,
      "grad_norm": 0.622468888759613,
      "learning_rate": 1.7080511161339097e-06,
      "loss": 1.1004,
      "step": 2912
    },
    {
      "epoch": 0.8676259796347661,
      "grad_norm": 0.6293972730636597,
      "learning_rate": 1.7006082362334409e-06,
      "loss": 1.1057,
      "step": 2913
    },
    {
      "epoch": 0.8679238258344347,
      "grad_norm": 0.6402567028999329,
      "learning_rate": 1.6931801007220117e-06,
      "loss": 1.1115,
      "step": 2914
    },
    {
      "epoch": 0.8682216720341034,
      "grad_norm": 0.6238921880722046,
      "learning_rate": 1.685766722796174e-06,
      "loss": 1.1035,
      "step": 2915
    },
    {
      "epoch": 0.868519518233772,
      "grad_norm": 0.6372717618942261,
      "learning_rate": 1.678368115626252e-06,
      "loss": 1.1119,
      "step": 2916
    },
    {
      "epoch": 0.8688173644334407,
      "grad_norm": 0.6510122418403625,
      "learning_rate": 1.670984292356338e-06,
      "loss": 1.1208,
      "step": 2917
    },
    {
      "epoch": 0.8691152106331094,
      "grad_norm": 0.6227189302444458,
      "learning_rate": 1.6636152661042549e-06,
      "loss": 1.1106,
      "step": 2918
    },
    {
      "epoch": 0.8694130568327779,
      "grad_norm": 0.6487370729446411,
      "learning_rate": 1.6562610499615416e-06,
      "loss": 1.1231,
      "step": 2919
    },
    {
      "epoch": 0.8697109030324466,
      "grad_norm": 0.6215865015983582,
      "learning_rate": 1.6489216569934197e-06,
      "loss": 1.1007,
      "step": 2920
    },
    {
      "epoch": 0.8700087492321152,
      "grad_norm": 0.649468183517456,
      "learning_rate": 1.6415971002387832e-06,
      "loss": 1.1266,
      "step": 2921
    },
    {
      "epoch": 0.8703065954317839,
      "grad_norm": 0.6304287910461426,
      "learning_rate": 1.6342873927101655e-06,
      "loss": 1.1226,
      "step": 2922
    },
    {
      "epoch": 0.8706044416314526,
      "grad_norm": 0.6425588726997375,
      "learning_rate": 1.6269925473937177e-06,
      "loss": 1.0865,
      "step": 2923
    },
    {
      "epoch": 0.8709022878311212,
      "grad_norm": 0.6347776055335999,
      "learning_rate": 1.6197125772491896e-06,
      "loss": 1.1024,
      "step": 2924
    },
    {
      "epoch": 0.8712001340307899,
      "grad_norm": 0.6095098257064819,
      "learning_rate": 1.6124474952099067e-06,
      "loss": 1.1018,
      "step": 2925
    },
    {
      "epoch": 0.8714979802304585,
      "grad_norm": 0.6230059862136841,
      "learning_rate": 1.6051973141827381e-06,
      "loss": 1.1112,
      "step": 2926
    },
    {
      "epoch": 0.8717958264301271,
      "grad_norm": 0.6278257369995117,
      "learning_rate": 1.5979620470480861e-06,
      "loss": 1.1137,
      "step": 2927
    },
    {
      "epoch": 0.8720936726297958,
      "grad_norm": 0.6492486000061035,
      "learning_rate": 1.5907417066598562e-06,
      "loss": 1.1324,
      "step": 2928
    },
    {
      "epoch": 0.8723915188294644,
      "grad_norm": 0.6288237571716309,
      "learning_rate": 1.5835363058454368e-06,
      "loss": 1.104,
      "step": 2929
    },
    {
      "epoch": 0.8726893650291331,
      "grad_norm": 0.6501293778419495,
      "learning_rate": 1.576345857405669e-06,
      "loss": 1.1131,
      "step": 2930
    },
    {
      "epoch": 0.8729872112288017,
      "grad_norm": 0.6003174185752869,
      "learning_rate": 1.5691703741148378e-06,
      "loss": 1.0958,
      "step": 2931
    },
    {
      "epoch": 0.8732850574284704,
      "grad_norm": 0.6315889358520508,
      "learning_rate": 1.5620098687206398e-06,
      "loss": 1.1133,
      "step": 2932
    },
    {
      "epoch": 0.8735829036281391,
      "grad_norm": 0.6182097792625427,
      "learning_rate": 1.5548643539441566e-06,
      "loss": 1.1029,
      "step": 2933
    },
    {
      "epoch": 0.8738807498278076,
      "grad_norm": 0.619224488735199,
      "learning_rate": 1.547733842479845e-06,
      "loss": 1.103,
      "step": 2934
    },
    {
      "epoch": 0.8741785960274763,
      "grad_norm": 0.6435920596122742,
      "learning_rate": 1.5406183469955072e-06,
      "loss": 1.1085,
      "step": 2935
    },
    {
      "epoch": 0.8744764422271449,
      "grad_norm": 0.6220390796661377,
      "learning_rate": 1.5335178801322615e-06,
      "loss": 1.1133,
      "step": 2936
    },
    {
      "epoch": 0.8747742884268136,
      "grad_norm": 0.6406397819519043,
      "learning_rate": 1.5264324545045372e-06,
      "loss": 1.1215,
      "step": 2937
    },
    {
      "epoch": 0.8750721346264823,
      "grad_norm": 0.6263743042945862,
      "learning_rate": 1.5193620827000311e-06,
      "loss": 1.0972,
      "step": 2938
    },
    {
      "epoch": 0.8753699808261509,
      "grad_norm": 0.6496587991714478,
      "learning_rate": 1.5123067772797072e-06,
      "loss": 1.1019,
      "step": 2939
    },
    {
      "epoch": 0.8756678270258196,
      "grad_norm": 0.6363325119018555,
      "learning_rate": 1.505266550777751e-06,
      "loss": 1.1182,
      "step": 2940
    },
    {
      "epoch": 0.8759656732254882,
      "grad_norm": 0.6406357884407043,
      "learning_rate": 1.498241415701569e-06,
      "loss": 1.1005,
      "step": 2941
    },
    {
      "epoch": 0.8762635194251568,
      "grad_norm": 0.6097025275230408,
      "learning_rate": 1.491231384531756e-06,
      "loss": 1.1112,
      "step": 2942
    },
    {
      "epoch": 0.8765613656248255,
      "grad_norm": 0.6210368871688843,
      "learning_rate": 1.4842364697220657e-06,
      "loss": 1.0951,
      "step": 2943
    },
    {
      "epoch": 0.8768592118244941,
      "grad_norm": 0.6374768614768982,
      "learning_rate": 1.477256683699404e-06,
      "loss": 1.1044,
      "step": 2944
    },
    {
      "epoch": 0.8771570580241628,
      "grad_norm": 0.6505922079086304,
      "learning_rate": 1.4702920388637998e-06,
      "loss": 1.11,
      "step": 2945
    },
    {
      "epoch": 0.8774549042238314,
      "grad_norm": 0.6403271555900574,
      "learning_rate": 1.4633425475883812e-06,
      "loss": 1.0996,
      "step": 2946
    },
    {
      "epoch": 0.8777527504235001,
      "grad_norm": 0.6128795742988586,
      "learning_rate": 1.456408222219351e-06,
      "loss": 1.1143,
      "step": 2947
    },
    {
      "epoch": 0.8780505966231688,
      "grad_norm": 0.6175402402877808,
      "learning_rate": 1.449489075075974e-06,
      "loss": 1.0844,
      "step": 2948
    },
    {
      "epoch": 0.8783484428228373,
      "grad_norm": 0.6280140280723572,
      "learning_rate": 1.4425851184505501e-06,
      "loss": 1.115,
      "step": 2949
    },
    {
      "epoch": 0.878646289022506,
      "grad_norm": 0.6125122308731079,
      "learning_rate": 1.4356963646083865e-06,
      "loss": 1.1244,
      "step": 2950
    },
    {
      "epoch": 0.8789441352221746,
      "grad_norm": 0.6540136337280273,
      "learning_rate": 1.428822825787789e-06,
      "loss": 1.0965,
      "step": 2951
    },
    {
      "epoch": 0.8792419814218433,
      "grad_norm": 0.6565504670143127,
      "learning_rate": 1.4219645142000293e-06,
      "loss": 1.1253,
      "step": 2952
    },
    {
      "epoch": 0.879539827621512,
      "grad_norm": 0.6296058893203735,
      "learning_rate": 1.4151214420293258e-06,
      "loss": 1.1158,
      "step": 2953
    },
    {
      "epoch": 0.8798376738211806,
      "grad_norm": 0.6475600600242615,
      "learning_rate": 1.4082936214328236e-06,
      "loss": 1.1192,
      "step": 2954
    },
    {
      "epoch": 0.8801355200208493,
      "grad_norm": 0.6430360674858093,
      "learning_rate": 1.4014810645405764e-06,
      "loss": 1.1194,
      "step": 2955
    },
    {
      "epoch": 0.8804333662205179,
      "grad_norm": 0.6358931660652161,
      "learning_rate": 1.3946837834555183e-06,
      "loss": 1.1063,
      "step": 2956
    },
    {
      "epoch": 0.8807312124201865,
      "grad_norm": 0.6485573053359985,
      "learning_rate": 1.387901790253442e-06,
      "loss": 1.1164,
      "step": 2957
    },
    {
      "epoch": 0.8810290586198551,
      "grad_norm": 0.6196106672286987,
      "learning_rate": 1.3811350969829828e-06,
      "loss": 1.0951,
      "step": 2958
    },
    {
      "epoch": 0.8813269048195238,
      "grad_norm": 0.627497673034668,
      "learning_rate": 1.3743837156655992e-06,
      "loss": 1.1139,
      "step": 2959
    },
    {
      "epoch": 0.8816247510191925,
      "grad_norm": 0.6298718452453613,
      "learning_rate": 1.3676476582955378e-06,
      "loss": 1.0986,
      "step": 2960
    },
    {
      "epoch": 0.8819225972188611,
      "grad_norm": 0.6181716918945312,
      "learning_rate": 1.3609269368398281e-06,
      "loss": 1.0928,
      "step": 2961
    },
    {
      "epoch": 0.8822204434185298,
      "grad_norm": 0.6487413048744202,
      "learning_rate": 1.354221563238255e-06,
      "loss": 1.1259,
      "step": 2962
    },
    {
      "epoch": 0.8825182896181984,
      "grad_norm": 0.6475627422332764,
      "learning_rate": 1.3475315494033304e-06,
      "loss": 1.1016,
      "step": 2963
    },
    {
      "epoch": 0.882816135817867,
      "grad_norm": 0.6545590162277222,
      "learning_rate": 1.3408569072202836e-06,
      "loss": 1.1033,
      "step": 2964
    },
    {
      "epoch": 0.8831139820175357,
      "grad_norm": 0.6336657404899597,
      "learning_rate": 1.3341976485470353e-06,
      "loss": 1.1224,
      "step": 2965
    },
    {
      "epoch": 0.8834118282172043,
      "grad_norm": 0.6509076356887817,
      "learning_rate": 1.3275537852141773e-06,
      "loss": 1.1202,
      "step": 2966
    },
    {
      "epoch": 0.883709674416873,
      "grad_norm": 0.6229913830757141,
      "learning_rate": 1.3209253290249447e-06,
      "loss": 1.1151,
      "step": 2967
    },
    {
      "epoch": 0.8840075206165416,
      "grad_norm": 0.6244238018989563,
      "learning_rate": 1.3143122917552077e-06,
      "loss": 1.1175,
      "step": 2968
    },
    {
      "epoch": 0.8843053668162103,
      "grad_norm": 0.6365113854408264,
      "learning_rate": 1.3077146851534417e-06,
      "loss": 1.1081,
      "step": 2969
    },
    {
      "epoch": 0.884603213015879,
      "grad_norm": 0.6344272494316101,
      "learning_rate": 1.3011325209407056e-06,
      "loss": 1.1194,
      "step": 2970
    },
    {
      "epoch": 0.8849010592155476,
      "grad_norm": 0.6259227991104126,
      "learning_rate": 1.2945658108106263e-06,
      "loss": 1.1068,
      "step": 2971
    },
    {
      "epoch": 0.8851989054152162,
      "grad_norm": 0.645117998123169,
      "learning_rate": 1.2880145664293753e-06,
      "loss": 1.1193,
      "step": 2972
    },
    {
      "epoch": 0.8854967516148848,
      "grad_norm": 0.630929172039032,
      "learning_rate": 1.2814787994356515e-06,
      "loss": 1.1001,
      "step": 2973
    },
    {
      "epoch": 0.8857945978145535,
      "grad_norm": 0.6358749270439148,
      "learning_rate": 1.274958521440649e-06,
      "loss": 1.112,
      "step": 2974
    },
    {
      "epoch": 0.8860924440142222,
      "grad_norm": 0.6345223784446716,
      "learning_rate": 1.268453744028053e-06,
      "loss": 1.1163,
      "step": 2975
    },
    {
      "epoch": 0.8863902902138908,
      "grad_norm": 0.653692364692688,
      "learning_rate": 1.2619644787540108e-06,
      "loss": 1.117,
      "step": 2976
    },
    {
      "epoch": 0.8866881364135595,
      "grad_norm": 0.6455109715461731,
      "learning_rate": 1.2554907371471015e-06,
      "loss": 1.1126,
      "step": 2977
    },
    {
      "epoch": 0.8869859826132281,
      "grad_norm": 0.612790584564209,
      "learning_rate": 1.2490325307083363e-06,
      "loss": 1.1026,
      "step": 2978
    },
    {
      "epoch": 0.8872838288128967,
      "grad_norm": 0.6446644067764282,
      "learning_rate": 1.242589870911125e-06,
      "loss": 1.1269,
      "step": 2979
    },
    {
      "epoch": 0.8875816750125654,
      "grad_norm": 0.6364877223968506,
      "learning_rate": 1.2361627692012523e-06,
      "loss": 1.1111,
      "step": 2980
    },
    {
      "epoch": 0.887879521212234,
      "grad_norm": 0.641802966594696,
      "learning_rate": 1.2297512369968678e-06,
      "loss": 1.1196,
      "step": 2981
    },
    {
      "epoch": 0.8881773674119027,
      "grad_norm": 0.646426260471344,
      "learning_rate": 1.223355285688461e-06,
      "loss": 1.1247,
      "step": 2982
    },
    {
      "epoch": 0.8884752136115713,
      "grad_norm": 0.6606464982032776,
      "learning_rate": 1.2169749266388408e-06,
      "loss": 1.1013,
      "step": 2983
    },
    {
      "epoch": 0.88877305981124,
      "grad_norm": 0.620144248008728,
      "learning_rate": 1.2106101711831108e-06,
      "loss": 1.0966,
      "step": 2984
    },
    {
      "epoch": 0.8890709060109087,
      "grad_norm": 0.6316250562667847,
      "learning_rate": 1.2042610306286595e-06,
      "loss": 1.1111,
      "step": 2985
    },
    {
      "epoch": 0.8893687522105773,
      "grad_norm": 0.6289860606193542,
      "learning_rate": 1.197927516255135e-06,
      "loss": 1.1012,
      "step": 2986
    },
    {
      "epoch": 0.8896665984102459,
      "grad_norm": 0.6404206156730652,
      "learning_rate": 1.1916096393144183e-06,
      "loss": 1.12,
      "step": 2987
    },
    {
      "epoch": 0.8899644446099145,
      "grad_norm": 0.634384036064148,
      "learning_rate": 1.1853074110306139e-06,
      "loss": 1.0854,
      "step": 2988
    },
    {
      "epoch": 0.8902622908095832,
      "grad_norm": 0.6221024990081787,
      "learning_rate": 1.1790208426000283e-06,
      "loss": 1.1112,
      "step": 2989
    },
    {
      "epoch": 0.8905601370092519,
      "grad_norm": 0.6264985203742981,
      "learning_rate": 1.1727499451911396e-06,
      "loss": 1.0997,
      "step": 2990
    },
    {
      "epoch": 0.8908579832089205,
      "grad_norm": 0.6340038180351257,
      "learning_rate": 1.1664947299445917e-06,
      "loss": 1.0911,
      "step": 2991
    },
    {
      "epoch": 0.8911558294085892,
      "grad_norm": 0.6352702379226685,
      "learning_rate": 1.1602552079731644e-06,
      "loss": 1.1242,
      "step": 2992
    },
    {
      "epoch": 0.8914536756082578,
      "grad_norm": 0.6384543776512146,
      "learning_rate": 1.154031390361764e-06,
      "loss": 1.1163,
      "step": 2993
    },
    {
      "epoch": 0.8917515218079264,
      "grad_norm": 0.6127960085868835,
      "learning_rate": 1.1478232881673856e-06,
      "loss": 1.0962,
      "step": 2994
    },
    {
      "epoch": 0.892049368007595,
      "grad_norm": 0.6372852325439453,
      "learning_rate": 1.141630912419115e-06,
      "loss": 1.1053,
      "step": 2995
    },
    {
      "epoch": 0.8923472142072637,
      "grad_norm": 0.6500398516654968,
      "learning_rate": 1.1354542741180962e-06,
      "loss": 1.1204,
      "step": 2996
    },
    {
      "epoch": 0.8926450604069324,
      "grad_norm": 0.6289162039756775,
      "learning_rate": 1.1292933842375109e-06,
      "loss": 1.1242,
      "step": 2997
    },
    {
      "epoch": 0.892942906606601,
      "grad_norm": 0.6195435523986816,
      "learning_rate": 1.1231482537225669e-06,
      "loss": 1.1139,
      "step": 2998
    },
    {
      "epoch": 0.8932407528062697,
      "grad_norm": 0.621446967124939,
      "learning_rate": 1.1170188934904758e-06,
      "loss": 1.1097,
      "step": 2999
    },
    {
      "epoch": 0.8935385990059383,
      "grad_norm": 0.6209551692008972,
      "learning_rate": 1.110905314430425e-06,
      "loss": 1.1052,
      "step": 3000
    },
    {
      "epoch": 0.8935385990059383,
      "eval_loss": 1.2393983602523804,
      "eval_runtime": 15.1841,
      "eval_samples_per_second": 114.199,
      "eval_steps_per_second": 14.291,
      "step": 3000
    },
    {
      "epoch": 0.8938364452056069,
      "grad_norm": 0.6350305676460266,
      "learning_rate": 1.104807527403574e-06,
      "loss": 1.1045,
      "step": 3001
    },
    {
      "epoch": 0.8941342914052756,
      "grad_norm": 0.6090103387832642,
      "learning_rate": 1.0987255432430233e-06,
      "loss": 1.0997,
      "step": 3002
    },
    {
      "epoch": 0.8944321376049442,
      "grad_norm": 0.6186074018478394,
      "learning_rate": 1.092659372753796e-06,
      "loss": 1.1007,
      "step": 3003
    },
    {
      "epoch": 0.8947299838046129,
      "grad_norm": 0.6227138638496399,
      "learning_rate": 1.086609026712826e-06,
      "loss": 1.0842,
      "step": 3004
    },
    {
      "epoch": 0.8950278300042815,
      "grad_norm": 0.6527139544487,
      "learning_rate": 1.0805745158689297e-06,
      "loss": 1.1098,
      "step": 3005
    },
    {
      "epoch": 0.8953256762039502,
      "grad_norm": 0.6440016031265259,
      "learning_rate": 1.0745558509427968e-06,
      "loss": 1.116,
      "step": 3006
    },
    {
      "epoch": 0.8956235224036189,
      "grad_norm": 0.6108453869819641,
      "learning_rate": 1.0685530426269574e-06,
      "loss": 1.0947,
      "step": 3007
    },
    {
      "epoch": 0.8959213686032875,
      "grad_norm": 0.6392330527305603,
      "learning_rate": 1.0625661015857802e-06,
      "loss": 1.1344,
      "step": 3008
    },
    {
      "epoch": 0.8962192148029561,
      "grad_norm": 0.6233141422271729,
      "learning_rate": 1.056595038455438e-06,
      "loss": 1.1003,
      "step": 3009
    },
    {
      "epoch": 0.8965170610026247,
      "grad_norm": 0.6284676790237427,
      "learning_rate": 1.050639863843903e-06,
      "loss": 1.1115,
      "step": 3010
    },
    {
      "epoch": 0.8968149072022934,
      "grad_norm": 0.6428810358047485,
      "learning_rate": 1.0447005883309103e-06,
      "loss": 1.1124,
      "step": 3011
    },
    {
      "epoch": 0.8971127534019621,
      "grad_norm": 0.6229919791221619,
      "learning_rate": 1.0387772224679572e-06,
      "loss": 1.0954,
      "step": 3012
    },
    {
      "epoch": 0.8974105996016307,
      "grad_norm": 0.6367018818855286,
      "learning_rate": 1.0328697767782748e-06,
      "loss": 1.1076,
      "step": 3013
    },
    {
      "epoch": 0.8977084458012994,
      "grad_norm": 0.6084621548652649,
      "learning_rate": 1.026978261756808e-06,
      "loss": 1.0965,
      "step": 3014
    },
    {
      "epoch": 0.898006292000968,
      "grad_norm": 0.6289997696876526,
      "learning_rate": 1.0211026878702024e-06,
      "loss": 1.0905,
      "step": 3015
    },
    {
      "epoch": 0.8983041382006366,
      "grad_norm": 0.6507461667060852,
      "learning_rate": 1.0152430655567858e-06,
      "loss": 1.1011,
      "step": 3016
    },
    {
      "epoch": 0.8986019844003053,
      "grad_norm": 0.6312487721443176,
      "learning_rate": 1.009399405226541e-06,
      "loss": 1.1228,
      "step": 3017
    },
    {
      "epoch": 0.8988998305999739,
      "grad_norm": 0.6332783699035645,
      "learning_rate": 1.003571717261096e-06,
      "loss": 1.1151,
      "step": 3018
    },
    {
      "epoch": 0.8991976767996426,
      "grad_norm": 0.6212210059165955,
      "learning_rate": 9.977600120137054e-07,
      "loss": 1.0901,
      "step": 3019
    },
    {
      "epoch": 0.8994955229993112,
      "grad_norm": 0.6225468516349792,
      "learning_rate": 9.919642998092284e-07,
      "loss": 1.0985,
      "step": 3020
    },
    {
      "epoch": 0.8997933691989799,
      "grad_norm": 0.6402681469917297,
      "learning_rate": 9.861845909441059e-07,
      "loss": 1.1204,
      "step": 3021
    },
    {
      "epoch": 0.9000912153986486,
      "grad_norm": 0.6287094950675964,
      "learning_rate": 9.80420895686356e-07,
      "loss": 1.1079,
      "step": 3022
    },
    {
      "epoch": 0.9003890615983172,
      "grad_norm": 0.6275615096092224,
      "learning_rate": 9.746732242755464e-07,
      "loss": 1.1085,
      "step": 3023
    },
    {
      "epoch": 0.9006869077979858,
      "grad_norm": 0.6476185917854309,
      "learning_rate": 9.689415869227692e-07,
      "loss": 1.1287,
      "step": 3024
    },
    {
      "epoch": 0.9009847539976544,
      "grad_norm": 0.6304489970207214,
      "learning_rate": 9.632259938106403e-07,
      "loss": 1.1191,
      "step": 3025
    },
    {
      "epoch": 0.9012826001973231,
      "grad_norm": 0.6205864548683167,
      "learning_rate": 9.575264550932705e-07,
      "loss": 1.1118,
      "step": 3026
    },
    {
      "epoch": 0.9015804463969918,
      "grad_norm": 0.6337959170341492,
      "learning_rate": 9.518429808962438e-07,
      "loss": 1.1152,
      "step": 3027
    },
    {
      "epoch": 0.9018782925966604,
      "grad_norm": 0.6122902035713196,
      "learning_rate": 9.461755813166085e-07,
      "loss": 1.1033,
      "step": 3028
    },
    {
      "epoch": 0.9021761387963291,
      "grad_norm": 0.6495623588562012,
      "learning_rate": 9.405242664228576e-07,
      "loss": 1.1176,
      "step": 3029
    },
    {
      "epoch": 0.9024739849959977,
      "grad_norm": 0.6257151365280151,
      "learning_rate": 9.348890462549021e-07,
      "loss": 1.0939,
      "step": 3030
    },
    {
      "epoch": 0.9027718311956663,
      "grad_norm": 0.6587477326393127,
      "learning_rate": 9.292699308240649e-07,
      "loss": 1.1214,
      "step": 3031
    },
    {
      "epoch": 0.903069677395335,
      "grad_norm": 0.6076934337615967,
      "learning_rate": 9.236669301130563e-07,
      "loss": 1.1072,
      "step": 3032
    },
    {
      "epoch": 0.9033675235950036,
      "grad_norm": 0.6269726753234863,
      "learning_rate": 9.180800540759604e-07,
      "loss": 1.1254,
      "step": 3033
    },
    {
      "epoch": 0.9036653697946723,
      "grad_norm": 0.6333617568016052,
      "learning_rate": 9.125093126382078e-07,
      "loss": 1.1011,
      "step": 3034
    },
    {
      "epoch": 0.9039632159943409,
      "grad_norm": 0.6375496983528137,
      "learning_rate": 9.069547156965708e-07,
      "loss": 1.1091,
      "step": 3035
    },
    {
      "epoch": 0.9042610621940096,
      "grad_norm": 0.6122974157333374,
      "learning_rate": 9.014162731191411e-07,
      "loss": 1.0977,
      "step": 3036
    },
    {
      "epoch": 0.9045589083936783,
      "grad_norm": 0.6412339806556702,
      "learning_rate": 8.95893994745306e-07,
      "loss": 1.1122,
      "step": 3037
    },
    {
      "epoch": 0.9048567545933469,
      "grad_norm": 0.6411993503570557,
      "learning_rate": 8.90387890385741e-07,
      "loss": 1.1114,
      "step": 3038
    },
    {
      "epoch": 0.9051546007930155,
      "grad_norm": 0.6309710144996643,
      "learning_rate": 8.848979698223814e-07,
      "loss": 1.0882,
      "step": 3039
    },
    {
      "epoch": 0.9054524469926841,
      "grad_norm": 0.6223629713058472,
      "learning_rate": 8.794242428084198e-07,
      "loss": 1.0885,
      "step": 3040
    },
    {
      "epoch": 0.9057502931923528,
      "grad_norm": 0.6255239844322205,
      "learning_rate": 8.739667190682699e-07,
      "loss": 1.111,
      "step": 3041
    },
    {
      "epoch": 0.9060481393920214,
      "grad_norm": 0.6251360774040222,
      "learning_rate": 8.685254082975669e-07,
      "loss": 1.1119,
      "step": 3042
    },
    {
      "epoch": 0.9063459855916901,
      "grad_norm": 0.6260493397712708,
      "learning_rate": 8.631003201631405e-07,
      "loss": 1.1105,
      "step": 3043
    },
    {
      "epoch": 0.9066438317913588,
      "grad_norm": 0.6468093991279602,
      "learning_rate": 8.576914643029977e-07,
      "loss": 1.1185,
      "step": 3044
    },
    {
      "epoch": 0.9069416779910274,
      "grad_norm": 0.641973078250885,
      "learning_rate": 8.522988503263108e-07,
      "loss": 1.1038,
      "step": 3045
    },
    {
      "epoch": 0.907239524190696,
      "grad_norm": 0.6205921769142151,
      "learning_rate": 8.469224878133953e-07,
      "loss": 1.0939,
      "step": 3046
    },
    {
      "epoch": 0.9075373703903646,
      "grad_norm": 0.636685311794281,
      "learning_rate": 8.415623863156985e-07,
      "loss": 1.1125,
      "step": 3047
    },
    {
      "epoch": 0.9078352165900333,
      "grad_norm": 0.623885452747345,
      "learning_rate": 8.362185553557723e-07,
      "loss": 1.1067,
      "step": 3048
    },
    {
      "epoch": 0.908133062789702,
      "grad_norm": 0.635087251663208,
      "learning_rate": 8.308910044272689e-07,
      "loss": 1.1022,
      "step": 3049
    },
    {
      "epoch": 0.9084309089893706,
      "grad_norm": 0.6462403535842896,
      "learning_rate": 8.255797429949175e-07,
      "loss": 1.1116,
      "step": 3050
    },
    {
      "epoch": 0.9087287551890393,
      "grad_norm": 0.6541346311569214,
      "learning_rate": 8.202847804945025e-07,
      "loss": 1.1395,
      "step": 3051
    },
    {
      "epoch": 0.9090266013887079,
      "grad_norm": 0.6225289106369019,
      "learning_rate": 8.150061263328601e-07,
      "loss": 1.1242,
      "step": 3052
    },
    {
      "epoch": 0.9093244475883766,
      "grad_norm": 0.6424941420555115,
      "learning_rate": 8.097437898878491e-07,
      "loss": 1.1179,
      "step": 3053
    },
    {
      "epoch": 0.9096222937880452,
      "grad_norm": 0.6417063474655151,
      "learning_rate": 8.044977805083376e-07,
      "loss": 1.1129,
      "step": 3054
    },
    {
      "epoch": 0.9099201399877138,
      "grad_norm": 0.6244187951087952,
      "learning_rate": 7.992681075141906e-07,
      "loss": 1.094,
      "step": 3055
    },
    {
      "epoch": 0.9102179861873825,
      "grad_norm": 0.644004225730896,
      "learning_rate": 7.940547801962506e-07,
      "loss": 1.1208,
      "step": 3056
    },
    {
      "epoch": 0.9105158323870511,
      "grad_norm": 0.6287850737571716,
      "learning_rate": 7.888578078163211e-07,
      "loss": 1.1276,
      "step": 3057
    },
    {
      "epoch": 0.9108136785867198,
      "grad_norm": 0.6331035494804382,
      "learning_rate": 7.836771996071457e-07,
      "loss": 1.1185,
      "step": 3058
    },
    {
      "epoch": 0.9111115247863885,
      "grad_norm": 0.6252988576889038,
      "learning_rate": 7.78512964772401e-07,
      "loss": 1.1019,
      "step": 3059
    },
    {
      "epoch": 0.9114093709860571,
      "grad_norm": 0.6192378401756287,
      "learning_rate": 7.733651124866736e-07,
      "loss": 1.1077,
      "step": 3060
    },
    {
      "epoch": 0.9117072171857257,
      "grad_norm": 0.6319203972816467,
      "learning_rate": 7.682336518954447e-07,
      "loss": 1.1056,
      "step": 3061
    },
    {
      "epoch": 0.9120050633853943,
      "grad_norm": 0.6407369375228882,
      "learning_rate": 7.631185921150741e-07,
      "loss": 1.1348,
      "step": 3062
    },
    {
      "epoch": 0.912302909585063,
      "grad_norm": 0.6226566433906555,
      "learning_rate": 7.580199422327905e-07,
      "loss": 1.1212,
      "step": 3063
    },
    {
      "epoch": 0.9126007557847317,
      "grad_norm": 0.6152005195617676,
      "learning_rate": 7.529377113066582e-07,
      "loss": 1.1136,
      "step": 3064
    },
    {
      "epoch": 0.9128986019844003,
      "grad_norm": 0.6494110822677612,
      "learning_rate": 7.478719083655827e-07,
      "loss": 1.1045,
      "step": 3065
    },
    {
      "epoch": 0.913196448184069,
      "grad_norm": 0.6177676320075989,
      "learning_rate": 7.428225424092794e-07,
      "loss": 1.1202,
      "step": 3066
    },
    {
      "epoch": 0.9134942943837376,
      "grad_norm": 0.6392973065376282,
      "learning_rate": 7.377896224082626e-07,
      "loss": 1.1125,
      "step": 3067
    },
    {
      "epoch": 0.9137921405834063,
      "grad_norm": 0.654961347579956,
      "learning_rate": 7.327731573038288e-07,
      "loss": 1.1161,
      "step": 3068
    },
    {
      "epoch": 0.9140899867830748,
      "grad_norm": 0.6143835783004761,
      "learning_rate": 7.277731560080436e-07,
      "loss": 1.0998,
      "step": 3069
    },
    {
      "epoch": 0.9143878329827435,
      "grad_norm": 0.6157786250114441,
      "learning_rate": 7.227896274037238e-07,
      "loss": 1.1044,
      "step": 3070
    },
    {
      "epoch": 0.9146856791824122,
      "grad_norm": 0.6683198809623718,
      "learning_rate": 7.178225803444183e-07,
      "loss": 1.0906,
      "step": 3071
    },
    {
      "epoch": 0.9149835253820808,
      "grad_norm": 0.611257016658783,
      "learning_rate": 7.128720236543951e-07,
      "loss": 1.0946,
      "step": 3072
    },
    {
      "epoch": 0.9152813715817495,
      "grad_norm": 0.6340640783309937,
      "learning_rate": 7.079379661286301e-07,
      "loss": 1.1197,
      "step": 3073
    },
    {
      "epoch": 0.9155792177814182,
      "grad_norm": 0.6264047622680664,
      "learning_rate": 7.030204165327869e-07,
      "loss": 1.0931,
      "step": 3074
    },
    {
      "epoch": 0.9158770639810868,
      "grad_norm": 0.6438785791397095,
      "learning_rate": 6.981193836031974e-07,
      "loss": 1.1037,
      "step": 3075
    },
    {
      "epoch": 0.9161749101807554,
      "grad_norm": 0.6159545183181763,
      "learning_rate": 6.932348760468543e-07,
      "loss": 1.109,
      "step": 3076
    },
    {
      "epoch": 0.916472756380424,
      "grad_norm": 0.6369842290878296,
      "learning_rate": 6.883669025413942e-07,
      "loss": 1.1006,
      "step": 3077
    },
    {
      "epoch": 0.9167706025800927,
      "grad_norm": 0.6296409368515015,
      "learning_rate": 6.835154717350712e-07,
      "loss": 1.1249,
      "step": 3078
    },
    {
      "epoch": 0.9170684487797613,
      "grad_norm": 0.6328773498535156,
      "learning_rate": 6.786805922467588e-07,
      "loss": 1.103,
      "step": 3079
    },
    {
      "epoch": 0.91736629497943,
      "grad_norm": 0.649982213973999,
      "learning_rate": 6.73862272665925e-07,
      "loss": 1.1025,
      "step": 3080
    },
    {
      "epoch": 0.9176641411790987,
      "grad_norm": 0.6266394853591919,
      "learning_rate": 6.690605215526114e-07,
      "loss": 1.1046,
      "step": 3081
    },
    {
      "epoch": 0.9179619873787673,
      "grad_norm": 0.6444719433784485,
      "learning_rate": 6.642753474374331e-07,
      "loss": 1.113,
      "step": 3082
    },
    {
      "epoch": 0.9182598335784359,
      "grad_norm": 0.5989339351654053,
      "learning_rate": 6.595067588215509e-07,
      "loss": 1.1024,
      "step": 3083
    },
    {
      "epoch": 0.9185576797781045,
      "grad_norm": 0.6209490895271301,
      "learning_rate": 6.547547641766616e-07,
      "loss": 1.103,
      "step": 3084
    },
    {
      "epoch": 0.9188555259777732,
      "grad_norm": 0.6198128461837769,
      "learning_rate": 6.500193719449787e-07,
      "loss": 1.1058,
      "step": 3085
    },
    {
      "epoch": 0.9191533721774419,
      "grad_norm": 0.6332685947418213,
      "learning_rate": 6.453005905392251e-07,
      "loss": 1.1072,
      "step": 3086
    },
    {
      "epoch": 0.9194512183771105,
      "grad_norm": 0.6391420960426331,
      "learning_rate": 6.405984283426125e-07,
      "loss": 1.1097,
      "step": 3087
    },
    {
      "epoch": 0.9197490645767792,
      "grad_norm": 0.6304422616958618,
      "learning_rate": 6.359128937088222e-07,
      "loss": 1.1213,
      "step": 3088
    },
    {
      "epoch": 0.9200469107764478,
      "grad_norm": 0.6432411670684814,
      "learning_rate": 6.312439949620031e-07,
      "loss": 1.1226,
      "step": 3089
    },
    {
      "epoch": 0.9203447569761165,
      "grad_norm": 0.6412847638130188,
      "learning_rate": 6.26591740396747e-07,
      "loss": 1.1089,
      "step": 3090
    },
    {
      "epoch": 0.9206426031757851,
      "grad_norm": 0.6170780658721924,
      "learning_rate": 6.219561382780715e-07,
      "loss": 1.1152,
      "step": 3091
    },
    {
      "epoch": 0.9209404493754537,
      "grad_norm": 0.6374718546867371,
      "learning_rate": 6.173371968414165e-07,
      "loss": 1.1132,
      "step": 3092
    },
    {
      "epoch": 0.9212382955751224,
      "grad_norm": 0.6159054040908813,
      "learning_rate": 6.127349242926217e-07,
      "loss": 1.1016,
      "step": 3093
    },
    {
      "epoch": 0.921536141774791,
      "grad_norm": 0.6337563991546631,
      "learning_rate": 6.081493288079099e-07,
      "loss": 1.0988,
      "step": 3094
    },
    {
      "epoch": 0.9218339879744597,
      "grad_norm": 0.6211062669754028,
      "learning_rate": 6.03580418533879e-07,
      "loss": 1.1162,
      "step": 3095
    },
    {
      "epoch": 0.9221318341741284,
      "grad_norm": 0.6577699780464172,
      "learning_rate": 5.99028201587486e-07,
      "loss": 1.1354,
      "step": 3096
    },
    {
      "epoch": 0.922429680373797,
      "grad_norm": 0.6113946437835693,
      "learning_rate": 5.944926860560285e-07,
      "loss": 1.0964,
      "step": 3097
    },
    {
      "epoch": 0.9227275265734656,
      "grad_norm": 0.623714029788971,
      "learning_rate": 5.899738799971321e-07,
      "loss": 1.1086,
      "step": 3098
    },
    {
      "epoch": 0.9230253727731342,
      "grad_norm": 0.6342601776123047,
      "learning_rate": 5.854717914387398e-07,
      "loss": 1.1215,
      "step": 3099
    },
    {
      "epoch": 0.9233232189728029,
      "grad_norm": 0.6229255795478821,
      "learning_rate": 5.809864283790956e-07,
      "loss": 1.0928,
      "step": 3100
    },
    {
      "epoch": 0.9236210651724716,
      "grad_norm": 0.6384140849113464,
      "learning_rate": 5.765177987867255e-07,
      "loss": 1.1202,
      "step": 3101
    },
    {
      "epoch": 0.9239189113721402,
      "grad_norm": 0.6185979247093201,
      "learning_rate": 5.720659106004289e-07,
      "loss": 1.1046,
      "step": 3102
    },
    {
      "epoch": 0.9242167575718089,
      "grad_norm": 0.6437633633613586,
      "learning_rate": 5.676307717292673e-07,
      "loss": 1.0935,
      "step": 3103
    },
    {
      "epoch": 0.9245146037714775,
      "grad_norm": 0.6350952982902527,
      "learning_rate": 5.632123900525388e-07,
      "loss": 1.1148,
      "step": 3104
    },
    {
      "epoch": 0.9248124499711462,
      "grad_norm": 0.6284207701683044,
      "learning_rate": 5.58810773419779e-07,
      "loss": 1.127,
      "step": 3105
    },
    {
      "epoch": 0.9251102961708147,
      "grad_norm": 0.6169633269309998,
      "learning_rate": 5.544259296507314e-07,
      "loss": 1.121,
      "step": 3106
    },
    {
      "epoch": 0.9254081423704834,
      "grad_norm": 0.6315242648124695,
      "learning_rate": 5.500578665353484e-07,
      "loss": 1.1049,
      "step": 3107
    },
    {
      "epoch": 0.9257059885701521,
      "grad_norm": 0.6418565511703491,
      "learning_rate": 5.457065918337645e-07,
      "loss": 1.1054,
      "step": 3108
    },
    {
      "epoch": 0.9260038347698207,
      "grad_norm": 0.6239096522331238,
      "learning_rate": 5.41372113276295e-07,
      "loss": 1.1132,
      "step": 3109
    },
    {
      "epoch": 0.9263016809694894,
      "grad_norm": 0.6422299742698669,
      "learning_rate": 5.370544385634102e-07,
      "loss": 1.0933,
      "step": 3110
    },
    {
      "epoch": 0.926599527169158,
      "grad_norm": 0.6410685777664185,
      "learning_rate": 5.327535753657309e-07,
      "loss": 1.1154,
      "step": 3111
    },
    {
      "epoch": 0.9268973733688267,
      "grad_norm": 0.6194297075271606,
      "learning_rate": 5.284695313240096e-07,
      "loss": 1.1127,
      "step": 3112
    },
    {
      "epoch": 0.9271952195684953,
      "grad_norm": 0.6170351505279541,
      "learning_rate": 5.24202314049117e-07,
      "loss": 1.1052,
      "step": 3113
    },
    {
      "epoch": 0.9274930657681639,
      "grad_norm": 0.6435105800628662,
      "learning_rate": 5.199519311220347e-07,
      "loss": 1.1084,
      "step": 3114
    },
    {
      "epoch": 0.9277909119678326,
      "grad_norm": 0.6373679041862488,
      "learning_rate": 5.157183900938311e-07,
      "loss": 1.1175,
      "step": 3115
    },
    {
      "epoch": 0.9280887581675012,
      "grad_norm": 0.6320210099220276,
      "learning_rate": 5.11501698485658e-07,
      "loss": 1.1165,
      "step": 3116
    },
    {
      "epoch": 0.9283866043671699,
      "grad_norm": 0.6280572414398193,
      "learning_rate": 5.073018637887339e-07,
      "loss": 1.0874,
      "step": 3117
    },
    {
      "epoch": 0.9286844505668386,
      "grad_norm": 0.6119697093963623,
      "learning_rate": 5.03118893464325e-07,
      "loss": 1.1021,
      "step": 3118
    },
    {
      "epoch": 0.9289822967665072,
      "grad_norm": 0.6185587644577026,
      "learning_rate": 4.989527949437411e-07,
      "loss": 1.1019,
      "step": 3119
    },
    {
      "epoch": 0.9292801429661759,
      "grad_norm": 0.6228737235069275,
      "learning_rate": 4.948035756283198e-07,
      "loss": 1.1124,
      "step": 3120
    },
    {
      "epoch": 0.9295779891658444,
      "grad_norm": 0.6466836333274841,
      "learning_rate": 4.906712428894033e-07,
      "loss": 1.124,
      "step": 3121
    },
    {
      "epoch": 0.9298758353655131,
      "grad_norm": 0.6354174613952637,
      "learning_rate": 4.865558040683438e-07,
      "loss": 1.1263,
      "step": 3122
    },
    {
      "epoch": 0.9301736815651818,
      "grad_norm": 0.624755859375,
      "learning_rate": 4.824572664764748e-07,
      "loss": 1.0812,
      "step": 3123
    },
    {
      "epoch": 0.9304715277648504,
      "grad_norm": 0.6278039813041687,
      "learning_rate": 4.783756373951054e-07,
      "loss": 1.1065,
      "step": 3124
    },
    {
      "epoch": 0.9307693739645191,
      "grad_norm": 0.6287121176719666,
      "learning_rate": 4.7431092407550397e-07,
      "loss": 1.1061,
      "step": 3125
    },
    {
      "epoch": 0.9310672201641877,
      "grad_norm": 0.6353144645690918,
      "learning_rate": 4.7026313373888856e-07,
      "loss": 1.1223,
      "step": 3126
    },
    {
      "epoch": 0.9313650663638564,
      "grad_norm": 0.6300527453422546,
      "learning_rate": 4.6623227357641466e-07,
      "loss": 1.1055,
      "step": 3127
    },
    {
      "epoch": 0.931662912563525,
      "grad_norm": 0.6557642221450806,
      "learning_rate": 4.622183507491529e-07,
      "loss": 1.1324,
      "step": 3128
    },
    {
      "epoch": 0.9319607587631936,
      "grad_norm": 0.6094654202461243,
      "learning_rate": 4.5822137238809126e-07,
      "loss": 1.1168,
      "step": 3129
    },
    {
      "epoch": 0.9322586049628623,
      "grad_norm": 0.6215384006500244,
      "learning_rate": 4.5424134559411413e-07,
      "loss": 1.105,
      "step": 3130
    },
    {
      "epoch": 0.9325564511625309,
      "grad_norm": 0.614746630191803,
      "learning_rate": 4.5027827743798435e-07,
      "loss": 1.1038,
      "step": 3131
    },
    {
      "epoch": 0.9328542973621996,
      "grad_norm": 0.6315264701843262,
      "learning_rate": 4.4633217496034107e-07,
      "loss": 1.1242,
      "step": 3132
    },
    {
      "epoch": 0.9331521435618683,
      "grad_norm": 0.6403162479400635,
      "learning_rate": 4.424030451716843e-07,
      "loss": 1.1034,
      "step": 3133
    },
    {
      "epoch": 0.9334499897615369,
      "grad_norm": 0.6402631998062134,
      "learning_rate": 4.384908950523603e-07,
      "loss": 1.1221,
      "step": 3134
    },
    {
      "epoch": 0.9337478359612056,
      "grad_norm": 0.6208403706550598,
      "learning_rate": 4.3459573155254396e-07,
      "loss": 1.1216,
      "step": 3135
    },
    {
      "epoch": 0.9340456821608741,
      "grad_norm": 0.6113629341125488,
      "learning_rate": 4.3071756159224096e-07,
      "loss": 1.1038,
      "step": 3136
    },
    {
      "epoch": 0.9343435283605428,
      "grad_norm": 0.6284322142601013,
      "learning_rate": 4.268563920612623e-07,
      "loss": 1.1054,
      "step": 3137
    },
    {
      "epoch": 0.9346413745602115,
      "grad_norm": 0.616893470287323,
      "learning_rate": 4.230122298192163e-07,
      "loss": 1.1228,
      "step": 3138
    },
    {
      "epoch": 0.9349392207598801,
      "grad_norm": 0.6393241882324219,
      "learning_rate": 4.191850816955001e-07,
      "loss": 1.1144,
      "step": 3139
    },
    {
      "epoch": 0.9352370669595488,
      "grad_norm": 0.6272119879722595,
      "learning_rate": 4.153749544892782e-07,
      "loss": 1.1115,
      "step": 3140
    },
    {
      "epoch": 0.9355349131592174,
      "grad_norm": 0.6323702335357666,
      "learning_rate": 4.115818549694839e-07,
      "loss": 1.1088,
      "step": 3141
    },
    {
      "epoch": 0.9358327593588861,
      "grad_norm": 0.6250095367431641,
      "learning_rate": 4.0780578987479225e-07,
      "loss": 1.1049,
      "step": 3142
    },
    {
      "epoch": 0.9361306055585547,
      "grad_norm": 0.626417338848114,
      "learning_rate": 4.040467659136227e-07,
      "loss": 1.1125,
      "step": 3143
    },
    {
      "epoch": 0.9364284517582233,
      "grad_norm": 0.6197050213813782,
      "learning_rate": 4.003047897641155e-07,
      "loss": 1.1187,
      "step": 3144
    },
    {
      "epoch": 0.936726297957892,
      "grad_norm": 0.6203882098197937,
      "learning_rate": 3.96579868074124e-07,
      "loss": 1.1023,
      "step": 3145
    },
    {
      "epoch": 0.9370241441575606,
      "grad_norm": 0.6324443221092224,
      "learning_rate": 3.928720074612069e-07,
      "loss": 1.1195,
      "step": 3146
    },
    {
      "epoch": 0.9373219903572293,
      "grad_norm": 0.6441107988357544,
      "learning_rate": 3.8918121451260947e-07,
      "loss": 1.1171,
      "step": 3147
    },
    {
      "epoch": 0.937619836556898,
      "grad_norm": 0.6024614572525024,
      "learning_rate": 3.855074957852578e-07,
      "loss": 1.1071,
      "step": 3148
    },
    {
      "epoch": 0.9379176827565666,
      "grad_norm": 0.6250506639480591,
      "learning_rate": 3.818508578057412e-07,
      "loss": 1.1229,
      "step": 3149
    },
    {
      "epoch": 0.9382155289562353,
      "grad_norm": 0.638670802116394,
      "learning_rate": 3.782113070703075e-07,
      "loss": 1.1135,
      "step": 3150
    },
    {
      "epoch": 0.9385133751559038,
      "grad_norm": 0.6242445111274719,
      "learning_rate": 3.7458885004484693e-07,
      "loss": 1.1122,
      "step": 3151
    },
    {
      "epoch": 0.9388112213555725,
      "grad_norm": 0.6038979291915894,
      "learning_rate": 3.7098349316487816e-07,
      "loss": 1.1021,
      "step": 3152
    },
    {
      "epoch": 0.9391090675552412,
      "grad_norm": 0.6449660658836365,
      "learning_rate": 3.673952428355465e-07,
      "loss": 1.1139,
      "step": 3153
    },
    {
      "epoch": 0.9394069137549098,
      "grad_norm": 0.6329522728919983,
      "learning_rate": 3.638241054316027e-07,
      "loss": 1.0958,
      "step": 3154
    },
    {
      "epoch": 0.9397047599545785,
      "grad_norm": 0.6268729567527771,
      "learning_rate": 3.602700872973952e-07,
      "loss": 1.1198,
      "step": 3155
    },
    {
      "epoch": 0.9400026061542471,
      "grad_norm": 0.6201335787773132,
      "learning_rate": 3.5673319474685885e-07,
      "loss": 1.1051,
      "step": 3156
    },
    {
      "epoch": 0.9403004523539158,
      "grad_norm": 0.6330424547195435,
      "learning_rate": 3.532134340635085e-07,
      "loss": 1.1035,
      "step": 3157
    },
    {
      "epoch": 0.9405982985535843,
      "grad_norm": 0.6249594688415527,
      "learning_rate": 3.497108115004144e-07,
      "loss": 1.1014,
      "step": 3158
    },
    {
      "epoch": 0.940896144753253,
      "grad_norm": 0.6228363513946533,
      "learning_rate": 3.462253332802068e-07,
      "loss": 1.128,
      "step": 3159
    },
    {
      "epoch": 0.9411939909529217,
      "grad_norm": 0.6407414078712463,
      "learning_rate": 3.4275700559505687e-07,
      "loss": 1.1006,
      "step": 3160
    },
    {
      "epoch": 0.9414918371525903,
      "grad_norm": 0.6436894536018372,
      "learning_rate": 3.3930583460666576e-07,
      "loss": 1.082,
      "step": 3161
    },
    {
      "epoch": 0.941789683352259,
      "grad_norm": 0.6276578903198242,
      "learning_rate": 3.358718264462535e-07,
      "loss": 1.1069,
      "step": 3162
    },
    {
      "epoch": 0.9420875295519276,
      "grad_norm": 0.6382056474685669,
      "learning_rate": 3.3245498721455104e-07,
      "loss": 1.1069,
      "step": 3163
    },
    {
      "epoch": 0.9423853757515963,
      "grad_norm": 0.6349144577980042,
      "learning_rate": 3.290553229817894e-07,
      "loss": 1.1007,
      "step": 3164
    },
    {
      "epoch": 0.9426832219512649,
      "grad_norm": 0.641930878162384,
      "learning_rate": 3.256728397876807e-07,
      "loss": 1.1109,
      "step": 3165
    },
    {
      "epoch": 0.9429810681509335,
      "grad_norm": 0.6366417407989502,
      "learning_rate": 3.223075436414214e-07,
      "loss": 1.1258,
      "step": 3166
    },
    {
      "epoch": 0.9432789143506022,
      "grad_norm": 0.64521723985672,
      "learning_rate": 3.1895944052167004e-07,
      "loss": 1.119,
      "step": 3167
    },
    {
      "epoch": 0.9435767605502708,
      "grad_norm": 0.6396946310997009,
      "learning_rate": 3.15628536376541e-07,
      "loss": 1.1057,
      "step": 3168
    },
    {
      "epoch": 0.9438746067499395,
      "grad_norm": 0.6097322106361389,
      "learning_rate": 3.123148371235929e-07,
      "loss": 1.1022,
      "step": 3169
    },
    {
      "epoch": 0.9441724529496082,
      "grad_norm": 0.6136809587478638,
      "learning_rate": 3.0901834864982217e-07,
      "loss": 1.1043,
      "step": 3170
    },
    {
      "epoch": 0.9444702991492768,
      "grad_norm": 0.6105096936225891,
      "learning_rate": 3.057390768116475e-07,
      "loss": 1.0907,
      "step": 3171
    },
    {
      "epoch": 0.9447681453489455,
      "grad_norm": 0.6306178569793701,
      "learning_rate": 3.024770274348976e-07,
      "loss": 1.1286,
      "step": 3172
    },
    {
      "epoch": 0.945065991548614,
      "grad_norm": 0.6353450417518616,
      "learning_rate": 2.99232206314809e-07,
      "loss": 1.1046,
      "step": 3173
    },
    {
      "epoch": 0.9453638377482827,
      "grad_norm": 0.6323910355567932,
      "learning_rate": 2.960046192160082e-07,
      "loss": 1.1045,
      "step": 3174
    },
    {
      "epoch": 0.9456616839479514,
      "grad_norm": 0.6275209188461304,
      "learning_rate": 2.9279427187250963e-07,
      "loss": 1.0952,
      "step": 3175
    },
    {
      "epoch": 0.94595953014762,
      "grad_norm": 0.6380491256713867,
      "learning_rate": 2.8960116998769103e-07,
      "loss": 1.117,
      "step": 3176
    },
    {
      "epoch": 0.9462573763472887,
      "grad_norm": 0.6314483880996704,
      "learning_rate": 2.864253192343014e-07,
      "loss": 1.1186,
      "step": 3177
    },
    {
      "epoch": 0.9465552225469573,
      "grad_norm": 0.6177024841308594,
      "learning_rate": 2.8326672525443853e-07,
      "loss": 1.1158,
      "step": 3178
    },
    {
      "epoch": 0.946853068746626,
      "grad_norm": 0.6190733313560486,
      "learning_rate": 2.8012539365953937e-07,
      "loss": 1.0992,
      "step": 3179
    },
    {
      "epoch": 0.9471509149462946,
      "grad_norm": 0.6492108106613159,
      "learning_rate": 2.7700133003037864e-07,
      "loss": 1.1157,
      "step": 3180
    },
    {
      "epoch": 0.9474487611459632,
      "grad_norm": 0.6259175539016724,
      "learning_rate": 2.7389453991705226e-07,
      "loss": 1.0932,
      "step": 3181
    },
    {
      "epoch": 0.9477466073456319,
      "grad_norm": 0.6463065147399902,
      "learning_rate": 2.70805028838963e-07,
      "loss": 1.1119,
      "step": 3182
    },
    {
      "epoch": 0.9480444535453005,
      "grad_norm": 0.6355845332145691,
      "learning_rate": 2.677328022848236e-07,
      "loss": 1.1141,
      "step": 3183
    },
    {
      "epoch": 0.9483422997449692,
      "grad_norm": 0.6375442743301392,
      "learning_rate": 2.646778657126381e-07,
      "loss": 1.1044,
      "step": 3184
    },
    {
      "epoch": 0.9486401459446379,
      "grad_norm": 0.6302236318588257,
      "learning_rate": 2.616402245496896e-07,
      "loss": 1.1034,
      "step": 3185
    },
    {
      "epoch": 0.9489379921443065,
      "grad_norm": 0.6385701894760132,
      "learning_rate": 2.5861988419253914e-07,
      "loss": 1.1092,
      "step": 3186
    },
    {
      "epoch": 0.9492358383439752,
      "grad_norm": 0.6214167475700378,
      "learning_rate": 2.5561685000700996e-07,
      "loss": 1.1128,
      "step": 3187
    },
    {
      "epoch": 0.9495336845436437,
      "grad_norm": 0.6344258785247803,
      "learning_rate": 2.526311273281823e-07,
      "loss": 1.1076,
      "step": 3188
    },
    {
      "epoch": 0.9498315307433124,
      "grad_norm": 0.6324419975280762,
      "learning_rate": 2.496627214603764e-07,
      "loss": 1.1135,
      "step": 3189
    },
    {
      "epoch": 0.950129376942981,
      "grad_norm": 0.6292755007743835,
      "learning_rate": 2.467116376771528e-07,
      "loss": 1.1181,
      "step": 3190
    },
    {
      "epoch": 0.9504272231426497,
      "grad_norm": 0.6483851671218872,
      "learning_rate": 2.4377788122129765e-07,
      "loss": 1.1207,
      "step": 3191
    },
    {
      "epoch": 0.9507250693423184,
      "grad_norm": 0.6358315348625183,
      "learning_rate": 2.408614573048107e-07,
      "loss": 1.1183,
      "step": 3192
    },
    {
      "epoch": 0.951022915541987,
      "grad_norm": 0.6153903603553772,
      "learning_rate": 2.3796237110890297e-07,
      "loss": 1.0979,
      "step": 3193
    },
    {
      "epoch": 0.9513207617416557,
      "grad_norm": 0.6204462647438049,
      "learning_rate": 2.3508062778398234e-07,
      "loss": 1.0981,
      "step": 3194
    },
    {
      "epoch": 0.9516186079413242,
      "grad_norm": 0.6354315280914307,
      "learning_rate": 2.3221623244964576e-07,
      "loss": 1.0995,
      "step": 3195
    },
    {
      "epoch": 0.9519164541409929,
      "grad_norm": 0.6269338726997375,
      "learning_rate": 2.293691901946704e-07,
      "loss": 1.0957,
      "step": 3196
    },
    {
      "epoch": 0.9522143003406616,
      "grad_norm": 0.611155092716217,
      "learning_rate": 2.2653950607700704e-07,
      "loss": 1.0979,
      "step": 3197
    },
    {
      "epoch": 0.9525121465403302,
      "grad_norm": 0.6385710835456848,
      "learning_rate": 2.2372718512376545e-07,
      "loss": 1.1293,
      "step": 3198
    },
    {
      "epoch": 0.9528099927399989,
      "grad_norm": 0.6105167269706726,
      "learning_rate": 2.209322323312102e-07,
      "loss": 1.1096,
      "step": 3199
    },
    {
      "epoch": 0.9531078389396676,
      "grad_norm": 0.6441154479980469,
      "learning_rate": 2.1815465266474822e-07,
      "loss": 1.1222,
      "step": 3200
    },
    {
      "epoch": 0.9534056851393362,
      "grad_norm": 0.6489999294281006,
      "learning_rate": 2.1539445105892676e-07,
      "loss": 1.1324,
      "step": 3201
    },
    {
      "epoch": 0.9537035313390049,
      "grad_norm": 0.6379176378250122,
      "learning_rate": 2.1265163241741437e-07,
      "loss": 1.1047,
      "step": 3202
    },
    {
      "epoch": 0.9540013775386734,
      "grad_norm": 0.6353323459625244,
      "learning_rate": 2.0992620161299993e-07,
      "loss": 1.0997,
      "step": 3203
    },
    {
      "epoch": 0.9542992237383421,
      "grad_norm": 0.6120100021362305,
      "learning_rate": 2.0721816348758473e-07,
      "loss": 1.1077,
      "step": 3204
    },
    {
      "epoch": 0.9545970699380107,
      "grad_norm": 0.6662302017211914,
      "learning_rate": 2.045275228521637e-07,
      "loss": 1.1108,
      "step": 3205
    },
    {
      "epoch": 0.9548949161376794,
      "grad_norm": 0.627516508102417,
      "learning_rate": 2.018542844868332e-07,
      "loss": 1.1012,
      "step": 3206
    },
    {
      "epoch": 0.9551927623373481,
      "grad_norm": 0.627473771572113,
      "learning_rate": 1.9919845314076426e-07,
      "loss": 1.1015,
      "step": 3207
    },
    {
      "epoch": 0.9554906085370167,
      "grad_norm": 0.6258399486541748,
      "learning_rate": 1.9656003353221044e-07,
      "loss": 1.1103,
      "step": 3208
    },
    {
      "epoch": 0.9557884547366854,
      "grad_norm": 0.6296773552894592,
      "learning_rate": 1.9393903034848672e-07,
      "loss": 1.1004,
      "step": 3209
    },
    {
      "epoch": 0.9560863009363539,
      "grad_norm": 0.6335697770118713,
      "learning_rate": 1.9133544824597172e-07,
      "loss": 1.1084,
      "step": 3210
    },
    {
      "epoch": 0.9563841471360226,
      "grad_norm": 0.6075793504714966,
      "learning_rate": 1.8874929185009215e-07,
      "loss": 1.0819,
      "step": 3211
    },
    {
      "epoch": 0.9566819933356913,
      "grad_norm": 0.659818172454834,
      "learning_rate": 1.8618056575531506e-07,
      "loss": 1.1014,
      "step": 3212
    },
    {
      "epoch": 0.9569798395353599,
      "grad_norm": 0.6229186058044434,
      "learning_rate": 1.8362927452514335e-07,
      "loss": 1.1073,
      "step": 3213
    },
    {
      "epoch": 0.9572776857350286,
      "grad_norm": 0.6328487396240234,
      "learning_rate": 1.810954226921058e-07,
      "loss": 1.1139,
      "step": 3214
    },
    {
      "epoch": 0.9575755319346972,
      "grad_norm": 0.6437397003173828,
      "learning_rate": 1.785790147577504e-07,
      "loss": 1.1076,
      "step": 3215
    },
    {
      "epoch": 0.9578733781343659,
      "grad_norm": 0.6062115430831909,
      "learning_rate": 1.7608005519263116e-07,
      "loss": 1.091,
      "step": 3216
    },
    {
      "epoch": 0.9581712243340346,
      "grad_norm": 0.605218231678009,
      "learning_rate": 1.7359854843630786e-07,
      "loss": 1.1107,
      "step": 3217
    },
    {
      "epoch": 0.9584690705337031,
      "grad_norm": 0.6255162954330444,
      "learning_rate": 1.711344988973318e-07,
      "loss": 1.1153,
      "step": 3218
    },
    {
      "epoch": 0.9587669167333718,
      "grad_norm": 0.626694917678833,
      "learning_rate": 1.6868791095324132e-07,
      "loss": 1.0971,
      "step": 3219
    },
    {
      "epoch": 0.9590647629330404,
      "grad_norm": 0.6161735653877258,
      "learning_rate": 1.6625878895055293e-07,
      "loss": 1.0967,
      "step": 3220
    },
    {
      "epoch": 0.9593626091327091,
      "grad_norm": 0.6536877155303955,
      "learning_rate": 1.6384713720475454e-07,
      "loss": 1.1212,
      "step": 3221
    },
    {
      "epoch": 0.9596604553323778,
      "grad_norm": 0.6381300091743469,
      "learning_rate": 1.6145296000029564e-07,
      "loss": 1.125,
      "step": 3222
    },
    {
      "epoch": 0.9599583015320464,
      "grad_norm": 0.6323751211166382,
      "learning_rate": 1.5907626159058275e-07,
      "loss": 1.1097,
      "step": 3223
    },
    {
      "epoch": 0.9602561477317151,
      "grad_norm": 0.6388651132583618,
      "learning_rate": 1.5671704619796835e-07,
      "loss": 1.0945,
      "step": 3224
    },
    {
      "epoch": 0.9605539939313836,
      "grad_norm": 0.6361817121505737,
      "learning_rate": 1.543753180137475e-07,
      "loss": 1.1111,
      "step": 3225
    },
    {
      "epoch": 0.9608518401310523,
      "grad_norm": 0.617210865020752,
      "learning_rate": 1.5205108119814572e-07,
      "loss": 1.1051,
      "step": 3226
    },
    {
      "epoch": 0.961149686330721,
      "grad_norm": 0.6430678963661194,
      "learning_rate": 1.4974433988031666e-07,
      "loss": 1.1209,
      "step": 3227
    },
    {
      "epoch": 0.9614475325303896,
      "grad_norm": 0.6460803747177124,
      "learning_rate": 1.4745509815833002e-07,
      "loss": 1.1021,
      "step": 3228
    },
    {
      "epoch": 0.9617453787300583,
      "grad_norm": 0.6125198602676392,
      "learning_rate": 1.451833600991659e-07,
      "loss": 1.0946,
      "step": 3229
    },
    {
      "epoch": 0.9620432249297269,
      "grad_norm": 0.6078023910522461,
      "learning_rate": 1.429291297387103e-07,
      "loss": 1.0921,
      "step": 3230
    },
    {
      "epoch": 0.9623410711293956,
      "grad_norm": 0.6453407406806946,
      "learning_rate": 1.4069241108174425e-07,
      "loss": 1.0984,
      "step": 3231
    },
    {
      "epoch": 0.9626389173290643,
      "grad_norm": 0.6130032539367676,
      "learning_rate": 1.384732081019391e-07,
      "loss": 1.1128,
      "step": 3232
    },
    {
      "epoch": 0.9629367635287328,
      "grad_norm": 0.6166512966156006,
      "learning_rate": 1.3627152474184669e-07,
      "loss": 1.1053,
      "step": 3233
    },
    {
      "epoch": 0.9632346097284015,
      "grad_norm": 0.6331944465637207,
      "learning_rate": 1.3408736491289598e-07,
      "loss": 1.1061,
      "step": 3234
    },
    {
      "epoch": 0.9635324559280701,
      "grad_norm": 0.6062200665473938,
      "learning_rate": 1.3192073249538527e-07,
      "loss": 1.1,
      "step": 3235
    },
    {
      "epoch": 0.9638303021277388,
      "grad_norm": 0.6297964453697205,
      "learning_rate": 1.2977163133847004e-07,
      "loss": 1.1208,
      "step": 3236
    },
    {
      "epoch": 0.9641281483274075,
      "grad_norm": 0.6453106999397278,
      "learning_rate": 1.2764006526016391e-07,
      "loss": 1.0959,
      "step": 3237
    },
    {
      "epoch": 0.9644259945270761,
      "grad_norm": 0.6375758051872253,
      "learning_rate": 1.2552603804732888e-07,
      "loss": 1.112,
      "step": 3238
    },
    {
      "epoch": 0.9647238407267448,
      "grad_norm": 0.622739851474762,
      "learning_rate": 1.2342955345566398e-07,
      "loss": 1.1042,
      "step": 3239
    },
    {
      "epoch": 0.9650216869264133,
      "grad_norm": 0.6220073103904724,
      "learning_rate": 1.213506152097077e-07,
      "loss": 1.1199,
      "step": 3240
    },
    {
      "epoch": 0.965319533126082,
      "grad_norm": 0.6175379157066345,
      "learning_rate": 1.192892270028223e-07,
      "loss": 1.1005,
      "step": 3241
    },
    {
      "epoch": 0.9656173793257506,
      "grad_norm": 0.6419153213500977,
      "learning_rate": 1.1724539249719502e-07,
      "loss": 1.0992,
      "step": 3242
    },
    {
      "epoch": 0.9659152255254193,
      "grad_norm": 0.6424520015716553,
      "learning_rate": 1.1521911532382357e-07,
      "loss": 1.0999,
      "step": 3243
    },
    {
      "epoch": 0.966213071725088,
      "grad_norm": 0.6317712664604187,
      "learning_rate": 1.132103990825173e-07,
      "loss": 1.0929,
      "step": 3244
    },
    {
      "epoch": 0.9665109179247566,
      "grad_norm": 0.6123984456062317,
      "learning_rate": 1.1121924734188716e-07,
      "loss": 1.1016,
      "step": 3245
    },
    {
      "epoch": 0.9668087641244253,
      "grad_norm": 0.6378028988838196,
      "learning_rate": 1.0924566363933686e-07,
      "loss": 1.0832,
      "step": 3246
    },
    {
      "epoch": 0.9671066103240938,
      "grad_norm": 0.6491348147392273,
      "learning_rate": 1.0728965148106174e-07,
      "loss": 1.0742,
      "step": 3247
    },
    {
      "epoch": 0.9674044565237625,
      "grad_norm": 0.6289087533950806,
      "learning_rate": 1.0535121434204209e-07,
      "loss": 1.1109,
      "step": 3248
    },
    {
      "epoch": 0.9677023027234312,
      "grad_norm": 0.6241329908370972,
      "learning_rate": 1.0343035566603099e-07,
      "loss": 1.1084,
      "step": 3249
    },
    {
      "epoch": 0.9680001489230998,
      "grad_norm": 0.6303631067276001,
      "learning_rate": 1.0152707886555423e-07,
      "loss": 1.1019,
      "step": 3250
    },
    {
      "epoch": 0.9682979951227685,
      "grad_norm": 0.6296857595443726,
      "learning_rate": 9.964138732190266e-08,
      "loss": 1.1149,
      "step": 3251
    },
    {
      "epoch": 0.9685958413224371,
      "grad_norm": 0.644363284111023,
      "learning_rate": 9.777328438512535e-08,
      "loss": 1.1148,
      "step": 3252
    },
    {
      "epoch": 0.9688936875221058,
      "grad_norm": 0.6293293833732605,
      "learning_rate": 9.592277337402311e-08,
      "loss": 1.1184,
      "step": 3253
    },
    {
      "epoch": 0.9691915337217745,
      "grad_norm": 0.6270025372505188,
      "learning_rate": 9.408985757614619e-08,
      "loss": 1.1045,
      "step": 3254
    },
    {
      "epoch": 0.969489379921443,
      "grad_norm": 0.6304171681404114,
      "learning_rate": 9.227454024778426e-08,
      "loss": 1.1143,
      "step": 3255
    },
    {
      "epoch": 0.9697872261211117,
      "grad_norm": 0.6242616176605225,
      "learning_rate": 9.04768246139598e-08,
      "loss": 1.085,
      "step": 3256
    },
    {
      "epoch": 0.9700850723207803,
      "grad_norm": 0.6361697316169739,
      "learning_rate": 8.869671386842915e-08,
      "loss": 1.1229,
      "step": 3257
    },
    {
      "epoch": 0.970382918520449,
      "grad_norm": 0.6258002519607544,
      "learning_rate": 8.693421117367151e-08,
      "loss": 1.1146,
      "step": 3258
    },
    {
      "epoch": 0.9706807647201177,
      "grad_norm": 0.626517653465271,
      "learning_rate": 8.518931966088106e-08,
      "loss": 1.1112,
      "step": 3259
    },
    {
      "epoch": 0.9709786109197863,
      "grad_norm": 0.6407362818717957,
      "learning_rate": 8.346204242996703e-08,
      "loss": 1.1313,
      "step": 3260
    },
    {
      "epoch": 0.971276457119455,
      "grad_norm": 0.6160361170768738,
      "learning_rate": 8.175238254954587e-08,
      "loss": 1.1052,
      "step": 3261
    },
    {
      "epoch": 0.9715743033191235,
      "grad_norm": 0.6360178589820862,
      "learning_rate": 8.006034305693688e-08,
      "loss": 1.1157,
      "step": 3262
    },
    {
      "epoch": 0.9718721495187922,
      "grad_norm": 0.6147614121437073,
      "learning_rate": 7.838592695815327e-08,
      "loss": 1.0973,
      "step": 3263
    },
    {
      "epoch": 0.9721699957184609,
      "grad_norm": 0.6301278471946716,
      "learning_rate": 7.672913722789888e-08,
      "loss": 1.1217,
      "step": 3264
    },
    {
      "epoch": 0.9724678419181295,
      "grad_norm": 0.6168976426124573,
      "learning_rate": 7.508997680956587e-08,
      "loss": 1.0982,
      "step": 3265
    },
    {
      "epoch": 0.9727656881177982,
      "grad_norm": 0.6298949122428894,
      "learning_rate": 7.346844861522374e-08,
      "loss": 1.109,
      "step": 3266
    },
    {
      "epoch": 0.9730635343174668,
      "grad_norm": 0.6333845853805542,
      "learning_rate": 7.186455552562033e-08,
      "loss": 1.1019,
      "step": 3267
    },
    {
      "epoch": 0.9733613805171355,
      "grad_norm": 0.6192238926887512,
      "learning_rate": 7.027830039017192e-08,
      "loss": 1.1034,
      "step": 3268
    },
    {
      "epoch": 0.9736592267168042,
      "grad_norm": 0.6329990029335022,
      "learning_rate": 6.870968602695871e-08,
      "loss": 1.1056,
      "step": 3269
    },
    {
      "epoch": 0.9739570729164727,
      "grad_norm": 0.6322906017303467,
      "learning_rate": 6.715871522272376e-08,
      "loss": 1.1006,
      "step": 3270
    },
    {
      "epoch": 0.9742549191161414,
      "grad_norm": 0.6158894896507263,
      "learning_rate": 6.562539073286411e-08,
      "loss": 1.0886,
      "step": 3271
    },
    {
      "epoch": 0.97455276531581,
      "grad_norm": 0.624439537525177,
      "learning_rate": 6.41097152814263e-08,
      "loss": 1.1102,
      "step": 3272
    },
    {
      "epoch": 0.9748506115154787,
      "grad_norm": 0.624197781085968,
      "learning_rate": 6.261169156110192e-08,
      "loss": 1.1051,
      "step": 3273
    },
    {
      "epoch": 0.9751484577151474,
      "grad_norm": 0.6293463706970215,
      "learning_rate": 6.113132223322327e-08,
      "loss": 1.1155,
      "step": 3274
    },
    {
      "epoch": 0.975446303914816,
      "grad_norm": 0.6274569630622864,
      "learning_rate": 5.966860992776214e-08,
      "loss": 1.1019,
      "step": 3275
    },
    {
      "epoch": 0.9757441501144847,
      "grad_norm": 0.627772867679596,
      "learning_rate": 5.822355724331541e-08,
      "loss": 1.105,
      "step": 3276
    },
    {
      "epoch": 0.9760419963141532,
      "grad_norm": 0.6279445886611938,
      "learning_rate": 5.6796166747110636e-08,
      "loss": 1.1201,
      "step": 3277
    },
    {
      "epoch": 0.9763398425138219,
      "grad_norm": 0.6300987005233765,
      "learning_rate": 5.538644097499601e-08,
      "loss": 1.1003,
      "step": 3278
    },
    {
      "epoch": 0.9766376887134905,
      "grad_norm": 0.6322870254516602,
      "learning_rate": 5.3994382431439285e-08,
      "loss": 1.1219,
      "step": 3279
    },
    {
      "epoch": 0.9769355349131592,
      "grad_norm": 0.6523212194442749,
      "learning_rate": 5.2619993589518856e-08,
      "loss": 1.1262,
      "step": 3280
    },
    {
      "epoch": 0.9772333811128279,
      "grad_norm": 0.6235091686248779,
      "learning_rate": 5.126327689092048e-08,
      "loss": 1.1066,
      "step": 3281
    },
    {
      "epoch": 0.9775312273124965,
      "grad_norm": 0.6292470693588257,
      "learning_rate": 4.992423474593944e-08,
      "loss": 1.1025,
      "step": 3282
    },
    {
      "epoch": 0.9778290735121652,
      "grad_norm": 0.6195799112319946,
      "learning_rate": 4.860286953346616e-08,
      "loss": 1.1199,
      "step": 3283
    },
    {
      "epoch": 0.9781269197118339,
      "grad_norm": 0.634871244430542,
      "learning_rate": 4.72991836009884e-08,
      "loss": 1.1086,
      "step": 3284
    },
    {
      "epoch": 0.9784247659115024,
      "grad_norm": 0.6036664247512817,
      "learning_rate": 4.6013179264587924e-08,
      "loss": 1.1036,
      "step": 3285
    },
    {
      "epoch": 0.9787226121111711,
      "grad_norm": 0.61818927526474,
      "learning_rate": 4.4744858808929424e-08,
      "loss": 1.0975,
      "step": 3286
    },
    {
      "epoch": 0.9790204583108397,
      "grad_norm": 0.6205064654350281,
      "learning_rate": 4.3494224487264925e-08,
      "loss": 1.1106,
      "step": 3287
    },
    {
      "epoch": 0.9793183045105084,
      "grad_norm": 0.6472957134246826,
      "learning_rate": 4.226127852142381e-08,
      "loss": 1.1093,
      "step": 3288
    },
    {
      "epoch": 0.979616150710177,
      "grad_norm": 0.6243222951889038,
      "learning_rate": 4.1046023101812824e-08,
      "loss": 1.0918,
      "step": 3289
    },
    {
      "epoch": 0.9799139969098457,
      "grad_norm": 0.627770721912384,
      "learning_rate": 3.984846038740831e-08,
      "loss": 1.1133,
      "step": 3290
    },
    {
      "epoch": 0.9802118431095144,
      "grad_norm": 0.615742027759552,
      "learning_rate": 3.866859250575616e-08,
      "loss": 1.0971,
      "step": 3291
    },
    {
      "epoch": 0.9805096893091829,
      "grad_norm": 0.6772239804267883,
      "learning_rate": 3.750642155296525e-08,
      "loss": 1.1057,
      "step": 3292
    },
    {
      "epoch": 0.9808075355088516,
      "grad_norm": 0.651216983795166,
      "learning_rate": 3.636194959370398e-08,
      "loss": 1.1044,
      "step": 3293
    },
    {
      "epoch": 0.9811053817085202,
      "grad_norm": 0.6263548135757446,
      "learning_rate": 3.523517866120041e-08,
      "loss": 1.0941,
      "step": 3294
    },
    {
      "epoch": 0.9814032279081889,
      "grad_norm": 0.636808454990387,
      "learning_rate": 3.4126110757232155e-08,
      "loss": 1.1163,
      "step": 3295
    },
    {
      "epoch": 0.9817010741078576,
      "grad_norm": 0.6443750858306885,
      "learning_rate": 3.303474785212868e-08,
      "loss": 1.1101,
      "step": 3296
    },
    {
      "epoch": 0.9819989203075262,
      "grad_norm": 0.6294336318969727,
      "learning_rate": 3.1961091884764596e-08,
      "loss": 1.117,
      "step": 3297
    },
    {
      "epoch": 0.9822967665071949,
      "grad_norm": 0.6188221573829651,
      "learning_rate": 3.090514476255635e-08,
      "loss": 1.1117,
      "step": 3298
    },
    {
      "epoch": 0.9825946127068635,
      "grad_norm": 0.6323879361152649,
      "learning_rate": 2.986690836145889e-08,
      "loss": 1.0986,
      "step": 3299
    },
    {
      "epoch": 0.9828924589065321,
      "grad_norm": 0.620199978351593,
      "learning_rate": 2.8846384525966775e-08,
      "loss": 1.1038,
      "step": 3300
    },
    {
      "epoch": 0.9831903051062008,
      "grad_norm": 0.6270275115966797,
      "learning_rate": 2.7843575069103067e-08,
      "loss": 1.1146,
      "step": 3301
    },
    {
      "epoch": 0.9834881513058694,
      "grad_norm": 0.6158193945884705,
      "learning_rate": 2.6858481772421563e-08,
      "loss": 1.1058,
      "step": 3302
    },
    {
      "epoch": 0.9837859975055381,
      "grad_norm": 0.6240172386169434,
      "learning_rate": 2.589110638600234e-08,
      "loss": 1.1033,
      "step": 3303
    },
    {
      "epoch": 0.9840838437052067,
      "grad_norm": 0.61282879114151,
      "learning_rate": 2.4941450628448438e-08,
      "loss": 1.0873,
      "step": 3304
    },
    {
      "epoch": 0.9843816899048754,
      "grad_norm": 0.6148067116737366,
      "learning_rate": 2.400951618688252e-08,
      "loss": 1.1066,
      "step": 3305
    },
    {
      "epoch": 0.9846795361045441,
      "grad_norm": 0.6301677823066711,
      "learning_rate": 2.3095304716944655e-08,
      "loss": 1.1072,
      "step": 3306
    },
    {
      "epoch": 0.9849773823042126,
      "grad_norm": 0.6311284303665161,
      "learning_rate": 2.2198817842788988e-08,
      "loss": 1.1423,
      "step": 3307
    },
    {
      "epoch": 0.9852752285038813,
      "grad_norm": 0.6061777472496033,
      "learning_rate": 2.13200571570793e-08,
      "loss": 1.0987,
      "step": 3308
    },
    {
      "epoch": 0.9855730747035499,
      "grad_norm": 0.6452845335006714,
      "learning_rate": 2.045902422099122e-08,
      "loss": 1.1177,
      "step": 3309
    },
    {
      "epoch": 0.9858709209032186,
      "grad_norm": 0.6421290040016174,
      "learning_rate": 1.9615720564201134e-08,
      "loss": 1.1046,
      "step": 3310
    },
    {
      "epoch": 0.9861687671028873,
      "grad_norm": 0.605431318283081,
      "learning_rate": 1.8790147684892847e-08,
      "loss": 1.11,
      "step": 3311
    },
    {
      "epoch": 0.9864666133025559,
      "grad_norm": 0.6400191187858582,
      "learning_rate": 1.798230704974646e-08,
      "loss": 1.1039,
      "step": 3312
    },
    {
      "epoch": 0.9867644595022246,
      "grad_norm": 0.6396893262863159,
      "learning_rate": 1.7192200093943955e-08,
      "loss": 1.1153,
      "step": 3313
    },
    {
      "epoch": 0.9870623057018932,
      "grad_norm": 0.6280839443206787,
      "learning_rate": 1.6419828221156953e-08,
      "loss": 1.1115,
      "step": 3314
    },
    {
      "epoch": 0.9873601519015618,
      "grad_norm": 0.6262692213058472,
      "learning_rate": 1.5665192803554498e-08,
      "loss": 1.0999,
      "step": 3315
    },
    {
      "epoch": 0.9876579981012304,
      "grad_norm": 0.6309360861778259,
      "learning_rate": 1.4928295181791953e-08,
      "loss": 1.1025,
      "step": 3316
    },
    {
      "epoch": 0.9879558443008991,
      "grad_norm": 0.6229701638221741,
      "learning_rate": 1.4209136665015444e-08,
      "loss": 1.1157,
      "step": 3317
    },
    {
      "epoch": 0.9882536905005678,
      "grad_norm": 0.6568220257759094,
      "learning_rate": 1.3507718530855196e-08,
      "loss": 1.121,
      "step": 3318
    },
    {
      "epoch": 0.9885515367002364,
      "grad_norm": 0.6244164705276489,
      "learning_rate": 1.2824042025423311e-08,
      "loss": 1.0927,
      "step": 3319
    },
    {
      "epoch": 0.9888493828999051,
      "grad_norm": 0.5994888544082642,
      "learning_rate": 1.215810836331488e-08,
      "loss": 1.0832,
      "step": 3320
    },
    {
      "epoch": 0.9891472290995738,
      "grad_norm": 0.6252540946006775,
      "learning_rate": 1.1509918727602432e-08,
      "loss": 1.0886,
      "step": 3321
    },
    {
      "epoch": 0.9894450752992423,
      "grad_norm": 0.6171160340309143,
      "learning_rate": 1.0879474269835933e-08,
      "loss": 1.0892,
      "step": 3322
    },
    {
      "epoch": 0.989742921498911,
      "grad_norm": 0.6351056098937988,
      "learning_rate": 1.0266776110038346e-08,
      "loss": 1.1042,
      "step": 3323
    },
    {
      "epoch": 0.9900407676985796,
      "grad_norm": 0.6355272531509399,
      "learning_rate": 9.671825336705631e-09,
      "loss": 1.08,
      "step": 3324
    },
    {
      "epoch": 0.9903386138982483,
      "grad_norm": 0.608055591583252,
      "learning_rate": 9.094623006807857e-09,
      "loss": 1.1043,
      "step": 3325
    },
    {
      "epoch": 0.990636460097917,
      "grad_norm": 0.6255134344100952,
      "learning_rate": 8.535170145779204e-09,
      "loss": 1.0995,
      "step": 3326
    },
    {
      "epoch": 0.9909343062975856,
      "grad_norm": 0.6263388395309448,
      "learning_rate": 7.99346774752241e-09,
      "loss": 1.0959,
      "step": 3327
    },
    {
      "epoch": 0.9912321524972543,
      "grad_norm": 0.6263720989227295,
      "learning_rate": 7.469516774406549e-09,
      "loss": 1.097,
      "step": 3328
    },
    {
      "epoch": 0.9915299986969229,
      "grad_norm": 0.6450942754745483,
      "learning_rate": 6.9633181572636985e-09,
      "loss": 1.1,
      "step": 3329
    },
    {
      "epoch": 0.9918278448965915,
      "grad_norm": 0.6288226246833801,
      "learning_rate": 6.474872795386722e-09,
      "loss": 1.1009,
      "step": 3330
    },
    {
      "epoch": 0.9921256910962601,
      "grad_norm": 0.6205242872238159,
      "learning_rate": 6.004181556529265e-09,
      "loss": 1.0983,
      "step": 3331
    },
    {
      "epoch": 0.9924235372959288,
      "grad_norm": 0.647709310054779,
      "learning_rate": 5.551245276903538e-09,
      "loss": 1.1252,
      "step": 3332
    },
    {
      "epoch": 0.9927213834955975,
      "grad_norm": 0.6193277835845947,
      "learning_rate": 5.116064761179207e-09,
      "loss": 1.1081,
      "step": 3333
    },
    {
      "epoch": 0.9930192296952661,
      "grad_norm": 0.6222624182701111,
      "learning_rate": 4.698640782483388e-09,
      "loss": 1.1028,
      "step": 3334
    },
    {
      "epoch": 0.9933170758949348,
      "grad_norm": 0.6340826749801636,
      "learning_rate": 4.298974082393992e-09,
      "loss": 1.103,
      "step": 3335
    },
    {
      "epoch": 0.9936149220946034,
      "grad_norm": 0.6353055238723755,
      "learning_rate": 3.9170653709441616e-09,
      "loss": 1.1107,
      "step": 3336
    },
    {
      "epoch": 0.993912768294272,
      "grad_norm": 0.6355594396591187,
      "learning_rate": 3.5529153266200542e-09,
      "loss": 1.1132,
      "step": 3337
    },
    {
      "epoch": 0.9942106144939407,
      "grad_norm": 0.6177548766136169,
      "learning_rate": 3.206524596355287e-09,
      "loss": 1.0927,
      "step": 3338
    },
    {
      "epoch": 0.9945084606936093,
      "grad_norm": 0.631292462348938,
      "learning_rate": 2.87789379553538e-09,
      "loss": 1.1062,
      "step": 3339
    },
    {
      "epoch": 0.994806306893278,
      "grad_norm": 0.629156231880188,
      "learning_rate": 2.567023507994426e-09,
      "loss": 1.1011,
      "step": 3340
    },
    {
      "epoch": 0.9951041530929466,
      "grad_norm": 0.6388160586357117,
      "learning_rate": 2.2739142860106476e-09,
      "loss": 1.1254,
      "step": 3341
    },
    {
      "epoch": 0.9954019992926153,
      "grad_norm": 0.6306875348091125,
      "learning_rate": 1.998566650313061e-09,
      "loss": 1.0928,
      "step": 3342
    },
    {
      "epoch": 0.995699845492284,
      "grad_norm": 0.6228340864181519,
      "learning_rate": 1.7409810900737012e-09,
      "loss": 1.1128,
      "step": 3343
    },
    {
      "epoch": 0.9959976916919525,
      "grad_norm": 0.6215200424194336,
      "learning_rate": 1.5011580629087364e-09,
      "loss": 1.1123,
      "step": 3344
    },
    {
      "epoch": 0.9962955378916212,
      "grad_norm": 0.6278738379478455,
      "learning_rate": 1.2790979948784643e-09,
      "loss": 1.1297,
      "step": 3345
    },
    {
      "epoch": 0.9965933840912898,
      "grad_norm": 0.6252317428588867,
      "learning_rate": 1.0748012804884245e-09,
      "loss": 1.1152,
      "step": 3346
    },
    {
      "epoch": 0.9968912302909585,
      "grad_norm": 0.6451824903488159,
      "learning_rate": 8.882682826816258e-10,
      "loss": 1.1137,
      "step": 3347
    },
    {
      "epoch": 0.9971890764906272,
      "grad_norm": 0.6302453279495239,
      "learning_rate": 7.194993328485389e-10,
      "loss": 1.1087,
      "step": 3348
    },
    {
      "epoch": 0.9974869226902958,
      "grad_norm": 0.6144789457321167,
      "learning_rate": 5.684947308159939e-10,
      "loss": 1.1054,
      "step": 3349
    },
    {
      "epoch": 0.9977847688899645,
      "grad_norm": 0.6399134397506714,
      "learning_rate": 4.352547448527311e-10,
      "loss": 1.1131,
      "step": 3350
    },
    {
      "epoch": 0.9980826150896331,
      "grad_norm": 0.6308466792106628,
      "learning_rate": 3.197796116694019e-10,
      "loss": 1.0987,
      "step": 3351
    },
    {
      "epoch": 0.9983804612893017,
      "grad_norm": 0.6232466697692871,
      "learning_rate": 2.2206953641412676e-10,
      "loss": 1.1183,
      "step": 3352
    },
    {
      "epoch": 0.9986783074889704,
      "grad_norm": 0.6292468309402466,
      "learning_rate": 1.4212469267582684e-10,
      "loss": 1.1014,
      "step": 3353
    },
    {
      "epoch": 0.998976153688639,
      "grad_norm": 0.6216614246368408,
      "learning_rate": 7.994522247978254e-11,
      "loss": 1.1195,
      "step": 3354
    },
    {
      "epoch": 0.9992739998883077,
      "grad_norm": 0.6190215945243835,
      "learning_rate": 3.5531236292074555e-11,
      "loss": 1.1197,
      "step": 3355
    },
    {
      "epoch": 0.9995718460879763,
      "grad_norm": 0.6289578080177307,
      "learning_rate": 8.882813018473713e-12,
      "loss": 1.1157,
      "step": 3356
    },
    {
      "epoch": 0.999869692287645,
      "grad_norm": 0.626802921295166,
      "learning_rate": 0.0,
      "loss": 1.1195,
      "step": 3357
    },
    {
      "epoch": 0.999869692287645,
      "step": 3357,
      "total_flos": 5.612488984343714e+19,
      "train_loss": 1.1850541065518332,
      "train_runtime": 1011044.2105,
      "train_samples_per_second": 1.7,
      "train_steps_per_second": 0.003
    }
  ],
  "logging_steps": 1,
  "max_steps": 3357,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 3,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.612488984343714e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}